Datasets:

WINGNUS
/

ACL-OCL

	{
	"paper_id": "2022",
	"header": {
	"generated_with": "S2ORC 1.0.0",
	"date_generated": "2023-01-19T06:07:06.503690Z"
	},
	"title": "Emotion Analysis of Writers and Readers of Japanese Tweets on Vaccinations",
	"authors": [
	{
	"first": "Patrick",
	"middle": [],
	"last": "Ramos",
	"suffix": "",
	"affiliation": {
	"laboratory": "",
	"institution": "Ateneo de Manila University",
	"location": {
	"country": "Philippines"
	}
	},
	"email": "[email protected]"
	},
	{
	"first": "Kiki",
	"middle": [],
	"last": "Ferawati",
	"suffix": "",
	"affiliation": {},
	"email": "[email protected]"
	},
	{
	"first": "Kongmeng",
	"middle": [],
	"last": "Liew",
	"suffix": "",
	"affiliation": {},
	"email": "[email protected]"
	},
	{
	"first": "Eiji",
	"middle": [],
	"last": "Aramaki",
	"suffix": "",
	"affiliation": {},
	"email": "[email protected]"
	},
	{
	"first": "Shoko",
	"middle": [],
	"last": "Wakamiya",
	"suffix": "",
	"affiliation": {},
	"email": "[email protected]"
	}
	],
	"year": "",
	"venue": null,
	"identifiers": {},
	"abstract": "Public opinion in social media is increasingly becoming a critical factor in pandemic control. Understanding the emotions of a population towards vaccinations and COVID-19 may be valuable in convincing members to become vaccinated. We investigated the emotions of Japanese Twitter users towards Tweets related to COVID-19 vaccination. Using the WRIME dataset, which provides emotion ratings for Japanese Tweets sourced from writers (Tweet posters) and readers, we fine-tuned a BERT model to predict levels of emotional intensity. This model achieved a training accuracy of M SE = 0.356. A separate dataset of 20,254 Japanese Tweets containing COVID-19 vaccine-related keywords was also collected, on which the fine-tuned BERT was used to perform emotion analysis. Afterwards, a correlation analysis between the extracted emotions and a set of vaccination measures in Japan was conducted. The results revealed that surprise and fear were the most intense emotions predicted by the model for writers and readers, respectively, on the vaccine-related Tweet dataset. The correlation analysis also showed that vaccinations were weakly positively correlated with predicted levels of writer joy, writer/reader anticipation, and writer/reader trust. Code will be made available at https: //github.com/PatrickJohnRamos/ BERT-Japan-vaccination.",
	"pdf_parse": {
	"paper_id": "2022",
	"_pdf_hash": "",
	"abstract": [
	{
	"text": "Public opinion in social media is increasingly becoming a critical factor in pandemic control. Understanding the emotions of a population towards vaccinations and COVID-19 may be valuable in convincing members to become vaccinated. We investigated the emotions of Japanese Twitter users towards Tweets related to COVID-19 vaccination. Using the WRIME dataset, which provides emotion ratings for Japanese Tweets sourced from writers (Tweet posters) and readers, we fine-tuned a BERT model to predict levels of emotional intensity. This model achieved a training accuracy of M SE = 0.356. A separate dataset of 20,254 Japanese Tweets containing COVID-19 vaccine-related keywords was also collected, on which the fine-tuned BERT was used to perform emotion analysis. Afterwards, a correlation analysis between the extracted emotions and a set of vaccination measures in Japan was conducted. The results revealed that surprise and fear were the most intense emotions predicted by the model for writers and readers, respectively, on the vaccine-related Tweet dataset. The correlation analysis also showed that vaccinations were weakly positively correlated with predicted levels of writer joy, writer/reader anticipation, and writer/reader trust. Code will be made available at https: //github.com/PatrickJohnRamos/ BERT-Japan-vaccination.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Abstract",
	"sec_num": null
	}
	],
	"body_text": [
	{
	"text": "Vaccination against COVID-19 has been demonstrated to reduce the spread of the virus (Jones et al., 2021; Hall et al., 2021; Voysey et al., 2021) . However, vaccine hesitancy can prevent vaccine uptake, increasing risk of infection. Searching for and understanding causes of vaccine hesitancy can lead to more effective methodologies in convincing community members to become vaccinated. One possible area of understanding vaccine hesitancy is the emotions felt towards vaccines and the COVID-19 pandemic. For example, fear felt towards vaccination might discourage one from receiving the vaccine. Meanwhile, fear felt towards contracting COVID-19 could encourage one to become vaccinated against it. Leveraging emotions has also been proposed in communication to reduce COVID-19 vaccine hesitancy (Chou and Budenz, 2020) .",
	"cite_spans": [
	{
	"start": 85,
	"end": 105,
	"text": "(Jones et al., 2021;",
	"ref_id": "BIBREF13"
	},
	{
	"start": 106,
	"end": 124,
	"text": "Hall et al., 2021;",
	"ref_id": "BIBREF11"
	},
	{
	"start": 125,
	"end": 145,
	"text": "Voysey et al., 2021)",
	"ref_id": "BIBREF28"
	},
	{
	"start": 798,
	"end": 821,
	"text": "(Chou and Budenz, 2020)",
	"ref_id": "BIBREF7"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction",
	"sec_num": "1"
	},
	{
	"text": "There have been several attempts at extracting emotions regarding COVID-19 vaccines, particularly from social media. The wealth of information available on social networking services such as Twitter already makes them a popular source for mining sentiment and emotions in other areas such as politics (Bose et al., 2019) and consumerism (Rathan et al., 2018) . Social media information extraction has seen continued research during the COVID-19 pandemic, with multiple works specifically seeking to mine sentiments and emotions surrounding the pandemic and vaccination (Boon-Itt and Skunkan, 2020; Sakti et al., 2021; Ayg\u00fcn et al., 2021; Niu et al., 2022) .",
	"cite_spans": [
	{
	"start": 301,
	"end": 320,
	"text": "(Bose et al., 2019)",
	"ref_id": "BIBREF6"
	},
	{
	"start": 337,
	"end": 358,
	"text": "(Rathan et al., 2018)",
	"ref_id": "BIBREF23"
	},
	{
	"start": 569,
	"end": 597,
	"text": "(Boon-Itt and Skunkan, 2020;",
	"ref_id": "BIBREF5"
	},
	{
	"start": 598,
	"end": 617,
	"text": "Sakti et al., 2021;",
	"ref_id": "BIBREF25"
	},
	{
	"start": 618,
	"end": 637,
	"text": "Ayg\u00fcn et al., 2021;",
	"ref_id": "BIBREF1"
	},
	{
	"start": 638,
	"end": 655,
	"text": "Niu et al., 2022)",
	"ref_id": "BIBREF20"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction",
	"sec_num": "1"
	},
	{
