Datasets:

MaartenGr

/

arxiv_nlp

like 1

Toward Network-based Keyword Extraction from Multitopic Web Documents

In this paper we analyse the selectivity measure calculated from the complex network in the task of the automatic keyword extraction. Texts, collected from different web sources (portals, forums), are represented as directed and weighted co-occurrence complex networks of words. Words are nodes and links are established between two nodes if they are directly co-occurring within the sentence. We test different centrality measures for ranking nodes - keyword candidates. The promising results are achieved using the selectivity measure. Then we propose an approach which enables extracting word pairs according to the values of the in/out selectivity and weight measures combined with filtering.

Computation and Language

Benchmarking Named Entity Disambiguation approaches for Streaming Graphs

Named Entity Disambiaguation (NED) is a central task for applications dealing with natural language text. Assume that we have a graph based knowledge base (subsequently referred as Knowledge Graph) where nodes represent various real world entities such as people, location, organization and concepts. Given data sources such as social media streams and web pages Entity Linking is the task of mapping named entities that are extracted from the data to those present in the Knowledge Graph. This is an inherently difficult task due to several reasons. Almost all these data sources are generated without any formal ontology; the unstructured nature of the input, limited context and the ambiguity involved when multiple entities are mapped to the same name make this a hard task. This report looks at two state of the art systems employing two distinctive approaches: graph based Accurate Online Disambiguation of Entities (AIDA) and Mined Evidence Named Entity Disambiguation (MENED), which employs a statistical inference approach. We compare both approaches using the data set and queries provided by the Knowledge Base Population (KBP) track at 2011 NIST Text Analytics Conference (TAC). This report begins with an overview of the respective approaches, followed by detailed description of the experimental setup. It concludes with our findings from the benchmarking exercise.

Computation and Language

Toward Selectivity Based Keyword Extraction for Croatian News

Preliminary report on network based keyword extraction for Croatian is an unsupervised method for keyword extraction from the complex network. We build our approach with a new network measure the node selectivity, motivated by the research of the graph based centrality approaches. The node selectivity is defined as the average weight distribution on the links of the single node. We extract nodes (keyword candidates) based on the selectivity value. Furthermore, we expand extracted nodes to word-tuples ranked with the highest in/out selectivity values. Selectivity based extraction does not require linguistic knowledge while it is purely derived from statistical and structural information en-compassed in the source text which is reflected into the structure of the network. Obtained sets are evaluated on a manually annotated keywords: for the set of extracted keyword candidates average F1 score is 24,63%, and average F2 score is 21,19%; for the exacted words-tuples candidates average F1 score is 25,9% and average F2 score is 24,47%.

Computation and Language

Modeling languages from graph networks

We model and compute the probability distribution of the letters in random generated words in a language by using the theory of set partitions, Young tableaux and graph theoretical representation methods. This has been of interest for several application areas such as network systems, bioinformatics, internet search, data mining and computacional linguistics.

Computation and Language

Autonomous requirements specification processing using natural language processing

We describe our ongoing research that centres on the application of natural language processing (NLP) to software engineering and systems development activities. In particular, this paper addresses the use of NLP in the requirements analysis and systems design processes. We have developed a prototype toolset that can assist the systems analyst or software engineer to select and verify terms relevant to a project. In this paper we describe the processes employed by the system to extract and classify objects of interest from requirements documents. These processes are illustrated using a small example.

Computation and Language

Substitute Based SCODE Word Embeddings in Supervised NLP Tasks

We analyze a word embedding method in supervised tasks. It maps words on a sphere such that words co-occurring in similar contexts lie closely. The similarity of contexts is measured by the distribution of substitutes that can fill them. We compared word embeddings, including more recent representations, in Named Entity Recognition (NER), Chunking, and Dependency Parsing. We examine our framework in multilingual dependency parsing as well. The results show that the proposed method achieves as good as or better results compared to the other word embeddings in the tasks we investigate. It achieves state-of-the-art results in multilingual dependency parsing. Word embeddings in 7 languages are available for public use.

Computation and Language

Interpretable Low-Rank Document Representations with Label-Dependent Sparsity Patterns

In context of document classification, where in a corpus of documents their label tags are readily known, an opportunity lies in utilizing label information to learn document representation spaces with better discriminative properties. To this end, in this paper application of a Variational Bayesian Supervised Nonnegative Matrix Factorization (supervised vbNMF) with label-driven sparsity structure of coefficients is proposed for learning of discriminative nonsubtractive latent semantic components occuring in TF-IDF document representations. Constraints are such that the components pursued are made to be frequently occuring in a small set of labels only, making it possible to yield document representations with distinctive label-specific sparse activation patterns. A simple measure of quality of this kind of sparsity structure, dubbed inter-label sparsity, is introduced and experimentally brought into tight connection with classification performance. Representing a great practical convenience, inter-label sparsity is shown to be easily controlled in supervised vbNMF by a single parameter.

Computation and Language

Principles and Parameters: a coding theory perspective

We propose an approach to Longobardi's parametric comparison method (PCM) via the theory of error-correcting codes. One associates to a collection of languages to be analyzed with the PCM a binary (or ternary) code with one code words for each language in the family and each word consisting of the binary values of the syntactic parameters of the language, with the ternary case allowing for an additional parameter state that takes into account phenomena of entailment of parameters. The code parameters of the resulting code can be compared with some classical bounds in coding theory: the asymptotic bound, the Gilbert-Varshamov bound, etc. The position of the code parameters with respect to some of these bounds provides quantitative information on the variability of syntactic parameters within and across historical-linguistic families. While computations carried out for languages belonging to the same family yield codes below the GV curve, comparisons across different historical families can give examples of isolated codes lying above the asymptotic bound.

Computation and Language

Two-pass Discourse Segmentation with Pairing and Global Features

Previous attempts at RST-style discourse segmentation typically adopt features centered on a single token to predict whether to insert a boundary before that token. In contrast, we develop a discourse segmenter utilizing a set of pairing features, which are centered on a pair of adjacent tokens in the sentence, by equally taking into account the information from both tokens. Moreover, we propose a novel set of global features, which encode characteristics of the segmentation as a whole, once we have an initial segmentation. We show that both the pairing and global features are useful on their own, and their combination achieved an $F_1$ of 92.6% of identifying in-sentence discourse boundaries, which is a 17.8% error-rate reduction over the state-of-the-art performance, approaching 95% of human performance. In addition, similar improvement is observed across different classification frameworks.

Computation and Language

Architecture of a Web-based Predictive Editor for Controlled Natural Language Processing

In this paper, we describe the architecture of a web-based predictive text editor being developed for the controlled natural language PENG$^{ASP)$. This controlled language can be used to write non-monotonic specifications that have the same expressive power as Answer Set Programs. In order to support the writing process of these specifications, the predictive text editor communicates asynchronously with the controlled natural language processor that generates lookahead categories and additional auxiliary information for the author of a specification text. The text editor can display multiple sets of lookahead categories simultaneously for different possible sentence completions, anaphoric expressions, and supports the addition of new content words to the lexicon.

Computation and Language

Targetable Named Entity Recognition in Social Media

We present a novel approach for recognizing what we call targetable named entities; that is, named entities in a targeted set (e.g, movies, books, TV shows). Unlike many other NER systems that need to retrain their statistical models as new entities arrive, our approach does not require such retraining, which makes it more adaptable for types of entities that are frequently updated. For this preliminary study, we focus on one entity type, movie title, using data collected from Twitter. Our system is tested on two evaluation sets, one including only entities corresponding to movies in our training set, and the other excluding any of those entities. Our final model shows F1-scores of 76.19% and 78.70% on these evaluation sets, which gives strong evidence that our approach is completely unbiased to any par- ticular set of entities found during training.

Computation and Language

Text to Multi-level MindMaps: A Novel Method for Hierarchical Visual Abstraction of Natural Language Text

MindMapping is a well-known technique used in note taking, which encourages learning and studying. MindMapping has been manually adopted to help present knowledge and concepts in a visual form. Unfortunately, there is no reliable automated approach to generate MindMaps from Natural Language text. This work firstly introduces MindMap Multilevel Visualization concept which is to jointly visualize and summarize textual information. The visualization is achieved pictorially across multiple levels using semantic information (i.e. ontology), while the summarization is achieved by the information in the highest levels as they represent abstract information in the text. This work also presents the first automated approach that takes a text input and generates a MindMap visualization out of it. The approach could visualize text documents in multilevel MindMaps, in which a high-level MindMap node could be expanded into child MindMaps. \ignore{ As far as we know, this is the first work that view MindMapping as a new approach to jointly summarize and visualize textual information.} The proposed method involves understanding of the input text and converting it into intermediate Detailed Meaning Representation (DMR). The DMR is then visualized with two modes; Single level or Multiple levels, which is convenient for larger text. The generated MindMaps from both approaches were evaluated based on Human Subject experiments performed on Amazon Mechanical Turk with various parameter settings.

Computation and Language

Beyond description. Comment on "Approaching human language with complex networks" by Cong & Liu

Comment on "Approaching human language with complex networks" by Cong & Liu

Computation and Language

Microtask crowdsourcing for disease mention annotation in PubMed abstracts

Identifying concepts and relationships in biomedical text enables knowledge to be applied in computational analyses. Many biological natural language process (BioNLP) projects attempt to address this challenge, but the state of the art in BioNLP still leaves much room for improvement. Progress in BioNLP research depends on large, annotated corpora for evaluating information extraction systems and training machine learning models. Traditionally, such corpora are created by small numbers of expert annotators often working over extended periods of time. Recent studies have shown that workers on microtask crowdsourcing platforms such as Amazon's Mechanical Turk (AMT) can, in aggregate, generate high-quality annotations of biomedical text. Here, we investigated the use of the AMT in capturing disease mentions in PubMed abstracts. We used the NCBI Disease corpus as a gold standard for refining and benchmarking our crowdsourcing protocol. After several iterations, we arrived at a protocol that reproduced the annotations of the 593 documents in the training set of this gold standard with an overall F measure of 0.872 (precision 0.862, recall 0.883). The output can also be tuned to optimize for precision (max = 0.984 when recall = 0.269) or recall (max = 0.980 when precision = 0.436). Each document was examined by 15 workers, and their annotations were merged based on a simple voting method. In total 145 workers combined to complete all 593 documents in the span of 1 week at a cost of $.06 per abstract per worker. The quality of the annotations, as judged with the F measure, increases with the number of workers assigned to each task such that the system can be tuned to balance cost against quality. These results demonstrate that microtask crowdsourcing can be a valuable tool for generating well-annotated corpora in BioNLP.

Computation and Language

A model of grassroots changes in linguistic systems

Linguistic norms emerge in human communities because people imitate each other. A shared linguistic system provides people with the benefits of shared knowledge and coordinated planning. Once norms are in place, why would they ever change? This question, echoing broad questions in the theory of social dynamics, has particular force in relation to language. By definition, an innovator is in the minority when the innovation first occurs. In some areas of social dynamics, important minorities can strongly influence the majority through their power, fame, or use of broadcast media. But most linguistic changes are grassroots developments that originate with ordinary people. Here, we develop a novel model of communicative behavior in communities, and identify a mechanism for arbitrary innovations by ordinary people to have a good chance of being widely adopted. To imitate each other, people must form a mental representation of what other people do. Each time they speak, they must also decide which form to produce themselves. We introduce a new decision function that enables us to smoothly explore the space between two types of behavior: probability matching (matching the probabilities of incoming experience) and regularization (producing some forms disproportionately often). Using Monte Carlo methods, we explore the interactions amongst the degree of regularization, the distribution of biases in a network, and the network position of the innovator. We identify two regimes for the widespread adoption of arbritrary innovations, viewed as informational cascades in the network. With moderate regularization of experienced input, average people (not well-connected people) are the most likely source of successful innovations. Our results shed light on a major outstanding puzzle in the theory of language change. The framework also holds promise for understanding the dynamics of other social norms.

Computation and Language

Gap-weighted subsequences for automatic cognate identification and phylogenetic inference

In this paper, we describe the problem of cognate identification and its relation to phylogenetic inference. We introduce subsequence based features for discriminating cognates from non-cognates. We show that subsequence based features perform better than the state-of-the-art string similarity measures for the purpose of cognate identification. We use the cognate judgments for the purpose of phylogenetic inference and observe that these classifiers infer a tree which is close to the gold standard tree. The contribution of this paper is the use of subsequence features for cognate identification and to employ the cognate judgments for phylogenetic inference.

Computation and Language

Controlled Natural Language Processing as Answer Set Programming: an Experiment

Most controlled natural languages (CNLs) are processed with the help of a pipeline architecture that relies on different software components. We investigate in this paper in an experimental way how well answer set programming (ASP) is suited as a unifying framework for parsing a CNL, deriving a formal representation for the resulting syntax trees, and for reasoning with that representation. We start from a list of input tokens in ASP notation and show how this input can be transformed into a syntax tree using an ASP grammar and then into reified ASP rules in form of a set of facts. These facts are then processed by an ASP meta-interpreter that allows us to infer new knowledge.

Computation and Language

First-Pass Large Vocabulary Continuous Speech Recognition using Bi-Directional Recurrent DNNs

We present a method to perform first-pass large vocabulary continuous speech recognition using only a neural network and language model. Deep neural network acoustic models are now commonplace in HMM-based speech recognition systems, but building such systems is a complex, domain-specific task. Recent work demonstrated the feasibility of discarding the HMM sequence modeling framework by directly predicting transcript text from audio. This paper extends this approach in two ways. First, we demonstrate that a straightforward recurrent neural network architecture can achieve a high level of accuracy. Second, we propose and evaluate a modified prefix-search decoding algorithm. This approach to decoding enables first-pass speech recognition with a language model, completely unaided by the cumbersome infrastructure of HMM-based systems. Experiments on the Wall Street Journal corpus demonstrate fairly competitive word error rates, and the importance of bi-directional network recurrence.

Computation and Language

Detection is the central problem in real-word spelling correction

Real-word spelling correction differs from non-word spelling correction in its aims and its challenges. Here we show that the central problem in real-word spelling correction is detection. Methods from non-word spelling correction, which focus instead on selection among candidate corrections, do not address detection adequately, because detection is either assumed in advance or heavily constrained. As we demonstrate in this paper, merely discriminating between the intended word and a random close variation of it within the context of a sentence is a task that can be performed with high accuracy using straightforward models. Trigram models are sufficient in almost all cases. The difficulty comes when every word in the sentence is a potential error, with a large set of possible candidate corrections. Despite their strengths, trigram models cannot reliably find true errors without introducing many more, at least not when used in the obvious sequential way without added structure. The detection task exposes weakness not visible in the selection task.

Computation and Language

SimLex-999: Evaluating Semantic Models with (Genuine) Similarity Estimation

We present SimLex-999, a gold standard resource for evaluating distributional semantic models that improves on existing resources in several important ways. First, in contrast to gold standards such as WordSim-353 and MEN, it explicitly quantifies similarity rather than association or relatedness, so that pairs of entities that are associated but not actually similar [Freud, psychology] have a low rating. We show that, via this focus on similarity, SimLex-999 incentivizes the development of models with a different, and arguably wider range of applications than those which reflect conceptual association. Second, SimLex-999 contains a range of concrete and abstract adjective, noun and verb pairs, together with an independent rating of concreteness and (free) association strength for each pair. This diversity enables fine-grained analyses of the performance of models on concepts of different types, and consequently greater insight into how architectures can be improved. Further, unlike existing gold standard evaluations, for which automatic approaches have reached or surpassed the inter-annotator agreement ceiling, state-of-the-art models perform well below this ceiling on SimLex-999. There is therefore plenty of scope for SimLex-999 to quantify future improvements to distributional semantic models, guiding the development of the next generation of representation-learning architectures.

Computation and Language

Unsupervised Keyword Extraction from Polish Legal Texts

In this work, we present an application of the recently proposed unsupervised keyword extraction algorithm RAKE to a corpus of Polish legal texts from the field of public procurement. RAKE is essentially a language and domain independent method. Its only language-specific input is a stoplist containing a set of non-content words. The performance of the method heavily depends on the choice of such a stoplist, which should be domain adopted. Therefore, we complement RAKE algorithm with an automatic approach to selecting non-content words, which is based on the statistical properties of term distribution.

Computation and Language

On Detecting Messaging Abuse in Short Text Messages using Linguistic and Behavioral patterns

The use of short text messages in social media and instant messaging has become a popular communication channel during the last years. This rising popularity has caused an increment in messaging threats such as spam, phishing or malware as well as other threats. The processing of these short text message threats could pose additional challenges such as the presence of lexical variants, SMS-like contractions or advanced obfuscations which can degrade the performance of traditional filtering solutions. By using a real-world SMS data set from a large telecommunications operator from the US and a social media corpus, in this paper we analyze the effectiveness of machine learning filters based on linguistic and behavioral patterns in order to detect short text spam and abusive users in the network. We have also explored different ways to deal with short text message challenges such as tokenization and entity detection by using text normalization and substring clustering techniques. The obtained results show the validity of the proposed solution by enhancing baseline approaches.

Computation and Language

Be Careful When Assuming the Obvious: Commentary on "The placement of the head that minimizes online memory: a complex systems approach"

Ferrer-i-Cancho (2015) presents a mathematical model of both the synchronic and diachronic nature of word order based on the assumption that memory costs are a never decreasing function of distance and a few very general linguistic assumptions. However, even these minimal and seemingly obvious assumptions are not as safe as they appear in light of recent typological and psycholinguistic evidence. The interaction of word order and memory has further depths to be explored.

Computation and Language

Convolutional Neural Networks for Sentence Classification

We report on a series of experiments with convolutional neural networks (CNN) trained on top of pre-trained word vectors for sentence-level classification tasks. We show that a simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks. Learning task-specific vectors through fine-tuning offers further gains in performance. We additionally propose a simple modification to the architecture to allow for the use of both task-specific and static vectors. The CNN models discussed herein improve upon the state of the art on 4 out of 7 tasks, which include sentiment analysis and question classification.