	"text": "Existing emotion analysis studies on COVID-19related Japanese Twitter corpora only focus on the emotions of writers, or those who post Tweets (Lee et al., 2020; Bashar, 2021) . However, the emotions of readers, or those who read Tweets, are not necessarily the same as those of writers. For instance, if a writer expresses disgust towards vaccination, a reader might express anger out of disagreement in response. These reader emotions may also contain useful information in understanding vaccine hesitancy. We contribute to this research area by using a BERT to extract emotions of both writers and readers towards Tweets related to COVID-19 vaccines and comparing the predicted emotions to vaccination uptake. We do this by fine-tuning a BERT (Devlin et al., 2019) to predict intensity scores for Plutchik's eight emotions (Plutchik, 1980) for writers and readers from Tweets containing keywords related to COVID-19 vaccination, and performing a correlation analysis between the mined emotions and a set of vaccination measures in Japan. Note that one limitation of our study is that the dataset of COVID-19 vaccine-related Tweets does not guarantee that COVID-19 or vaccination are the topics of the texts, or are necessarily the object of any inferred emotion.",
	"cite_spans": [
	{
	"start": 142,
	"end": 160,
	"text": "(Lee et al., 2020;",
	"ref_id": "BIBREF16"
	},
	{
	"start": 161,
	"end": 174,
	"text": "Bashar, 2021)",
	"ref_id": "BIBREF3"
	},
	{
	"start": 825,
	"end": 841,
	"text": "(Plutchik, 1980)",
	"ref_id": "BIBREF22"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction",
	"sec_num": "1"
	},
	{
	"text": "We find that the emotion most prominently predicted by the model for writers on the vaccinerelated Tweet dataset is surprise, with fear being the most intensely predicted emotion for readers. Additionally, writer joy, writer/reader anticipation, and writer/reader trust are weakly positively correlated with vaccinations.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction",
	"sec_num": "1"
	},
	{
	"text": "Prior to the adoption of deep learning, emotion analysis of Tweets was performed with a combination of feature engineering, lexicon-based approach, and traditional off-the-shelf classifiers. Balabantaray et al. (2012) and Wang et al. (2012) engineered overlapping sets of features from Tweets, with shared features including n-grams, POS, adjectives, and lexicon-based sentiment polarity scores. Balabantaray et al. (2012) fed the features to an SVM while Wang et al. (2012) used linear and Naive Bayes classifiers. EmpaTweet (2012) used a similar set of features and also an SVM, but exchanged sentiment polarity scores for synonym rings, hypernyms, and LDA topic scores.",
	"cite_spans": [
	{
	"start": 191,
	"end": 217,
	"text": "Balabantaray et al. (2012)",
	"ref_id": "BIBREF2"
	},
	{
	"start": 222,
	"end": 240,
	"text": "Wang et al. (2012)",
	"ref_id": "BIBREF29"
	},
	{
	"start": 456,
	"end": 474,
	"text": "Wang et al. (2012)",
	"ref_id": "BIBREF29"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Related Work",
	"sec_num": "2"
	},
	{
	"text": "However, these classical methods have been outperformed by more contemporary and dedicated sequence modelling techniques such as RNNs and LSTMs. Vateekul and Koomsubha (2016) demonstrated the superiority of LSTMs in emotion analysis over SVMs and Naive Bayes on Thai Twitter text, while Colneri\u010d and Dem\u0161ar (2018) showed the effectiveness of character-based RNNs.",
	"cite_spans": [
	{
	"start": 145,
	"end": 174,
	"text": "Vateekul and Koomsubha (2016)",
	"ref_id": "BIBREF27"
	},
	{
	"start": 287,
	"end": 313,
	"text": "Colneri\u010d and Dem\u0161ar (2018)",
	"ref_id": "BIBREF10"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Related Work",
	"sec_num": "2"
	},
	{
	"text": "The introduction of Transformers (Vaswani et al., 2017) as the new state-of-the-art sequence modelling architecture and their increasing ubiquity has also lead to their application in social media emotion analysis for a variety of languages. BERT models can outperform CNNs and BiLSTMs on English Twitter (Harb et al., 2020); Naive Bayes, logistic regression, and SVMs on Romanian Twitter (Ciobotaru and Dinu, 2021) ; and TF-IDF and word2vec on Arabic Twitter (Al-Twairesh, 2021). During the COVID-19 pandemic, a RoBERTa (Liu et al., 2019) was fine-tuned on emotion analysis to classify the emotions of Tweets containing hashtags related to the pandemic (Choudrie et al., 2021) .",
	"cite_spans": [
	{
	"start": 33,
	"end": 55,
	"text": "(Vaswani et al., 2017)",
	"ref_id": "BIBREF26"
	},
	{
	"start": 389,
	"end": 415,
	"text": "(Ciobotaru and Dinu, 2021)",
	"ref_id": "BIBREF9"
	},
	{
	"start": 521,
	"end": 539,
	"text": "(Liu et al., 2019)",
	"ref_id": "BIBREF17"
	},
	{
	"start": 654,
	"end": 677,
	"text": "(Choudrie et al., 2021)",
	"ref_id": null
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Related Work",
	"sec_num": "2"
	},
	{
	"text": "Meanwhile, emotion analyses of Japanese Tweets related to COVID-19 have been conducted using traditional techniques such as lexicon-based methods (Bashar, 2021) and frequency analysis (Lee et al., 2020) .",
	"cite_spans": [
	{
	"start": 146,
	"end": 160,
	"text": "(Bashar, 2021)",
	"ref_id": "BIBREF3"
	},
	{
	"start": 184,
	"end": 202,
	"text": "(Lee et al., 2020)",
	"ref_id": "BIBREF16"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Related Work",
	"sec_num": "2"
	},
	{
	"text": "The work most similar to ours is that of Niu et al. (2022) , who perform sentiment analysis of COVID-19-related Japanese Tweets; investigate the correlation of the mined sentiments with infections, deaths, and vaccinations; and conduct additional analyses comparing multiple vaccine brands. Our study differs from this by extracting emotions rather than sentiments from both writers and readers, and comparing these mined features to vaccinations, vaccinated people, and fully vaccinated people.",
	"cite_spans": [
	{
	"start": 41,
	"end": 58,
	"text": "Niu et al. (2022)",
	"ref_id": "BIBREF20"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Related Work",
	"sec_num": "2"
	},
	{