Computation and Language

Evaluating Neural Word Representations in Tensor-Based Compositional Settings

We provide a comparative study between neural word representations and traditional vector spaces based on co-occurrence counts, in a number of compositional tasks. We use three different semantic spaces and implement seven tensor-based compositional models, which we then test (together with simpler additive and multiplicative approaches) in tasks involving verb disambiguation and sentence similarity. To check their scalability, we additionally evaluate the spaces using simple compositional methods on larger-scale tasks with less constrained language: paraphrase detection and dialogue act tagging. In the more constrained tasks, co-occurrence vectors are competitive, although choice of compositional method is important; on the larger-scale tasks, they are outperformed by neural word embeddings, which show robust, stable performance across the tasks.

Computation and Language

Resolving Lexical Ambiguity in Tensor Regression Models of Meaning

This paper provides a method for improving tensor-based compositional distributional models of meaning by the addition of an explicit disambiguation step prior to composition. In contrast with previous research where this hypothesis has been successfully tested against relatively simple compositional models, in our work we use a robust model trained with linear regression. The results we get in two experiments show the superiority of the prior disambiguation method and suggest that the effectiveness of this approach is model-independent.

Computation and Language

Non-Standard Words as Features for Text Categorization

This paper presents categorization of Croatian texts using Non-Standard Words (NSW) as features. Non-Standard Words are: numbers, dates, acronyms, abbreviations, currency, etc. NSWs in Croatian language are determined according to Croatian NSW taxonomy. For the purpose of this research, 390 text documents were collected and formed the SKIPEZ collection with 6 classes: official, literary, informative, popular, educational and scientific. Text categorization experiment was conducted on three different representations of the SKIPEZ collection: in the first representation, the frequencies of NSWs are used as features; in the second representation, the statistic measures of NSWs (variance, coefficient of variation, standard deviation, etc.) are used as features; while the third representation combines the first two feature sets. Naive Bayes, CN2, C4.5, kNN, Classification Trees and Random Forest algorithms were used in text categorization experiments. The best categorization results are achieved using the first feature set (NSW frequencies) with the categorization accuracy of 87%. This suggests that the NSWs should be considered as features in highly inflectional languages, such as Croatian. NSW based features reduce the dimensionality of the feature space without standard lemmatization procedures, and therefore the bag-of-NSWs should be considered for further Croatian texts categorization experiments.

Computation and Language

Strongly Incremental Repair Detection

We present STIR (STrongly Incremental Repair detection), a system that detects speech repairs and edit terms on transcripts incrementally with minimal latency. STIR uses information-theoretic measures from n-gram models as its principal decision features in a pipeline of classifiers detecting the different stages of repairs. Results on the Switchboard disfluency tagged corpus show utterance-final accuracy on a par with state-of-the-art incremental repair detection methods, but with better incremental accuracy, faster time-to-detection and less computational overhead. We evaluate its performance using incremental metrics and propose new repair processing evaluation standards.

Computation and Language

Empirical Evaluation of Tree distances for Parser Evaluation

In this empirical study, I compare various tree distance measures -- originally developed in computational biology for the purpose of tree comparison -- for the purpose of parser evaluation. I will control for the parser setting by comparing the automatically generated parse trees from the state-of-the-art parser Charniak, 2000) with the gold-standard parse trees. The article describes two different tree distance measures (RF and QD) along with its variants (GRF and GQD) for the purpose of parser evaluation. The article will argue that RF measure captures similar information as the standard EvalB metric (Sekine and Collins, 1997) and the tree edit distance (Zhang and Shasha, 1989) applied by Tsarfaty et al. (2011). Finally, the article also provides empirical evidence by reporting high correlations between the different tree distances and EvalB metric's scores.

Computation and Language

Neural Machine Translation by Jointly Learning to Align and Translate

Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural machine translation often belong to a family of encoder-decoders and consists of an encoder that encodes a source sentence into a fixed-length vector from which a decoder generates a translation. In this paper, we conjecture that the use of a fixed-length vector is a bottleneck in improving the performance of this basic encoder-decoder architecture, and propose to extend this by allowing a model to automatically (soft-)search for parts of a source sentence that are relevant to predicting a target word, without having to form these parts as a hard segment explicitly. With this new approach, we achieve a translation performance comparable to the existing state-of-the-art phrase-based system on the task of English-to-French translation. Furthermore, qualitative analysis reveals that the (soft-)alignments found by the model agree well with our intuition.

Computation and Language

Overcoming the Curse of Sentence Length for Neural Machine Translation using Automatic Segmentation

The authors of (Cho et al., 2014a) have shown that the recently introduced neural network translation systems suffer from a significant drop in translation quality when translating long sentences, unlike existing phrase-based translation systems. In this paper, we propose a way to address this issue by automatically segmenting an input sentence into phrases that can be easily translated by the neural network translation model. Once each segment has been independently translated by the neural machine translation model, the translated clauses are concatenated to form a final translation. Empirical results show a significant improvement in translation quality for long sentences.

Computation and Language

On the Properties of Neural Machine Translation: Encoder-Decoder Approaches

Neural machine translation is a relatively new approach to statistical machine translation based purely on neural networks. The neural machine translation models often consist of an encoder and a decoder. The encoder extracts a fixed-length representation from a variable-length input sentence, and the decoder generates a correct translation from this representation. In this paper, we focus on analyzing the properties of the neural machine translation using two models; RNN Encoder--Decoder and a newly proposed gated recursive convolutional neural network. We show that the neural machine translation performs relatively well on short sentences without unknown words, but its performance degrades rapidly as the length of the sentence and the number of unknown words increase. Furthermore, we find that the proposed gated recursive convolutional network learns a grammatical structure of a sentence automatically.

Computation and Language

Semantic clustering of Russian web search results: possibilities and problems

The paper deals with word sense induction from lexical co-occurrence graphs. We construct such graphs on large Russian corpora and then apply this data to cluster Mail.ru Search results according to meanings of the query. We compare different methods of performing such clustering and different source corpora. Models of applying distributional semantics to big linguistic data are described.

Computation and Language

An NLP Assistant for Clide

This report describes an NLP assistant for the collaborative development environment Clide, that supports the development of NLP applications by providing easy access to some common NLP data structures. The assistant visualizes text fragments and their dependencies by displaying the semantic graph of a sentence, the coreference chain of a paragraph and mined triples that are extracted from a paragraph's semantic graphs and linked using its coreference chain. Using this information and a logic programming library, we create an NLP database which is used by a series of queries to mine the triples. The algorithm is tested by translating a natural language text describing a graph to an actual graph that is shown as an annotation in the text editor.

Computation and Language

Analyzing the Language of Food on Social Media

We investigate the predictive power behind the language of food on social media. We collect a corpus of over three million food-related posts from Twitter and demonstrate that many latent population characteristics can be directly predicted from this data: overweight rate, diabetes rate, political leaning, and home geographical location of authors. For all tasks, our language-based models significantly outperform the majority-class baselines. Performance is further improved with more complex natural language processing, such as topic modeling. We analyze which textual features have most predictive power for these datasets, providing insight into the connections between the language of food, geographic locale, and community characteristics. Lastly, we design and implement an online system for real-time query and visualization of the dataset. Visualization tools, such as geo-referenced heatmaps, semantics-preserving wordclouds and temporal histograms, allow us to discover more complex, global patterns mirrored in the language of food.

Computation and Language

Approximating solution structure of the Weighted Sentence Alignment problem

We study the complexity of approximating solution structure of the bijective weighted sentence alignment problem of DeNero and Klein (2008). In particular, we consider the complexity of finding an alignment that has a significant overlap with an optimal alignment. We discuss ways of representing the solution for the general weighted sentence alignment as well as phrases-to-words alignment problem, and show that computing a string which agrees with the optimal sentence partition on more than half (plus an arbitrarily small polynomial fraction) positions for the phrases-to-words alignment is NP-hard. For the general weighted sentence alignment we obtain such bound from the agreement on a little over 2/3 of the bits. Additionally, we generalize the Hamming distance approximation of a solution structure to approximating it with respect to the edit distance metric, obtaining similar lower bounds.

Computation and Language

A Study of Association Measures and their Combination for Arabic MWT Extraction

Automatic Multi-Word Term (MWT) extraction is a very important issue to many applications, such as information retrieval, question answering, and text categorization. Although many methods have been used for MWT extraction in English and other European languages, few studies have been applied to Arabic. In this paper, we propose a novel, hybrid method which combines linguistic and statistical approaches for Arabic Multi-Word Term extraction. The main contribution of our method is to consider contextual information and both termhood and unithood for association measures at the statistical filtering step. In addition, our technique takes into account the problem of MWT variation in the linguistic filtering step. The performance of the proposed statistical measure (NLC-value) is evaluated using an Arabic environment corpus by comparing it with some existing competitors. Experimental results show that our NLC-value measure outperforms the other ones in term of precision for both bi-grams and tri-grams.

Computation and Language

Sequence to Sequence Learning with Neural Networks

Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they cannot be used to map sequences to sequences. In this paper, we present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure. Our method uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to a vector of a fixed dimensionality, and then another deep LSTM to decode the target sequence from the vector. Our main result is that on an English to French translation task from the WMT'14 dataset, the translations produced by the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did not have difficulty on long sentences. For comparison, a phrase-based SMT system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM to rerank the 1000 hypotheses produced by the aforementioned SMT system, its BLEU score increases to 36.5, which is close to the previous best result on this task. The LSTM also learned sensible phrase and sentence representations that are sensitive to word order and are relatively invariant to the active and the passive voice. Finally, we found that reversing the order of the words in all source sentences (but not target sentences) improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier.

Computation and Language

Word Sense Disambiguation using WSD specific Wordnet of Polysemy Words

This paper presents a new model of WordNet that is used to disambiguate the correct sense of polysemy word based on the clue words. The related words for each sense of a polysemy word as well as single sense word are referred to as the clue words. The conventional WordNet organizes nouns, verbs, adjectives and adverbs together into sets of synonyms called synsets each expressing a different concept. In contrast to the structure of WordNet, we developed a new model of WordNet that organizes the different senses of polysemy words as well as the single sense words based on the clue words. These clue words for each sense of a polysemy word as well as for single sense word are used to disambiguate the correct meaning of the polysemy word in the given context using knowledge based Word Sense Disambiguation (WSD) algorithms. The clue word can be a noun, verb, adjective or adverb.

Computation and Language

Incorporating Semi-supervised Features into Discontinuous Easy-First Constituent Parsing

This paper describes adaptations for EaFi, a parser for easy-first parsing of discontinuous constituents, to adapt it to multiple languages as well as make use of the unlabeled data that was provided as part of the SPMRL shared task 2014.

Computation and Language

Text mixing shapes the anatomy of rank-frequency distributions: A modern Zipfian mechanics for natural language

Natural languages are full of rules and exceptions. One of the most famous quantitative rules is Zipf's law which states that the frequency of occurrence of a word is approximately inversely proportional to its rank. Though this `law' of ranks has been found to hold across disparate texts and forms of data, analyses of increasingly large corpora over the last 15 years have revealed the existence of two scaling regimes. These regimes have thus far been explained by a hypothesis suggesting a separability of languages into core and non-core lexica. Here, we present and defend an alternative hypothesis, that the two scaling regimes result from the act of aggregating texts. We observe that text mixing leads to an effective decay of word introduction, which we show provides accurate predictions of the location and severity of breaks in scaling. Upon examining large corpora from 10 languages in the Project Gutenberg eBooks collection (eBooks), we find emphatic empirical support for the universality of our claim.

Computation and Language

An Approach to Reducing Annotation Costs for BioNLP

There is a broad range of BioNLP tasks for which active learning (AL) can significantly reduce annotation costs and a specific AL algorithm we have developed is particularly effective in reducing annotation costs for these tasks. We have previously developed an AL algorithm called ClosestInitPA that works best with tasks that have the following characteristics: redundancy in training material, burdensome annotation costs, Support Vector Machines (SVMs) work well for the task, and imbalanced datasets (i.e. when set up as a binary classification problem, one class is substantially rarer than the other). Many BioNLP tasks have these characteristics and thus our AL algorithm is a natural approach to apply to BioNLP tasks.

Computation and Language

Polarity detection movie reviews in hindi language

Nowadays peoples are actively involved in giving comments and reviews on social networking websites and other websites like shopping websites, news websites etc. large number of people everyday share their opinion on the web, results is a large number of user data is collected .users also find it trivial task to read all the reviews and then reached into the decision. It would be better if these reviews are classified into some category so that the user finds it easier to read. Opinion Mining or Sentiment Analysis is a natural language processing task that mines information from various text forms such as reviews, news, and blogs and classify them on the basis of their polarity as positive, negative or neutral. But, from the last few years, user content in Hindi language is also increasing at a rapid rate on the Web. So it is very important to perform opinion mining in Hindi language as well. In this paper a Hindi language opinion mining system is proposed. The system classifies the reviews as positive, negative and neutral for Hindi language. Negation is also handled in the proposed system. Experimental results using reviews of movies show the effectiveness of the system

Computation and Language

An Algorithm Based on Empirical Methods, for Text-to-Tuneful-Speech Synthesis of Sanskrit Verse

The rendering of Sanskrit poetry from text to speech is a problem that has not been solved before. One reason may be the complications in the language itself. We present unique algorithms based on extensive empirical analysis, to synthesize speech from a given text input of Sanskrit verses. Using a pre-recorded audio units database which is itself tremendously reduced in size compared to the colossal size that would otherwise be required, the algorithms work on producing the best possible, tunefully rendered chanting of the given verse. His would enable the visually impaired and those with reading disabilities to easily access the contents of Sanskrit verses otherwise available only in writing.

Computation and Language

A Binary Schema and Computational Algorithms to Process Vowel-based Euphonic Conjunctions for Word Searches

Comprehensively searching for words in Sanskrit E-text is a non-trivial problem because words could change their forms in different contexts. One such context is sandhi or euphonic conjunctions, which cause a word to change owing to the presence of adjacent letters or words. The change wrought by these possible conjunctions can be so significant in Sanskrit that a simple search for the word in its given form alone can significantly reduce the success level of the search. This work presents a representational schema that represents letters in a binary format and reduces Paninian rules of euphonic conjunctions to simple bit set-unset operations. The work presents an efficient algorithm to process vowel-based sandhis using this schema. It further presents another algorithm that uses the sandhi processor to generate the possible transformed word forms of a given word to use in a comprehensive word search.

Computation and Language

Computational Algorithms Based on the Paninian System to Process Euphonic Conjunctions for Word Searches

Searching for words in Sanskrit E-text is a problem that is accompanied by complexities introduced by features of Sanskrit such as euphonic conjunctions or sandhis. A word could occur in an E-text in a transformed form owing to the operation of rules of sandhi. Simple word search would not yield these transformed forms of the word. Further, there is no search engine in the literature that can comprehensively search for words in Sanskrit E-texts taking euphonic conjunctions into account. This work presents an optimal binary representational schema for letters of the Sanskrit alphabet along with algorithms to efficiently process the sandhi rules of Sanskrit grammar. The work further presents an algorithm that uses the sandhi processing algorithm to perform a comprehensive word search on E-text.

Computation and Language

Voting for Deceptive Opinion Spam Detection

Consumers' purchase decisions are increasingly influenced by user-generated online reviews. Accordingly, there has been growing concern about the potential for posting deceptive opinion spam fictitious reviews that have been deliberately written to sound authentic, to deceive the readers. Existing approaches mainly focus on developing automatic supervised learning based methods to help users identify deceptive opinion spams. This work, we used the LSI and Sprinkled LSI technique to reduce the dimension for deception detection. We make our contribution to demonstrate what LSI is capturing in latent semantic space and reveal how deceptive opinions can be recognized automatically from truthful opinions. Finally, we proposed a voting scheme which integrates different approaches to further improve the classification performance.

Computation and Language

Lexical Normalisation of Twitter Data

Twitter with over 500 million users globally, generates over 100,000 tweets per minute . The 140 character limit per tweet, perhaps unintentionally, encourages users to use shorthand notations and to strip spellings to their bare minimum "syllables" or elisions e.g. "srsly". The analysis of twitter messages which typically contain misspellings, elisions, and grammatical errors, poses a challenge to established Natural Language Processing (NLP) tools which are generally designed with the assumption that the data conforms to the basic grammatical structure commonly used in English language. In order to make sense of Twitter messages it is necessary to first transform them into a canonical form, consistent with the dictionary or grammar. This process, performed at the level of individual tokens ("words"), is called lexical normalisation. This paper investigates various techniques for lexical normalisation of Twitter data and presents the findings as the techniques are applied to process raw data from Twitter.

Computation and Language

Modeling the average shortest path length in growth of word-adjacency networks

We investigate properties of evolving linguistic networks defined by the word-adjacency relation. Such networks belong to the category of networks with accelerated growth but their shortest path length appears to reveal the network size dependence of different functional form than the ones known so far. We thus compare the networks created from literary texts with their artificial substitutes based on different variants of the Dorogovtsev-Mendes model and observe that none of them is able to properly simulate the novel asymptotics of the shortest path length. Then, we identify the local chain-like linear growth induced by grammar and style as a missing element in this model and extend it by incorporating such effects. It is in this way that a satisfactory agreement with the empirical result is obtained.