	"text": "To gauge the emotions of the Japanese public towards COVID-19 vaccines and related topics, a dataset of 20,254 vaccine-related Tweets from December 2021 containing any of the keywords\"\u30ef \u30af\u30c1\u30f3\" (\"vaccine\"), \"\u30e2\u30c7\u30eb\u30ca\" (\"Moderna\"), \"\u30d5\u30a1\u30a4\u30b6\u30fc\" (\"Pfizer\"), or \"\u30aa\u30df\u30af\u30ed\u30f3\" (\"Omicron\") was constructed. \"Moderna\" and \"Pfizer\" were specifically selected as keywords as these are brands of COVID-19 vaccines commonly administered in Japan. The dataset was created by sampling 15 random minutes from each day of December 2021 for each keyword, and scraping all Tweets containing the assigned keyword for each sampled minute. A distribution of the dataset according to keyword is shown in Table 1 . Table 2 compares our constructed dataset against WRIME (Kajiwara et al., 2021) , the fine-tuning dataset we discuss in Section 3.2. Our dataset contains longer texts on average. Postpositional particles, nouns, punctuation marks, and verbs are among the most common parts of speech in both datasets. However, auxiliary verbs are more common in the vaccine-related Tweet dataset while non-punctuation symbols are more common in WRIME.",
	"cite_spans": [
	{
	"start": 731,
	"end": 754,
	"text": "(Kajiwara et al., 2021)",
	"ref_id": "BIBREF14"
	}
	],
	"ref_spans": [
	{
	"start": 666,
	"end": 673,
	"text": "Table 1",
	"ref_id": "TABREF1"
	},
	{
	"start": 676,
	"end": 683,
	"text": "Table 2",
	"ref_id": "TABREF2"
	}
	],
	"eq_spans": [],
	"section": "Method 3.1 Dataset",
	"sec_num": "3"
	},
	{
	"text": "We used BERT to perform emotion analysis by fine-tuning it to extract writer and reader emotional intensity scores from text, with higher scores indicating higher intensities. Writer emotions refer to emotions felt by the writers of a Tweet as they write the post, while reader emotions refer to the emotions felt by the readers of a Tweet as they read it. We followed Plutchik's (1980) framework and fine-tuned BERT to extract intensity scores for eight emotions: joy, sadness, anticipation, surprise, anger, fear, disgust, and trust.",
	"cite_spans": [
	{
	"start": 369,
	"end": 386,
	"text": "Plutchik's (1980)",
	"ref_id": "BIBREF22"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Fine-tuning BERT for Emotion Analysis",
	"sec_num": "3.2"
	},
	{
	"text": "Using BERT for emotion analysis was straightforward and required only a simple modification to the head of the BERT model. First, input texts were tokenized using MeCab (Kudo et al., 2004) and WordPiece (Wu et al., 2016) . The tokenized inputs were then each prepended with [CLS] classification tokens and fed through the BERT model. After the last layer, each [CLS] token was linearly projected into 16 class scores, representing the 8 emotions of writers and readers.",
	"cite_spans": [
	{
	"start": 169,
	"end": 188,
	"text": "(Kudo et al., 2004)",
	"ref_id": "BIBREF15"
	},
	{
	"start": 203,
	"end": 220,
	"text": "(Wu et al., 2016)",
	"ref_id": "BIBREF31"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Fine-tuning BERT for Emotion Analysis",
	"sec_num": "3.2"
	},
	{
	"text": "A PyTorch (Paszke et al., 2019) implementation of BERT BASE (110M parameters) pre-trained on Japanese Wikipedia from HuggingFace (Wolf et al., 2020) 2 was fine-tuned on WRIME (Kajiwara et al., 2021) . WRIME is a dataset for emotional intensity estimation comprised of 43,200 Japanese Tweets annotated with Plutchik's 8 emotions by the posts' writers and 3 \"reader\" annotators hired by the dataset authors to read the posts. 3 Each emotion is annotated across 4 levels (0 to 3) of increasing intensity, with 0 referring to no presence and 3 re-ferring to strong intensity. To create only a single set of reader emotion scores per data point, we averaged the scores across the three readers. BERT was then trained with mean squared error loss to directly predict the emotional intensity scores of writers and readers for given Tweets. A sample data point can be seen in Table 3 .",
	"cite_spans": [
	{
	"start": 10,
	"end": 31,
	"text": "(Paszke et al., 2019)",
	"ref_id": "BIBREF21"
	},
	{
	"start": 175,
	"end": 198,
	"text": "(Kajiwara et al., 2021)",
	"ref_id": "BIBREF14"
	}
	],
	"ref_spans": [
	{
	"start": 868,
	"end": 875,
	"text": "Table 3",
	"ref_id": "TABREF4"
	}
	],
	"eq_spans": [],
	"section": "Fine-tuning BERT for Emotion Analysis",
	"sec_num": "3.2"
	},
	{
	"text": "Our fine-tuning was inspired by common BERT fine-tuning procedures (Devlin et al., 2019). We fine-tuned BERT on WRIME for 3 epochs using the AdamW optimizer (Loshchilov and Hutter, 2017) with a learning rate of 2e-5, \u03b2 1 =0.9, \u03b2 2 =0.999, weight decay of 0.01, linear decay, a warmup ratio of 0.01, and a batch size of 32. Training was conducted with an NVIDIA Tesla K80 and finished in 3 hours. The fine-tuned BERT was dubbed \"emotion analysis BERT.\"",
	"cite_spans": [
	{
	"start": 157,
	"end": 186,
	"text": "(Loshchilov and Hutter, 2017)",
	"ref_id": null
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Fine-tuning BERT for Emotion Analysis",
	"sec_num": "3.2"
	},
	{
	"text": "We evaluated emotion analysis BERT by comparing its mean squared errors to those of two baselines based on Kajiwara et al. (2021) . The first was a bag-of-words and linear regression model (BoW+LinReg). Each text was tokenized using MeCab, vectorized into a bag-of-words using the 2000 most common words in the vocabulary, and then fed into a linear regression model. While Kajiwara et al. (2021) used logistic regression, we used linear regression for a fairer comparison with emotion analysis BERT, which directly predicts intensity scores. The second model we compared to uses fastText (Bojanowski et al., 2017) and an SVM (fastText+SVM). Each word of was embedded using fastText, after which the average of all the embeddings in the sequence were used as input into an SVM that regresses the emotional intensity scores. We used a linear kernel and C = 100.",
	"cite_spans": [
	{
	"start": 107,
	"end": 129,
	"text": "Kajiwara et al. (2021)",
	"ref_id": "BIBREF14"
	},
	{
	"start": 589,
	"end": 614,
	"text": "(Bojanowski et al., 2017)",
	"ref_id": "BIBREF4"
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Fine-tuning BERT for Emotion Analysis",
	"sec_num": "3.2"
	},
	{
	"text": "Emotions from the vaccine-related Tweet dataset were extracted using emotion analysis BERT following the procedure described in Section 3.2. Tweets were tokenized, prepended with a [CLS] token, and processed through emotion analysis BERT, with the [CLS] token being projected into writer and reader emotion scores after the last layer. Note that this is purely inference and no training is done using the vaccine Tweet data. ",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Inferring Emotions from Vaccine-related Tweets",
	"sec_num": "3.3"
	},
	{
	"text": "Tweet \u8eca \u306e \u30bf \u30a4 \u30e4 \u304c \u30d1 \u30f3 \u30af \u3057 \u3066 \u305f \u3002 \u3002 \u3044 \u305f \u305a \u3089 \u306e \u53ef \u80fd \u6027 \u304c \u9ad8 \u3044 \u3093 \u3060 \u3063 \u3066\u3002\u3002(",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Inferring Emotions from Vaccine-related Tweets",