Computation and Language

Using crowdsourcing system for creating site-specific statistical machine translation engine

A crowdsourcing translation approach is an effective tool for globalization of site content, but it is also an important source of parallel linguistic data. For the given site, processed with a crowdsourcing system, a sentence-aligned corpus can be fetched, which covers a very narrow domain of terminology and language patterns - a site-specific domain. These data can be used for training and estimation of site-specific statistical machine translation engine

Computation and Language

Semantically-Informed Syntactic Machine Translation: A Tree-Grafting Approach

We describe a unified and coherent syntactic framework for supporting a semantically-informed syntactic approach to statistical machine translation. Semantically enriched syntactic tags assigned to the target-language training texts improved translation quality. The resulting system significantly outperformed a linguistically naive baseline model (Hiero), and reached the highest scores yet reported on the NIST 2009 Urdu-English translation task. This finding supports the hypothesis (posed by many researchers in the MT community, e.g., in DARPA GALE) that both syntactic and semantic information are critical for improving translation quality---and further demonstrates that large gains can be achieved for low-resource languages with different word order than English.

Computation and Language

The meaning-frequency law in Zipfian optimization models of communication

According to Zipf's meaning-frequency law, words that are more frequent tend to have more meanings. Here it is shown that a linear dependency between the frequency of a form and its number of meanings is found in a family of models of Zipf's law for word frequencies. This is evidence for a weak version of the meaning-frequency law. Interestingly, that weak law (a) is not an inevitable of property of the assumptions of the family and (b) is found at least in the narrow regime where those models exhibit Zipf's law for word frequencies.

Computation and Language

Performance of Stanford and Minipar Parser on Biomedical Texts

In this paper, the performance of two dependency parsers, namely Stanford and Minipar, on biomedical texts has been reported. The performance of te parsers to assignm dependencies between two biomedical concepts that are already proved to be connected is not satisfying. Both Stanford and Minipar, being statistical parsers, fail to assign dependency relation between two connected concepts if they are distant by at least one clause. Minipar's performance, in terms of precision, recall and the F-score of the attachment score (e.g., correctly identified head in a dependency), to parse biomedical text is also measured taking the Stanford's as a gold standard. The results suggest that Minipar is not suitable yet to parse biomedical texts. In addition, a qualitative investigation reveals that the difference between working principles of the parsers also play a vital role for Minipar's degraded performance.

Computation and Language

Topic Similarity Networks: Visual Analytics for Large Document Sets

We investigate ways in which to improve the interpretability of LDA topic models by better analyzing and visualizing their outputs. We focus on examining what we refer to as topic similarity networks: graphs in which nodes represent latent topics in text collections and links represent similarity among topics. We describe efficient and effective approaches to both building and labeling such networks. Visualizations of topic models based on these networks are shown to be a powerful means of exploring, characterizing, and summarizing large collections of unstructured text documents. They help to "tease out" non-obvious connections among different sets of documents and provide insights into how topics form larger themes. We demonstrate the efficacy and practicality of these approaches through two case studies: 1) NSF grants for basic research spanning a 14 year period and 2) the entire English portion of Wikipedia.

Computation and Language

Semi-supervised Classification for Natural Language Processing

Semi-supervised classification is an interesting idea where classification models are learned from both labeled and unlabeled data. It has several advantages over supervised classification in natural language processing domain. For instance, supervised classification exploits only labeled data that are expensive, often difficult to get, inadequate in quantity, and require human experts for annotation. On the other hand, unlabeled data are inexpensive and abundant. Despite the fact that many factors limit the wide-spread use of semi-supervised classification, it has become popular since its level of performance is empirically as good as supervised classification. This study explores the possibilities and achievements as well as complexity and limitations of semi-supervised classification for several natural langue processing tasks like parsing, biomedical information processing, text classification, and summarization.

Computation and Language

Generating Conceptual Metaphors from Proposition Stores

Contemporary research on computational processing of linguistic metaphors is divided into two main branches: metaphor recognition and metaphor interpretation. We take a different line of research and present an automated method for generating conceptual metaphors from linguistic data. Given the generated conceptual metaphors, we find corresponding linguistic metaphors in corpora. In this paper, we describe our approach and its evaluation using English and Russian data.

Computation and Language

The Utility of Text: The Case of Amicus Briefs and the Supreme Court

We explore the idea that authoring a piece of text is an act of maximizing one's expected utility. To make this idea concrete, we consider the societally important decisions of the Supreme Court of the United States. Extensive past work in quantitative political science provides a framework for empirically modeling the decisions of justices and how they relate to text. We incorporate into such a model texts authored by amici curiae ("friends of the court" separate from the litigants) who seek to weigh in on the decision, then explicitly model their goals in a random utility model. We demonstrate the benefits of this approach in improved vote prediction and the ability to perform counterfactual analysis.

Computation and Language

CRF-based Named Entity Recognition @ICON 2013

This paper describes performance of CRF based systems for Named Entity Recognition (NER) in Indian language as a part of ICON 2013 shared task. In this task we have considered a set of language independent features for all the languages. Only for English a language specific feature, i.e. capitalization, has been added. Next the use of gazetteer is explored for Bengali, Hindi and English. The gazetteers are built from Wikipedia and other sources. Test results show that the system achieves the highest F measure of 88% for English and the lowest F measure of 69% for both Tamil and Telugu. Note that for the least performing two languages no gazetteer was used. NER in Bengali and Hindi finds accuracy (F measure) of 87% and 79%, respectively.

Computation and Language

A Deep Learning Approach to Data-driven Parameterizations for Statistical Parametric Speech Synthesis

Nearly all Statistical Parametric Speech Synthesizers today use Mel Cepstral coefficients as the vocal tract parameterization of the speech signal. Mel Cepstral coefficients were never intended to work in a parametric speech synthesis framework, but as yet, there has been little success in creating a better parameterization that is more suited to synthesis. In this paper, we use deep learning algorithms to investigate a data-driven parameterization technique that is designed for the specific requirements of synthesis. We create an invertible, low-dimensional, noise-robust encoding of the Mel Log Spectrum by training a tapered Stacked Denoising Autoencoder (SDA). This SDA is then unwrapped and used as the initialization for a Multi-Layer Perceptron (MLP). The MLP is fine-tuned by training it to reconstruct the input at the output layer. This MLP is then split down the middle to form encoding and decoding networks. These networks produce a parameterization of the Mel Log Spectrum that is intended to better fulfill the requirements of synthesis. Results are reported for experiments conducted using this resulting parameterization with the ClusterGen speech synthesizer.

Computation and Language

Improving the Performance of English-Tamil Statistical Machine Translation System using Source-Side Pre-Processing

Machine Translation is one of the major oldest and the most active research area in Natural Language Processing. Currently, Statistical Machine Translation (SMT) dominates the Machine Translation research. Statistical Machine Translation is an approach to Machine Translation which uses models to learn translation patterns directly from data, and generalize them to translate a new unseen text. The SMT approach is largely language independent, i.e. the models can be applied to any language pair. Statistical Machine Translation (SMT) attempts to generate translations using statistical methods based on bilingual text corpora. Where such corpora are available, excellent results can be attained translating similar texts, but such corpora are still not available for many language pairs. Statistical Machine Translation systems, in general, have difficulty in handling the morphology on the source or the target side especially for morphologically rich languages. Errors in morphology or syntax in the target language can have severe consequences on meaning of the sentence. They change the grammatical function of words or the understanding of the sentence through the incorrect tense information in verb. Baseline SMT also known as Phrase Based Statistical Machine Translation (PBSMT) system does not use any linguistic information and it only operates on surface word form. Recent researches shown that adding linguistic information helps to improve the accuracy of the translation with less amount of bilingual corpora. Adding linguistic information can be done using the Factored Statistical Machine Translation system through pre-processing steps. This paper investigates about how English side pre-processing is used to improve the accuracy of English-Tamil SMT system.

Computation and Language

LAF-Fabric: a data analysis tool for Linguistic Annotation Framework with an application to the Hebrew Bible

The Linguistic Annotation Framework (LAF) provides a general, extensible stand-off markup system for corpora. This paper discusses LAF-Fabric, a new tool to analyse LAF resources in general with an extension to process the Hebrew Bible in particular. We first walk through the history of the Hebrew Bible as text database in decennium-wide steps. Then we describe how LAF-Fabric may serve as an analysis tool for this corpus. Finally, we describe three analytic projects/workflows that benefit from the new LAF representation: 1) the study of linguistic variation: extract cooccurrence data of common nouns between the books of the Bible (Martijn Naaijer); 2) the study of the grammar of Hebrew poetry in the Psalms: extract clause typology (Gino Kalkman); 3) construction of a parser of classical Hebrew by Data Oriented Parsing: generate tree structures from the database (Andreas van Cranenburgh).

Computation and Language

A Morphological Analyzer for Japanese Nouns, Verbs and Adjectives

We present an open source morphological analyzer for Japanese nouns, verbs and adjectives. The system builds upon the morphological analyzing capabilities of MeCab to incorporate finer details of classification such as politeness, tense, mood and voice attributes. We implemented our analyzer in the form of a finite state transducer using the open source finite state compiler FOMA toolkit. The source code and tool is available at https://bitbucket.org/skylander/yc-nlplab/.

Computation and Language

Not All Neural Embeddings are Born Equal

Neural language models learn word representations that capture rich linguistic and conceptual information. Here we investigate the embeddings learned by neural machine translation models. We show that translation-based embeddings outperform those learned by cutting-edge monolingual models at single-language tasks requiring knowledge of conceptual similarity and/or syntactic role. The findings suggest that, while monolingual models learn information about how concepts are related, neural-translation models better capture their true ontological status.

Computation and Language

Generating abbreviations using Google Books library

The article describes the original method of creating a dictionary of abbreviations based on the Google Books Ngram Corpus. The dictionary of abbreviations is designed for Russian, yet as its methodology is universal it can be applied to any language. The dictionary can be used to define the function of the period during text segmentation in various applied systems of text processing. The article describes difficulties encountered in the process of its construction as well as the ways to overcome them. A model of evaluating a probability of first and second type errors (extraction accuracy and fullness) is constructed. Certain statistical data for the use of abbreviations are provided.

Computation and Language

Corpora Preparation and Stopword List Generation for Arabic data in Social Network

This paper proposes a methodology to prepare corpora in Arabic language from online social network (OSN) and review site for Sentiment Analysis (SA) task. The paper also proposes a methodology for generating a stopword list from the prepared corpora. The aim of the paper is to investigate the effect of removing stopwords on the SA task. The problem is that the stopwords lists generated before were on Modern Standard Arabic (MSA) which is not the common language used in OSN. We have generated a stopword list of Egyptian dialect and a corpus-based list to be used with the OSN corpora. We compare the efficiency of text classification when using the generated lists along with previously generated lists of MSA and combining the Egyptian dialect list with the MSA list. The text classification was performed using Na\"ive Bayes and Decision Tree classifiers and two feature selection approaches, unigrams and bigram. The experiments show that the general lists containing the Egyptian dialects words give better performance than using lists of MSA stopwords only.

Computation and Language

Supervised learning Methods for Bangla Web Document Categorization

This paper explores the use of machine learning approaches, or more specifically, four supervised learning Methods, namely Decision Tree(C 4.5), K-Nearest Neighbour (KNN), Na\"ive Bays (NB), and Support Vector Machine (SVM) for categorization of Bangla web documents. This is a task of automatically sorting a set of documents into categories from a predefined set. Whereas a wide range of methods have been applied to English text categorization, relatively few studies have been conducted on Bangla language text categorization. Hence, we attempt to analyze the efficiency of those four methods for categorization of Bangla documents. In order to validate, Bangla corpus from various websites has been developed and used as examples for the experiment. For Bangla, empirical results support that all four methods produce satisfactory performance with SVM attaining good result in terms of high dimensional and relatively noisy document feature vectors.

Computation and Language

Contrastive Unsupervised Word Alignment with Non-Local Features

Word alignment is an important natural language processing task that indicates the correspondence between natural languages. Recently, unsupervised learning of log-linear models for word alignment has received considerable attention as it combines the merits of generative and discriminative approaches. However, a major challenge still remains: it is intractable to calculate the expectations of non-local features that are critical for capturing the divergence between natural languages. We propose a contrastive approach that aims to differentiate observed training examples from noises. It not only introduces prior knowledge to guide unsupervised learning but also cancels out partition functions. Based on the observation that the probability mass of log-linear models for word alignment is usually highly concentrated, we propose to use top-n alignments to approximate the expectations with respect to posterior distributions. This allows for efficient and accurate calculation of expectations of non-local features. Experiments show that our approach achieves significant improvements over state-of-the-art unsupervised word alignment methods.

Computation and Language

Language-based Examples in the Statistics Classroom

Statistics pedagogy values using a variety of examples. Thanks to text resources on the Web, and since statistical packages have the ability to analyze string data, it is now easy to use language-based examples in a statistics class. Three such examples are discussed here. First, many types of wordplay (e.g., crosswords and hangman) involve finding words with letters that satisfy a certain pattern. Second, linguistics has shown that idiomatic pairs of words often appear together more frequently than chance. For example, in the Brown Corpus, this is true of the phrasal verb to throw up (p-value=7.92E-10.) Third, a pangram contains all the letters of the alphabet at least once. These are searched for in Charles Dickens' A Christmas Carol, and their lengths are compared to the expected value given by the unequal probability coupon collector's problem as well as simulations.

Computation and Language

Spatial Diffuseness Features for DNN-Based Speech Recognition in Noisy and Reverberant Environments

We propose a spatial diffuseness feature for deep neural network (DNN)-based automatic speech recognition to improve recognition accuracy in reverberant and noisy environments. The feature is computed in real-time from multiple microphone signals without requiring knowledge or estimation of the direction of arrival, and represents the relative amount of diffuse noise in each time and frequency bin. It is shown that using the diffuseness feature as an additional input to a DNN-based acoustic model leads to a reduced word error rate for the REVERB challenge corpus, both compared to logmelspec features extracted from noisy signals, and features enhanced by spectral subtraction.

Computation and Language

Hybrid approaches for automatic vowelization of Arabic texts

Hybrid approaches for automatic vowelization of Arabic texts are presented in this article. The process is made up of two modules. In the first one, a morphological analysis of the text words is performed using the open source morphological Analyzer AlKhalil Morpho Sys. Outputs for each word analyzed out of context, are its different possible vowelizations. The integration of this Analyzer in our vowelization system required the addition of a lexical database containing the most frequent words in Arabic language. Using a statistical approach based on two hidden Markov models (HMM), the second module aims to eliminate the ambiguities. Indeed, for the first HMM, the unvowelized Arabic words are the observed states and the vowelized words are the hidden states. The observed states of the second HMM are identical to those of the first, but the hidden states are the lists of possible diacritics of the word without its Arabic letters. Our system uses Viterbi algorithm to select the optimal path among the solutions proposed by Al Khalil Morpho Sys. Our approach opens an important way to improve the performance of automatic vowelization of Arabic texts for other uses in automatic natural language processing.

Computation and Language

An Ontology for Comprehensive Tutoring of Euphonic Conjunctions of Sanskrit Grammar

Euphonic conjunctions (sandhis) form a very important aspect of Sanskrit morphology and phonology. The traditional and modern methods of studying about euphonic conjunctions in Sanskrit follow different methodologies. The former involves a rigorous study of the Paninian system embodied in Panini's Ashtadhyayi, while the latter usually involves the study of a few important sandhi rules with the use of examples. The former is not suitable for beginners, and the latter, not sufficient to gain a comprehensive understanding of the operation of sandhi rules. This is so since there are not only numerous sandhi rules and exceptions, but also complex precedence rules involved. The need for a new ontology for sandhi-tutoring was hence felt. This work presents a comprehensive ontology designed to enable a student-user to learn in stages all about euphonic conjunctions and the relevant aphorisms of Sanskrit grammar and to test and evaluate the progress of the student-user. The ontology forms the basis of a multimedia sandhi tutor that was given to different categories of users including Sanskrit scholars for extensive and rigorous testing.

Computation and Language

Sentiment Analysis based on User Tag for Traditional Chinese Medicine in Weibo

With the acceptance of Western culture and science, Traditional Chinese Medicine (TCM) has become a controversial issue in China. So, it's important to study the public's sentiment and opinion on TCM. The rapid development of online social network, such as twitter, make it convenient and efficient to sample hundreds of millions of people for the aforementioned sentiment study. To the best of our knowledge, the present work is the first attempt that applies sentiment analysis to the domain of TCM on Sina Weibo (a twitter-like microblogging service in China). In our work, firstly we collect tweets topic about TCM from Sina Weibo, and label the tweets as supporting TCM and opposing TCM automatically based on user tag. Then, a support vector machine classifier has been built to predict the sentiment of TCM tweets without labels. Finally, we present a method to adjust the classifier result. The performance of F-measure attained with our method is 97%.

Computation and Language

POLYGLOT-NER: Massive Multilingual Named Entity Recognition

The increasing diversity of languages used on the web introduces a new level of complexity to Information Retrieval (IR) systems. We can no longer assume that textual content is written in one language or even the same language family. In this paper, we demonstrate how to build massive multilingual annotators with minimal human expertise and intervention. We describe a system that builds Named Entity Recognition (NER) annotators for 40 major languages using Wikipedia and Freebase. Our approach does not require NER human annotated datasets or language specific resources like treebanks, parallel corpora, and orthographic rules. The novelty of approach lies therein - using only language agnostic techniques, while achieving competitive performance. Our method learns distributed word representations (word embeddings) which encode semantic and syntactic features of words in each language. Then, we automatically generate datasets from Wikipedia link structure and Freebase attributes. Finally, we apply two preprocessing stages (oversampling and exact surface form matching) which do not require any linguistic expertise. Our evaluation is two fold: First, we demonstrate the system performance on human annotated datasets. Second, for languages where no gold-standard benchmarks are available, we propose a new method, distant evaluation, based on statistical machine translation.

Computation and Language

Learning Distributed Word Representations for Natural Logic Reasoning

Natural logic offers a powerful relational conception of meaning that is a natural counterpart to distributed semantic representations, which have proven valuable in a wide range of sophisticated language tasks. However, it remains an open question whether it is possible to train distributed representations to support the rich, diverse logical reasoning captured by natural logic. We address this question using two neural network-based models for learning embeddings: plain neural networks and neural tensor networks. Our experiments evaluate the models' ability to learn the basic algebra of natural logic relations from simulated data and from the WordNet noun graph. The overall positive results are promising for the future of learned distributed representations in the applied modeling of logical semantics.