	"sec_num": "3.3"
	},
	{
	"text": "Results for emotional intensity prediction on WRIME are shown in Table 4 . The only emotion for which we do not achieve the best mean squared error is writer trust, where we still remain competitive a difference of 0.001 MSE from the best performing model. For all other emotions for both writers and readers, emotion analysis BERT outperforms both of the other models. We achieve 0.111 MSE and 0.106 MSE improvements on writer and reader emotion prediction respectively. For overall emotional intensity prediction, we outperform the next best model by 0.090 MSE. Table 5 presents inferred emotions for sample test entries from WRIME. We qualitatively examined these results, which showed that emotion analysis BERT is capable of detecting emotion even when emotions are not explicitly stated, like they would be in a sentence such as \"I am joyful.\" The model was able to infer anger and disgust from texts including phrases like \"I waited for an hour but no one came!!\", and joy and anticipation from \"Haven't had back-to-back holidays in a while. Tomorrow is a weekday break!!\" Figure 1 presents the distributions of emotions for both writers and readers predicted by emotion anal- ysis BERT. The highest average predicted score was surprise for writers and fear for readers. With the exception of fear, inferred writer scores were more intense, especially in sadness, anger, and trust, where median inferred writer scores were greater than the third quartile of inferred reader scores. Why the emotion intensities predicted for writers differed from those assumed for readers could be of interest to future studies.",
	"cite_spans": [],
	"ref_spans": [
	{
	"start": 65,
	"end": 72,
	"text": "Table 4",
	"ref_id": "TABREF5"
	},
	{
	"start": 564,
	"end": 571,
	"text": "Table 5",
	"ref_id": "TABREF7"
	},
	{
	"start": 1080,
	"end": 1088,
	"text": "Figure 1",
	"ref_id": "FIGREF0"
	}
	],
	"eq_spans": [],
	"section": "Fine-tuning results",
	"sec_num": "4.1"
	},
	{
	"text": "The distributions of predicted writer and reader emotions are compared in a Q-Q plot in Figure 2 . Anticipation and fear followed similar distributions for both writer and reader inferences, while other emotions showed lower values for readers, especially for trust. The percentage distributions of emotions for writers and readers inferred by the model for all keywords and per keyword are shown in Figure 3 . While predicted writer emotion scores tended to be higher, this did not necessarily mean the same for the proportion each emotion constituted of the sum of all emotion scores. Although writers had higher predicted average anticipation, surprise, and disgust, these emotions comprised a larger percentage of inferred reader emotions than they did writer emotions. Fear, which was predicted at higher levels for readers, also constituted a larger proportion of total inferred reader emotion.",
	"cite_spans": [],
	"ref_spans": [
	{
	"start": 88,
	"end": 97,
	"text": "Figure 2",
	"ref_id": "FIGREF1"
	},
	{
	"start": 401,
	"end": 409,
	"text": "Figure 3",
	"ref_id": null
	}
	],
	"eq_spans": [],
	"section": "Emotion inference results",
	"sec_num": "4.2"
	},
	{
	"text": "Predicted emotions of Tweets referencing Moderna were compared to those of Tweets referencing Pfizer to identify any differences in emotions towards the two brands of vaccines. The results of a Kolmogorov-Smirnov test for the intensities of emotions towards the two brands are presented in Table 6 . Only differences in predictions for writer joy (Pfizer higher), writer sadness (Moderna higher), and reader sadness (Moderna higher) could be considered attributable to differences in Figure 3 : Distributions of predicted emotions for writers and readers for all keywords and for each keyword. Emotion proportions were calculated by dividing the total score of each emotion by the sum of all emotion scores for the emotion's assigned subject (writer or reader).",
	"cite_spans": [],
	"ref_spans": [
	{
	"start": 290,
	"end": 297,
	"text": "Table 6",
	"ref_id": "TABREF9"
	},
	{
	"start": 484,
	"end": 492,
	"text": "Figure 3",
	"ref_id": null
	}
	],
	"eq_spans": [],
	"section": "Emotion inference results",
	"sec_num": "4.2"
	},
	{
	"text": "underlying distributions.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Emotion inference results",
	"sec_num": "4.2"
	},
	{
	"text": "Sample emotion inferences can be seen in Table 7 . Like with the qualitative results from WRIME, emotion analysis BERT was capable of detecting emotion even in a lack of explicitly mentioned emotions. The model could identify anger and disgust from Tweets containing phrases such as \"I can't call it anything other than foolish,\" and trust from \"Today, I finally got vaccinated. My arm hurts a little bit, but other than that there are no problems.\" ",
	"cite_spans": [],
	"ref_spans": [
	{
	"start": 41,
	"end": 49,
	"text": "Table 7",
	"ref_id": "TABREF11"
	}
	],
	"eq_spans": [],
	"section": "Emotion inference results",
	"sec_num": "4.2"
	},
	{
	"text": "Vaccination data was taken from Our World in Data (Mathieu et al., 2021) 4 . The vaccination data was comprised of a set of periodic vaccination measures for several countries. We filtered the dataset to only include entries from Japan in December 2021, matching the collection period of our vaccine-related Tweet dataset. For each date in the dataset, we focused on vaccinations, or the number of vaccinations administered on that date; new people vaccinated, or the number of people who received their first dose on said date; and new people fully vaccinated, or the number of people who received their second dose on that date. One thing to note is that vaccinations have plateaued starting November 2021, resulting from a lower number of vaccinations and a slower uptake of boosters (Mathieu et al., 2021) . Figure 4 compares the average predicted score for each emotion to the number of new vaccinations, people vaccinated, and people fully vaccinated across December 2021. No easily discernible trend common to both emotion scores and vaccination metrics was found.",
	"cite_spans": [
	{
	"start": 50,
	"end": 72,
	"text": "(Mathieu et al., 2021)",
	"ref_id": "BIBREF19"
	},
	{
	"start": 787,
	"end": 809,
	"text": "(Mathieu et al., 2021)",
	"ref_id": "BIBREF19"
	}
	],
	"ref_spans": [
	{
	"start": 812,
	"end": 820,
	"text": "Figure 4",
	"ref_id": "FIGREF2"
	}
	],
	"eq_spans": [],