Computation and Language

Constructing Long Short-Term Memory based Deep Recurrent Neural Networks for Large Vocabulary Speech Recognition

Long short-term memory (LSTM) based acoustic modeling methods have recently been shown to give state-of-the-art performance on some speech recognition tasks. To achieve a further performance improvement, in this research, deep extensions on LSTM are investigated considering that deep hierarchical model has turned out to be more efficient than a shallow one. Motivated by previous research on constructing deep recurrent neural networks (RNNs), alternative deep LSTM architectures are proposed and empirically evaluated on a large vocabulary conversational telephone speech recognition task. Meanwhile, regarding to multi-GPU devices, the training process for LSTM networks is introduced and discussed. Experimental results demonstrate that the deep LSTM networks benefit from the depth and yield the state-of-the-art performance on this task.

Computation and Language

Patterns in the English Language: Phonological Networks, Percolation and Assembly Models

In this paper we provide a quantitative framework for the study of phonological networks (PNs) for the English language by carrying out principled comparisons to null models, either based on site percolation, randomization techniques, or network growth models. In contrast to previous work, we mainly focus on null models that reproduce lower order characteristics of the empirical data. We find that artificial networks matching connectivity properties of the English PN are exceedingly rare: this leads to the hypothesis that the word repertoire might have been assembled over time by preferentially introducing new words which are small modifications of old words. Our null models are able to explain the "power-law-like" part of the degree distributions and generally retrieve qualitative features of the PN such as high clustering, high assortativity coefficient, and small-world characteristics. However, the detailed comparison to expectations from null models also points out significant differences, suggesting the presence of additional constraints in word assembly. Key constraints we identify are the avoidance of large degrees, the avoidance of triadic closure, and the avoidance of large non-percolating clusters.

Computation and Language

Dependent Types for Pragmatics

This paper proposes the use of dependent types for pragmatic phenomena such as pronoun binding and presupposition resolution as a type-theoretic alternative to formalisms such as Discourse Representation Theory and Dynamic Semantics.

Computation and Language

Arabic Language Text Classification Using Dependency Syntax-Based Feature Selection

We study the performance of Arabic text classification combining various techniques: (a) tfidf vs. dependency syntax, for feature selection and weighting; (b) class association rules vs. support vector machines, for classification. The Arabic text is used in two forms: rootified and lightly stemmed. The results we obtain show that lightly stemmed text leads to better performance than rootified text; that class association rules are better suited for small feature sets obtained by dependency syntax constraints; and, finally, that support vector machines are better suited for large feature sets based on morphological feature selection criteria.

Computation and Language

A Modality Lexicon and its use in Automatic Tagging

This paper describes our resource-building results for an eight-week JHU Human Language Technology Center of Excellence Summer Camp for Applied Language Exploration (SCALE-2009) on Semantically-Informed Machine Translation. Specifically, we describe the construction of a modality annotation scheme, a modality lexicon, and two automated modality taggers that were built using the lexicon and annotation scheme. Our annotation scheme is based on identifying three components of modality: a trigger, a target and a holder. We describe how our modality lexicon was produced semi-automatically, expanding from an initial hand-selected list of modality trigger words and phrases. The resulting expanded modality lexicon is being made publicly available. We demonstrate that one tagger---a structure-based tagger---results in precision around 86% (depending on genre) for tagging of a standard LDC data set. In a machine translation application, using the structure-based tagger to annotate English modalities on an English-Urdu training corpus improved the translation quality score for Urdu by 0.3 Bleu points in the face of sparse training data.

Computation and Language

The Visualization of Change in Word Meaning over Time using Temporal Word Embeddings

We describe a visualization tool that can be used to view the change in meaning of words over time. The tool makes use of existing (static) word embedding datasets together with a timestamped $n$-gram corpus to create {\em temporal} word embeddings.

Computation and Language

A stronger null hypothesis for crossing dependencies

The syntactic structure of a sentence can be modeled as a tree where vertices are words and edges indicate syntactic dependencies between words. It is well-known that those edges normally do not cross when drawn over the sentence. Here a new null hypothesis for the number of edge crossings of a sentence is presented. That null hypothesis takes into account the length of the pair of edges that may cross and predicts the relative number of crossings in random trees with a small error, suggesting that a ban of crossings or a principle of minimization of crossings are not needed in general to explain the origins of non-crossing dependencies. Our work paves the way for more powerful null hypotheses to investigate the origins of non-crossing dependencies in nature.

Computation and Language

Using Mechanical Turk to Build Machine Translation Evaluation Sets

Building machine translation (MT) test sets is a relatively expensive task. As MT becomes increasingly desired for more and more language pairs and more and more domains, it becomes necessary to build test sets for each case. In this paper, we investigate using Amazon's Mechanical Turk (MTurk) to make MT test sets cheaply. We find that MTurk can be used to make test sets much cheaper than professionally-produced test sets. More importantly, in experiments with multiple MT systems, we find that the MTurk-produced test sets yield essentially the same conclusions regarding system performance as the professionally-produced test sets yield.

Computation and Language

Bucking the Trend: Large-Scale Cost-Focused Active Learning for Statistical Machine Translation

We explore how to improve machine translation systems by adding more translation data in situations where we already have substantial resources. The main challenge is how to buck the trend of diminishing returns that is commonly encountered. We present an active learning-style data solicitation algorithm to meet this challenge. We test it, gathering annotations via Amazon Mechanical Turk, and find that we get an order of magnitude increase in performance rates of improvement.

Computation and Language

Clustering Words by Projection Entropy

We apply entropy agglomeration (EA), a recently introduced algorithm, to cluster the words of a literary text. EA is a greedy agglomerative procedure that minimizes projection entropy (PE), a function that can quantify the segmentedness of an element set. To apply it, the text is reduced to a feature allocation, a combinatorial object to represent the word occurences in the text's paragraphs. The experiment results demonstrate that EA, despite its reduction and simplicity, is useful in capturing significant relationships among the words in the text. This procedure was implemented in Python and published as a free software: REBUS.

Computation and Language

Analysis of Named Entity Recognition and Linking for Tweets

Applying natural language processing for mining and intelligent information access to tweets (a form of microblog) is a challenging, emerging research area. Unlike carefully authored news text and other longer content, tweets pose a number of new challenges, due to their short, noisy, context-dependent, and dynamic nature. Information extraction from tweets is typically performed in a pipeline, comprising consecutive stages of language identification, tokenisation, part-of-speech tagging, named entity recognition and entity disambiguation (e.g. with respect to DBpedia). In this work, we describe a new Twitter entity disambiguation dataset, and conduct an empirical analysis of named entity recognition and disambiguation, investigating how robust a number of state-of-the-art systems are on such noisy texts, what the main sources of error are, and which problems should be further investigated to improve the state of the art.

Computation and Language

Modified Mel Filter Bank to Compute MFCC of Subsampled Speech

Mel Frequency Cepstral Coefficients (MFCCs) are the most popularly used speech features in most speech and speaker recognition applications. In this work, we propose a modified Mel filter bank to extract MFCCs from subsampled speech. We also propose a stronger metric which effectively captures the correlation between MFCCs of original speech and MFCC of resampled speech. It is found that the proposed method of filter bank construction performs distinguishably well and gives recognition performance on resampled speech close to recognition accuracies on original speech.

Computation and Language

Correcting Errors in Digital Lexicographic Resources Using a Dictionary Manipulation Language

We describe a paradigm for combining manual and automatic error correction of noisy structured lexicographic data. Modifications to the structure and underlying text of the lexicographic data are expressed in a simple, interpreted programming language. Dictionary Manipulation Language (DML) commands identify nodes by unique identifiers, and manipulations are performed using simple commands such as create, move, set text, etc. Corrected lexicons are produced by applying sequences of DML commands to the source version of the lexicon. DML commands can be written manually to repair one-off errors or generated automatically to correct recurring problems. We discuss advantages of the paradigm for the task of editing digital bilingual dictionaries.

Computation and Language

Detecting Structural Irregularity in Electronic Dictionaries Using Language Modeling

Dictionaries are often developed using tools that save to Extensible Markup Language (XML)-based standards. These standards often allow high-level repeating elements to represent lexical entries, and utilize descendants of these repeating elements to represent the structure within each lexical entry, in the form of an XML tree. In many cases, dictionaries are published that have errors and inconsistencies that are expensive to find manually. This paper discusses a method for dictionary writers to quickly audit structural regularity across entries in a dictionary by using statistical language modeling. The approach learns the patterns of XML nodes that could occur within an XML tree, and then calculates the probability of each XML tree in the dictionary against these patterns to look for entries that diverge from the norm.