	"section": "Comparison to vaccinations in Japan",
	"sec_num": "4.3"
	},
	{
	"text": "For each writer and reader emotion, we performed a correlation analysis with each vaccination 4 Vaccination data is available at https: //github.com/owid/covid-19-data/tree/ master/public/data/vaccinations. measure by taking the Pearson correlation coefficient between the sums of the particular emotion's predicted intensities and their corresponding vaccination metric for each date recorded in the vaccination dataset. The results are presented in Figure 5 . The only results with satisfactory p-values were under vaccinations, which were positive correlated with writer joy (r = 0.36, p = 0.047), writer anticipation (r = 0.40, p = 0.027), writer trust (r = 0.40, p = 0.025), reader anticipation (r = 0.44, p = 0.013), and reader trust (r = 0.39, p = 0.031). We did not observe any significant correlations between the predicted emotion scores and people vaccinated or people fully vaccinated, as all p-values for these results were greater than or equal to 0.157, which is above the alpha of 0.05.",
	"cite_spans": [
	{
	"start": 94,
	"end": 95,
	"text": "4",
	"ref_id": null
	}
	],
	"ref_spans": [
	{
	"start": 451,
	"end": 460,
	"text": "Figure 5",
	"ref_id": "FIGREF3"
	}
	],
	"eq_spans": [],
	"section": "Comparison to vaccinations in Japan",
	"sec_num": "4.3"
	},
	{
	"text": "Research into understanding the emotions of a population towards vaccines could hold both positive and negative societal impacts. Any relationship discovered between emotions and vaccinations could be leveraged in campaigns aimed at convincing citizens to receive vaccinations, reducing the spread of COVID-19. On the other hand, knowledge of what emotions could affect vaccine acceptance can be used in efforts to increase vaccination hesitancy and slow down vaccination rates, which could prolong risks of infection.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Broader Impact",
	"sec_num": "5"
	},
	{
	"text": "We fine-tuned a BERT on the task of emotion analysis, and used the emotion analysis BERT to infer emotion intensities of writers and readers from a corpus of Tweets containing keywords related to COVID-19 and COVID-19 vaccination. Our results revealed that surprise and fear were respectively the most intensely predicted emotions for writers and readers. Furthermore, vaccinations were weakly positively correlated with writer joy, writer anticipation, writer trust, reader anticipation, and reader trust.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Conclusion",
	"sec_num": "6"
	},
	{
	"text": "Future works can extend this study by designing the emotion analysis to be aspect-based with respect to the keywords, as it is possible that the keywords are not the objects of the inferred emotions. Another possible area for further research could be correlation analyses with other COVID-19 metrics, such as infections and reproduction rate. Tweet \u30d0\u30ab\u3068\u3057\u304b\u8a00\u3044\u3088\u3046\u304c\u306a\u3044\u3002\u30a2\u30e1\u30ea\u30ab\u5e30\u308a\u306e\u5973\u304c\u81ea\u5b85\u5f85\u6a5f\u671f\u9593\u306b\u7537\u3068 \u4f1a\u3063\u3066\u3001\u305d\u306e\u7537\u304c\u30b5\u30c3\u30ab\u30fc\u5834\u3067\u6492\u304d\u6563\u3089\u3059\u3002\u3042\u3042\u3042\u3001\u30d0\u30ab\u3060\u306a\u3042\u3001 \u30aa\u30df\u30af\u30ed\u30f3\u682a\u63a5\u89e6\u8005 \u5929\u7687\u676f\u89b3\u6226 2021\u5e7412\u670816\u65e5 (I can't call it anything other than foolish. A woman who returned from the United States met with a man while she was still in quarantine, then that man went to a soccer game and infected others. Ah, it really is foolish. ",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Conclusion",
	"sec_num": "6"
	},
	{
	"text": "We use the cl-tohoku/bert-base-japanse-v2 Japanese BERT checkpoint available at https://huggingface.co/cl-tohoku/ bert-base-japanese-v2.3 As of February 2022, there are two versions of this dataset. We use Version 1. WRIME is available at https: //github.com/ids-cv/wrime.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "",
	"sec_num": null
	}
	],
	"back_matter": [
	{
	"text": "This work was supported by JST SICORP Grant Number JPMJSC2107, Japan.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Acknowledgements",
	"sec_num": null
	}
	],
	"bib_entries": {
	"BIBREF0": {
	"ref_id": "b0",
	"title": "The evolution of language models applied to emotion analysis of arabic tweets",
	"authors": [
	{
	"first": "Nora",
	"middle": [],
	"last": "Al-Twairesh",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "Information",
	"volume": "12",
	"issue": "2",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.3390/info12020084"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Nora Al-Twairesh. 2021. The evolution of language models applied to emotion analysis of arabic tweets. Information, 12(2):84.",
	"links": null
	},
	"BIBREF1": {
	"ref_id": "b1",
	"title": "Aspect based twitter sentiment analysis on vaccination and vaccine types in covid-19 pandemic with deep learning",
	"authors": [
	{
	"first": "Buket",
	"middle": [],
	"last": "Irfan Ayg\u00fcn",
	"suffix": ""
	},
	{
	"first": "Mehmet",
	"middle": [],
	"last": "Kaya",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Kaya",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "IEEE Journal of Biomedical and Health Informatics",
	"volume": "",
	"issue": "",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.1109/JBHI.2021.3133103"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Irfan Ayg\u00fcn, Buket Kaya, and Mehmet Kaya. 2021. As- pect based twitter sentiment analysis on vaccination and vaccine types in covid-19 pandemic with deep learning. IEEE Journal of Biomedical and Health Informatics.",
	"links": null
	},
	"BIBREF2": {
	"ref_id": "b2",
	"title": "Multi-class twitter emotion classification: A new approach",
	"authors": [
	{
	"first": "C",
	"middle": [],
	"last": "Rakesh",
	"suffix": ""
	},
	{
	"first": "Mudasir",
	"middle": [],
	"last": "Balabantaray",
	"suffix": ""
	},
	{
	"first": "Nibha",
	"middle": [],
	"last": "Mohammad",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Sharma",
	"suffix": ""
	}
	],
	"year": 2012,
	"venue": "International Journal of Applied Information Systems",
	"volume": "4",
	"issue": "1",
	"pages": "48--53",
	"other_ids": {
	"DOI": [
	"10.5120/ijais12-450651"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Rakesh C Balabantaray, Mudasir Mohammad, and Nibha Sharma. 2012. Multi-class twitter emotion classification: A new approach. International Jour- nal of Applied Information Systems, 4(1):48-53.",
	"links": null
	},
	"BIBREF3": {
	"ref_id": "b3",
	"title": "Exploration of public emotion dynamics in japan from twitter data during covid-19",
	"authors": [
	{
	"first": "Md Khayrul",
	"middle": [],
	"last": "Bashar",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "2021 6th International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)",
	"volume": "6",
	"issue": "",
	"pages": "310--315",
	"other_ids": {
	"DOI": [
	"10.1109/ICIIBMS52876.2021.9651600"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Md Khayrul Bashar. 2021. Exploration of public emo- tion dynamics in japan from twitter data during covid- 19. In 2021 6th International Conference on Intelli- gent Informatics and Biomedical Sciences (ICIIBMS), volume 6, pages 310-315. IEEE.",
	"links": null
	},
	"BIBREF4": {
	"ref_id": "b4",
	"title": "Enriching word vectors with subword information",
	"authors": [
	{
	"first": "Piotr",
	"middle": [],
	"last": "Bojanowski",
	"suffix": ""
	},
	{
	"first": "Edouard",
	"middle": [],
	"last": "Grave",
	"suffix": ""
	},
	{
	"first": "Armand",
	"middle": [],
	"last": "Joulin",
	"suffix": ""
	},
	{
	"first": "Tomas",
	"middle": [],
	"last": "Mikolov",
	"suffix": ""
	}
	],
	"year": 2017,
	"venue": "Transactions of the Association for Computational Linguistics",
	"volume": "5",
	"issue": "",
	"pages": "135--146",
	"other_ids": {
	"DOI": [
	"10.1162/tacl_a_00051"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2017. Enriching word vectors with subword information. Transactions of the Associa- tion for Computational Linguistics, 5:135-146.",
	"links": null
	},
	"BIBREF5": {
	"ref_id": "b5",
	"title": "Public perception of the covid-19 pandemic on twitter: Sentiment analysis and topic modeling study",
	"authors": [
	{
	"first": "Sakun",
	"middle": [],
	"last": "Boon",
	"suffix": ""
	},
	{
	"first": "-Itt",
	"middle": [],
	"last": "",
	"suffix": ""
	},
	{
	"first": "Yukolpat",
	"middle": [],
	"last": "Skunkan",
	"suffix": ""
	}
	],
	"year": 2020,
	"venue": "JMIR Public Health and Surveillance",
	"volume": "6",
	"issue": "4",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.2196/21978"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Sakun Boon-Itt and Yukolpat Skunkan. 2020. Public perception of the covid-19 pandemic on twitter: Sen- timent analysis and topic modeling study. JMIR Public Health and Surveillance, 6(4):e21978.",
	"links": null
	},
	"BIBREF6": {
	"ref_id": "b6",
	"title": "Analyzing political sentiment using twitter data",
	"authors": [
	{
	"first": "Rajesh",
	"middle": [],
	"last": "Bose",
	"suffix": ""
	},
	{
	"first": "Raktim",
	"middle": [],
	"last": "Kumar Dey",
	"suffix": ""
	},
	{
	"first": "Sandip",
	"middle": [],
	"last": "Roy",
	"suffix": ""
	},
	{
	"first": "Debabrata",
	"middle": [],
	"last": "Sarddar",
	"suffix": ""
	}
	],
	"year": 2019,
	"venue": "Information and communication technology for intelligent systems",
	"volume": "",
	"issue": "",
	"pages": "427--436",
	"other_ids": {
	"DOI": [
	"10.1007/978-981-13-1747-7_41"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Rajesh Bose, Raktim Kumar Dey, Sandip Roy, and De- babrata Sarddar. 2019. Analyzing political sentiment using twitter data. In Information and communica- tion technology for intelligent systems, pages 427- 436. Springer.",
	"links": null
	},
	"BIBREF7": {
	"ref_id": "b7",
	"title": "Considering emotion in covid-19 vaccine communication: addressing vaccine hesitancy and fostering vaccine confidence",
	"authors": [
	{
	"first": "Wen-Ying Sylvia",
	"middle": [],
	"last": "Chou",
	"suffix": ""
	},
	{
	"first": "Alexandra",
	"middle": [],
	"last": "Budenz",
	"suffix": ""
	}
	],
	"year": 2020,
	"venue": "Health communication",
	"volume": "35",
	"issue": "14",
	"pages": "1718--1722",
	"other_ids": {
	"DOI": [
	"10.1080/10410236.2020.1838096"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Wen-Ying Sylvia Chou and Alexandra Budenz. 2020. Considering emotion in covid-19 vaccine commu- nication: addressing vaccine hesitancy and foster- ing vaccine confidence. Health communication, 35(14):1718-1722.",
	"links": null
	},
	"BIBREF8": {
	"ref_id": "b8",
	"title": "Nikhil Matta, and Ilias Pappas. 2021. Applying and understanding an advanced, novel deep learning approach: A covid 19, text based, emotions analysis study",
	"authors": [
	{
	"first": "Jyoti",
	"middle": [],
	"last": "Choudrie",
	"suffix": ""
	},
	{
	"first": "Shruti",
	"middle": [],
	"last": "Patil",
	"suffix": ""
	},
	{
	"first": "Ketan",
	"middle": [],
	"last": "Kotecha",
	"suffix": ""
	}
	],
	"year": null,
	"venue": "Systems Frontiers",
	"volume": "23",
	"issue": "6",
	"pages": "1431--1465",
	"other_ids": {
	"DOI": [
	"10.1007/s10796-021-10152-6"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Jyoti Choudrie, Shruti Patil, Ketan Kotecha, Nikhil Matta, and Ilias Pappas. 2021. Applying and under- standing an advanced, novel deep learning approach: A covid 19, text based, emotions analysis study. In- formation Systems Frontiers, 23(6):1431-1465.",
	"links": null
	},
	"BIBREF9": {
	"ref_id": "b9",
	"title": "RED: A novel dataset for Romanian emotion detection from tweets",
	"authors": [
	{
	"first": "Alexandra",
	"middle": [],
	"last": "Ciobotaru",
	"suffix": ""
	},
	{
	"first": "P",
	"middle": [],
	"last": "Liviu",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Dinu",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2021)",
	"volume": "",
	"issue": "",
	"pages": "291--300",
	"other_ids": {},
	"num": null,
	"urls": [],
	"raw_text": "Alexandra Ciobotaru and Liviu P. Dinu. 2021. RED: A novel dataset for Romanian emotion detection from tweets. In Proceedings of the International Confer- ence on Recent Advances in Natural Language Pro- cessing (RANLP 2021), pages 291-300, Held Online. INCOMA Ltd.",
	"links": null
	},
	"BIBREF10": {
	"ref_id": "b10",
	"title": "Emotion recognition on twitter: Comparative study and training a",
	"authors": [
	{
	"first": "Niko",
	"middle": [],
	"last": "Colneri\u010d",
	"suffix": ""
	},
	{
	"first": "Janez",
	"middle": [],
	"last": "Dem\u0161ar",
	"suffix": ""
	}
	],
	"year": 2018,
	"venue": "",
	"volume": "",
	"issue": "",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.1109/TAFFC.2018.2807817"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Niko Colneri\u010d and Janez Dem\u0161ar. 2018. Emotion recog- nition on twitter: Comparative study and training a",
	"links": null
	},
	"BIBREF11": {
	"ref_id": "b11",
	"title": "Covid-19 vaccine coverage in health-care workers in england and effectiveness of bnt162b2 mrna vaccine against infection (siren): a prospective, multicentre, cohort study",