Computation and Language

Addressing the Rare Word Problem in Neural Machine Translation

Neural Machine Translation (NMT) is a new approach to machine translation that has shown promising results that are comparable to traditional approaches. A significant weakness in conventional NMT systems is their inability to correctly translate very rare words: end-to-end NMTs tend to have relatively small vocabularies with a single unk symbol that represents every possible out-of-vocabulary (OOV) word. In this paper, we propose and implement an effective technique to address this problem. We train an NMT system on data that is augmented by the output of a word alignment algorithm, allowing the NMT system to emit, for each OOV word in the target sentence, the position of its corresponding word in the source sentence. This information is later utilized in a post-processing step that translates every OOV word using a dictionary. Our experiments on the WMT14 English to French translation task show that this method provides a substantial improvement of up to 2.8 BLEU points over an equivalent NMT system that does not use this technique. With 37.5 BLEU points, our NMT system is the first to surpass the best result achieved on a WMT14 contest task.

Computation and Language

Training for Fast Sequential Prediction Using Dynamic Feature Selection

We present paired learning and inference algorithms for significantly reducing computation and increasing speed of the vector dot products in the classifiers that are at the heart of many NLP components. This is accomplished by partitioning the features into a sequence of templates which are ordered such that high confidence can often be reached using only a small fraction of all features. Parameter estimation is arranged to maximize accuracy and early confidence in this sequence. We present experiments in left-to-right part-of-speech tagging on WSJ, demonstrating that we can preserve accuracy above 97% with over a five-fold reduction in run-time.

Computation and Language

A random forest system combination approach for error detection in digital dictionaries

When digitizing a print bilingual dictionary, whether via optical character recognition or manual entry, it is inevitable that errors are introduced into the electronic version that is created. We investigate automating the process of detecting errors in an XML representation of a digitized print dictionary using a hybrid approach that combines rule-based, feature-based, and language model-based methods. We investigate combining methods and show that using random forests is a promising approach. We find that in isolation, unsupervised methods rival the performance of supervised methods. Random forests typically require training data so we investigate how we can apply random forests to combine individual base methods that are themselves unsupervised without requiring large amounts of training data. Experiments reveal empirically that a relatively small amount of data is sufficient and can potentially be further reduced through specific selection criteria.

Computation and Language

Semi-Automatic Construction of a Domain Ontology for Wind Energy Using Wikipedia Articles

Domain ontologies are important information sources for knowledge-based systems. Yet, building domain ontologies from scratch is known to be a very labor-intensive process. In this study, we present our semi-automatic approach to building an ontology for the domain of wind energy which is an important type of renewable energy with a growing share in electricity generation all over the world. Related Wikipedia articles are first processed in an automated manner to determine the basic concepts of the domain together with their properties and next the concepts, properties, and relationships are organized to arrive at the ultimate ontology. We also provide pointers to other engineering ontologies which could be utilized together with the proposed wind energy ontology in addition to its prospective application areas. The current study is significant as, to the best of our knowledge, it proposes the first considerably wide-coverage ontology for the wind energy domain and the ontology is built through a semi-automatic process which makes use of the related Web resources, thereby reducing the overall cost of the ontology building process.

Computation and Language

Experiments to Improve Named Entity Recognition on Turkish Tweets

Social media texts are significant information sources for several application areas including trend analysis, event monitoring, and opinion mining. Unfortunately, existing solutions for tasks such as named entity recognition that perform well on formal texts usually perform poorly when applied to social media texts. In this paper, we report on experiments that have the purpose of improving named entity recognition on Turkish tweets, using two different annotated data sets. In these experiments, starting with a baseline named entity recognition system, we adapt its recognition rules and resources to better fit Twitter language by relaxing its capitalization constraint and by diacritics-based expansion of its lexical resources, and we employ a simplistic normalization scheme on tweets to observe the effects of these on the overall named entity recognition performance on Turkish tweets. The evaluation results of the system with these different settings are provided with discussions of these results.

Computation and Language

Supervised learning model for parsing Arabic language

Parsing the Arabic language is a difficult task given the specificities of this language and given the scarcity of digital resources (grammars and annotated corpora). In this paper, we suggest a method for Arabic parsing based on supervised machine learning. We used the SVMs algorithm to select the syntactic labels of the sentence. Furthermore, we evaluated our parser following the cross validation method by using the Penn Arabic Treebank. The obtained results are very encouraging.

Computation and Language

Rapid Adaptation of POS Tagging for Domain Specific Uses

Part-of-speech (POS) tagging is a fundamental component for performing natural language tasks such as parsing, information extraction, and question answering. When POS taggers are trained in one domain and applied in significantly different domains, their performance can degrade dramatically. We present a methodology for rapid adaptation of POS taggers to new domains. Our technique is unsupervised in that a manually annotated corpus for the new domain is not necessary. We use suffix information gathered from large amounts of raw text as well as orthographic information to increase the lexical coverage. We present an experiment in the Biological domain where our POS tagger achieves results comparable to POS taggers specifically trained to this domain.

Computation and Language

The Latent Structure of Dictionaries

How many words (and which ones) are sufficient to define all other words? When dictionaries are analyzed as directed graphs with links from defining words to defined words, they reveal a latent structure. Recursively removing all words that are reachable by definition but that do not define any further words reduces the dictionary to a Kernel of about 10%. This is still not the smallest number of words that can define all the rest. About 75% of the Kernel turns out to be its Core, a Strongly Connected Subset of words with a definitional path to and from any pair of its words and no word's definition depending on a word outside the set. But the Core cannot define all the rest of the dictionary. The 25% of the Kernel surrounding the Core consists of small strongly connected subsets of words: the Satellites. The size of the smallest set of words that can define all the rest (the graph's Minimum Feedback Vertex Set or MinSet) is about 1% of the dictionary, 15% of the Kernel, and half-Core, half-Satellite. But every dictionary has a huge number of MinSets. The Core words are learned earlier, more frequent, and less concrete than the Satellites, which in turn are learned earlier and more frequent but more concrete than the rest of the Dictionary. In principle, only one MinSet's words would need to be grounded through the sensorimotor capacity to recognize and categorize their referents. In a dual-code sensorimotor-symbolic model of the mental lexicon, the symbolic code could do all the rest via re-combinatory definition.

Computation and Language

On Detecting Noun-Adjective Agreement Errors in Bulgarian Language Using GATE

In this article, we describe an approach for automatic detection of noun-adjective agreement errors in Bulgarian texts by explaining the necessary steps required to develop a simple Java-based language processing application. For this purpose, we use the GATE language processing framework, which is capable of analyzing texts in Bulgarian language and can be embedded in software applications, accessed through a set of Java APIs. In our example application we also demonstrate how to use the functionality of GATE to perform regular expressions over annotations for detecting agreement errors in simple noun phrases formed by two words - attributive adjective and a noun, where the attributive adjective precedes the noun. The provided code samples can also be used as a starting point for implementing natural language processing functionalities in software applications related to language processing tasks like detection, annotation and retrieval of word groups meeting a specific set of criteria.

Computation and Language

Detecting Suicidal Ideation in Chinese Microblogs with Psychological Lexicons

Suicide is among the leading causes of death in China. However, technical approaches toward preventing suicide are challenging and remaining under development. Recently, several actual suicidal cases were preceded by users who posted microblogs with suicidal ideation to Sina Weibo, a Chinese social media network akin to Twitter. It would therefore be desirable to detect suicidal ideations from microblogs in real-time, and immediately alert appropriate support groups, which may lead to successful prevention. In this paper, we propose a real-time suicidal ideation detection system deployed over Weibo, using machine learning and known psychological techniques. Currently, we have identified 53 known suicidal cases who posted suicide notes on Weibo prior to their deaths.We explore linguistic features of these known cases using a psychological lexicon dictionary, and train an effective suicidal Weibo post detection model. 6714 tagged posts and several classifiers are used to verify the model. By combining both machine learning and psychological knowledge, SVM classifier has the best performance of different classifiers, yielding an F-measure of 68:3%, a Precision of 78:9%, and a Recall of 60:3%.

Computation and Language

Tied Probabilistic Linear Discriminant Analysis for Speech Recognition

Acoustic models using probabilistic linear discriminant analysis (PLDA) capture the correlations within feature vectors using subspaces which do not vastly expand the model. This allows high dimensional and correlated feature spaces to be used, without requiring the estimation of multiple high dimension covariance matrices. In this letter we extend the recently presented PLDA mixture model for speech recognition through a tied PLDA approach, which is better able to control the model size to avoid overfitting. We carried out experiments using the Switchboard corpus, with both mel frequency cepstral coefficient features and bottleneck feature derived from a deep neural network. Reductions in word error rate were obtained by using tied PLDA, compared with the PLDA mixture model, subspace Gaussian mixture models, and deep neural networks.

Computation and Language

Modeling Word Relatedness in Latent Dirichlet Allocation

Standard LDA model suffers the problem that the topic assignment of each word is independent and word correlation hence is neglected. To address this problem, in this paper, we propose a model called Word Related Latent Dirichlet Allocation (WR-LDA) by incorporating word correlation into LDA topic models. This leads to new capabilities that standard LDA model does not have such as estimating infrequently occurring words or multi-language topic modeling. Experimental results demonstrate the effectiveness of our model compared with standard LDA.

Computation and Language