	"authors": [
	{
	"first": "Victoria",
	"middle": [
	"Jane"
	],
	"last": "Hall",
	"suffix": ""
	},
	{
	"first": "Sarah",
	"middle": [],
	"last": "Foulkes",
	"suffix": ""
	},
	{
	"first": "Ayoub",
	"middle": [],
	"last": "Saei",
	"suffix": ""
	},
	{
	"first": "Nick",
	"middle": [],
	"last": "Andrews",
	"suffix": ""
	},
	{
	"first": "Blanche",
	"middle": [],
	"last": "Oguti",
	"suffix": ""
	},
	{
	"first": "Andre",
	"middle": [],
	"last": "Charlett",
	"suffix": ""
	},
	{
	"first": "Edgar",
	"middle": [],
	"last": "Wellington",
	"suffix": ""
	},
	{
	"first": "Julia",
	"middle": [],
	"last": "Stowe",
	"suffix": ""
	},
	{
	"first": "Natalie",
	"middle": [],
	"last": "Gillson",
	"suffix": ""
	},
	{
	"first": "Ana",
	"middle": [],
	"last": "Atti",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "The Lancet",
	"volume": "397",
	"issue": "",
	"pages": "1725--1735",
	"other_ids": {
	"DOI": [
	"10.1016/S0140-6736(21)00790-X"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Victoria Jane Hall, Sarah Foulkes, Ayoub Saei, Nick Andrews, Blanche Oguti, Andre Charlett, Edgar Wellington, Julia Stowe, Natalie Gillson, Ana Atti, et al. 2021. Covid-19 vaccine coverage in health-care workers in england and effectiveness of bnt162b2 mrna vaccine against infection (siren): a prospective, multicentre, cohort study. The Lancet, 397(10286):1725-1735.",
	"links": null
	},
	"BIBREF12": {
	"ref_id": "b12",
	"title": "A framework to analyze the emotional reactions to mass violent events on twitter and influential factors",
	"authors": [
	{
	"first": "G",
	"middle": [
	"D"
	],
	"last": "Jonathas",
	"suffix": ""
	},
	{
	"first": "R\u00e9gis",
	"middle": [],
	"last": "Harb",
	"suffix": ""
	},
	{
	"first": "Karin",
	"middle": [],
	"last": "Ebeling",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Becker",
	"suffix": ""
	}
	],
	"year": 2020,
	"venue": "Information Processing & Management",
	"volume": "57",
	"issue": "6",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.1016/j.ipm.2020.102372"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Jonathas GD Harb, R\u00e9gis Ebeling, and Karin Becker. 2020. A framework to analyze the emotional reac- tions to mass violent events on twitter and influen- tial factors. Information Processing & Management, 57(6):102372.",
	"links": null
	},
	"BIBREF13": {
	"ref_id": "b13",
	"title": "Single-dose bnt162b2 vaccine protects against asymptomatic sars-cov-2 infection",
	"authors": [
	{
	"first": "K",
	"middle": [],
	"last": "Nick",
	"suffix": ""
	},
	{
	"first": "Lucy",
	"middle": [],
	"last": "Jones",
	"suffix": ""
	},
	{
	"first": "Shaun",
	"middle": [],
	"last": "Rivett",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Seaman",
	"suffix": ""
	},
	{
	"first": "J",
	"middle": [],
	"last": "Richard",
	"suffix": ""
	},
	{
	"first": "Ben",
	"middle": [],
	"last": "Samworth",
	"suffix": ""
	},
	{
	"first": "Chris",
	"middle": [],
	"last": "Warne",
	"suffix": ""
	},
	{
	"first": "Mark",
	"middle": [],
	"last": "Workman",
	"suffix": ""
	},
	{
	"first": "Jo",
	"middle": [],
	"last": "Ferris",
	"suffix": ""
	},
	{
	"first": "Natalie",
	"middle": [],
	"last": "Wright",
	"suffix": ""
	},
	{
	"first": "Ashley",
	"middle": [],
	"last": "Quinnell",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Shaw",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "Elife",
	"volume": "10",
	"issue": "",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.7554/eLife.68808"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Nick K Jones, Lucy Rivett, Shaun Seaman, Richard J Samworth, Ben Warne, Chris Workman, Mark Fer- ris, Jo Wright, Natalie Quinnell, Ashley Shaw, et al. 2021. Single-dose bnt162b2 vaccine protects against asymptomatic sars-cov-2 infection. Elife, 10:e68808.",
	"links": null
	},
	"BIBREF14": {
	"ref_id": "b14",
	"title": "WRIME: A new dataset for emotional intensity estimation with subjective and objective annotations",
	"authors": [
	{
	"first": "Tomoyuki",
	"middle": [],
	"last": "Kajiwara",
	"suffix": ""
	},
	{
	"first": "Chenhui",
	"middle": [],
	"last": "Chu",
	"suffix": ""
	},
	{
	"first": "Noriko",
	"middle": [],
	"last": "Takemura",
	"suffix": ""
	},
	{
	"first": "Yuta",
	"middle": [],
	"last": "Nakashima",
	"suffix": ""
	},
	{
	"first": "Hajime",
	"middle": [],
	"last": "Nagahara",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
	"volume": "",
	"issue": "",
	"pages": "2095--2104",
	"other_ids": {
	"DOI": [
	"10.18653/v1/2021.naacl-main.169"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Tomoyuki Kajiwara, Chenhui Chu, Noriko Take- mura, Yuta Nakashima, and Hajime Nagahara. 2021. WRIME: A new dataset for emotional intensity es- timation with subjective and objective annotations. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computa- tional Linguistics: Human Language Technologies, pages 2095-2104, Online. Association for Computa- tional Linguistics.",
	"links": null
	},
	"BIBREF15": {
	"ref_id": "b15",
	"title": "Applying conditional random fields to Japanese morphological analysis",
	"authors": [
	{
	"first": "Taku",
	"middle": [],
	"last": "Kudo",
	"suffix": ""
	},
	{
	"first": "Kaoru",
	"middle": [],
	"last": "Yamamoto",
	"suffix": ""
	},
	{
	"first": "Yuji",
	"middle": [],
	"last": "Matsumoto",
	"suffix": ""
	}
	],
	"year": 2004,
	"venue": "Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing",
	"volume": "",
	"issue": "",
	"pages": "230--237",
	"other_ids": {},
	"num": null,
	"urls": [],
	"raw_text": "Taku Kudo, Kaoru Yamamoto, and Yuji Matsumoto. 2004. Applying conditional random fields to Japanese morphological analysis. In Proceedings of the 2004 Conference on Empirical Methods in Natu- ral Language Processing, pages 230-237, Barcelona, Spain. Association for Computational Linguistics.",
	"links": null
	},
	"BIBREF16": {
	"ref_id": "b16",
	"title": "Determining public opinion of the covid-19 pandemic in south korea and japan: social network mining on twitter",
	"authors": [
	{
	"first": "Hocheol",
	"middle": [],
	"last": "Lee",
	"suffix": ""
	},
	{
	"first": "Eun",
	"middle": [
	"Bi"
	],
	"last": "Noh",
	"suffix": ""
	},
	{
	"first": "Bo",
	"middle": [],
	"last": "Sea Hwan Choi",
	"suffix": ""
	},
	{
	"first": "Eun",
	"middle": [
	"Woo"
	],
	"last": "Zhao",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Nam",
	"suffix": ""
	}
	],
	"year": 2020,
	"venue": "Healthcare Informatics Research",
	"volume": "26",
	"issue": "4",
	"pages": "335--343",
	"other_ids": {
	"DOI": [
	"10.4258/hir.2020.26.4.335"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Hocheol Lee, Eun Bi Noh, Sea Hwan Choi, Bo Zhao, and Eun Woo Nam. 2020. Determining public opin- ion of the covid-19 pandemic in south korea and japan: social network mining on twitter. Healthcare Informatics Research, 26(4):335-343.",
	"links": null
	},
	"BIBREF17": {
	"ref_id": "b17",
	"title": "Roberta: A robustly optimized bert pretraining approach",
	"authors": [
	{
	"first": "Yinhan",
	"middle": [],
	"last": "Liu",
	"suffix": ""
	},
	{
	"first": "Myle",
	"middle": [],
	"last": "Ott",
	"suffix": ""
	},
	{
	"first": "Naman",
	"middle": [],
	"last": "Goyal",
	"suffix": ""
	},
	{
	"first": "Jingfei",
	"middle": [],
	"last": "Du",
	"suffix": ""
	},
	{
	"first": "Mandar",
	"middle": [],
	"last": "Joshi",
	"suffix": ""
	},
	{
	"first": "Danqi",
	"middle": [],
	"last": "Chen",
	"suffix": ""
	},
	{
	"first": "Omer",
	"middle": [],
	"last": "Levy",
	"suffix": ""
	},
	{
	"first": "Mike",
	"middle": [],
	"last": "Lewis",
	"suffix": ""
	},
	{
	"first": "Luke",
	"middle": [],
	"last": "Zettlemoyer",
	"suffix": ""
	},
	{
	"first": "Veselin",
	"middle": [],
	"last": "Stoyanov",
	"suffix": ""
	}
	],
	"year": 2019,
	"venue": "",
	"volume": "",
	"issue": "",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.48550/arXiv.1907.11692"
	],
	"arXiv": [
	"arXiv:1907.11692"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Man- dar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining ap- proach. arXiv preprint arXiv:1907.11692.",
	"links": null
	},
	"BIBREF19": {
	"ref_id": "b19",
	"title": "A global database of covid-19 vaccinations",
	"authors": [
	{
	"first": "Edouard",
	"middle": [],
	"last": "Mathieu",
	"suffix": ""
	},
	{
	"first": "Hannah",
	"middle": [],
	"last": "Ritchie",
	"suffix": ""
	},
	{
	"first": "Esteban",
	"middle": [],
	"last": "Ortiz-Ospina",
	"suffix": ""
	},
	{
	"first": "Max",
	"middle": [],
	"last": "Roser",
	"suffix": ""
	},
	{
	"first": "Joe",
	"middle": [],
	"last": "Hasell",
	"suffix": ""
	},
	{
	"first": "Cameron",
	"middle": [],
	"last": "Appel",
	"suffix": ""
	},
	{
	"first": "Charlie",
	"middle": [],
	"last": "Giattino",
	"suffix": ""
	},
	{
	"first": "Lucas",
	"middle": [],
	"last": "Rod\u00e9s-Guirao",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "Nature human behaviour",
	"volume": "5",
	"issue": "7",
	"pages": "947--953",
	"other_ids": {
	"DOI": [
	"10.1038/s41562-021-01122-8"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Edouard Mathieu, Hannah Ritchie, Esteban Ortiz- Ospina, Max Roser, Joe Hasell, Cameron Appel, Charlie Giattino, and Lucas Rod\u00e9s-Guirao. 2021. A global database of covid-19 vaccinations. Nature human behaviour, 5(7):947-953.",
	"links": null
	},
	"BIBREF20": {
	"ref_id": "b20",
	"title": "Public opinion and sentiment before and at the beginning of covid-19 vaccinations in japan: Twitter analysis",
	"authors": [
	{
	"first": "Qian",
	"middle": [],
	"last": "Niu",
	"suffix": ""
	},
	{
	"first": "Junyu",
	"middle": [],
	"last": "Liu",
	"suffix": ""
	},
	{
	"first": "Masaya",
	"middle": [],
	"last": "Kato",
	"suffix": ""
	},
	{
	"first": "Yuki",
	"middle": [],
	"last": "Shinohara",
	"suffix": ""
	},
	{
	"first": "Natsuki",
	"middle": [],
	"last": "Matsumura",
	"suffix": ""
	},
	{
	"first": "Tomoki",
	"middle": [],
	"last": "Aoyama",
	"suffix": ""
	},
	{
	"first": "Momoko",
	"middle": [],
	"last": "Nagai-Tanima",
	"suffix": ""
	}
	],
	"year": 2022,
	"venue": "",
	"volume": "",
	"issue": "",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.1101/2021.07.19.21260735"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Qian Niu, Junyu Liu, Masaya Kato, Yuki Shinohara, Natsuki Matsumura, Tomoki Aoyama, and Momoko Nagai-Tanima. 2022. Public opinion and sentiment before and at the beginning of covid-19 vaccinations in japan: Twitter analysis. medRxiv.",
	"links": null
	},
	"BIBREF21": {
	"ref_id": "b21",
	"title": "Pytorch: An imperative style, high-performance deep learning library",
	"authors": [
	{
	"first": "Adam",
	"middle": [],
	"last": "Paszke",
	"suffix": ""
	},
	{
	"first": "Sam",
	"middle": [],
	"last": "Gross",
	"suffix": ""
	},
	{
	"first": "Francisco",
	"middle": [],
	"last": "Massa",
	"suffix": ""
	},
	{
	"first": "Adam",
	"middle": [],
	"last": "Lerer",
	"suffix": ""
	},
	{
	"first": "James",
	"middle": [],
	"last": "Bradbury",
	"suffix": ""
	},
	{
	"first": "Gregory",
	"middle": [],
	"last": "Chanan",
	"suffix": ""
	},
	{
	"first": "Trevor",
	"middle": [],
	"last": "Killeen",
	"suffix": ""
	},
	{
	"first": "Zeming",
	"middle": [],
	"last": "Lin",
	"suffix": ""
	},
	{
	"first": "Natalia",
	"middle": [],
	"last": "Gimelshein",
	"suffix": ""
	},
	{
	"first": "Luca",
	"middle": [],
	"last": "Antiga",
	"suffix": ""
	}
	],
	"year": 2019,
	"venue": "Advances in neural information processing systems",
	"volume": "",
	"issue": "",
	"pages": "",
	"other_ids": {},
	"num": null,
	"urls": [],
	"raw_text": "Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. 2019. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32.",
	"links": null
	},
	"BIBREF22": {
	"ref_id": "b22",
	"title": "A general psychoevolutionary theory of emotion",
	"authors": [
	{
	"first": "Robert",
	"middle": [],
	"last": "Plutchik",
	"suffix": ""
	}
	],
	"year": 1980,
	"venue": "Theories of emotion",
	"volume": "",
	"issue": "",
	"pages": "3--33",
	"other_ids": {
	"DOI": [
	"10.1016/B978-0-12-558701-3.50007-7"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Robert Plutchik. 1980. A general psychoevolutionary theory of emotion. In Theories of emotion, pages 3-33. Elsevier.",
	"links": null
	},
	"BIBREF23": {
	"ref_id": "b23",
	"title": "Consumer insight mining: aspect based twitter opinion mining of mobile phone reviews",
	"authors": [
	{
	"first": "M",
	"middle": [],
	"last": "Rathan",
	"suffix": ""
	},
	{
	"first": "R",
	"middle": [],
	"last": "Vishwanath",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Hulipalled",
	"suffix": ""
	},
	{
	"first": "L",
	"middle": [
	"M"
	],
	"last": "Venugopal",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Patnaik",
	"suffix": ""
	}
	],
	"year": 2018,
	"venue": "Applied Soft Computing",
	"volume": "68",
	"issue": "",
	"pages": "765--773",
	"other_ids": {
	"DOI": [
	"10.1016/j.asoc.2017.07.056"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "M Rathan, Vishwanath R Hulipalled, KR Venugopal, and LM Patnaik. 2018. Consumer insight mining: aspect based twitter opinion mining of mobile phone reviews. Applied Soft Computing, 68:765-773.",
	"links": null
	},
	"BIBREF24": {
	"ref_id": "b24",
	"title": "Em-paTweet: Annotating and detecting emotions on Twitter",
	"authors": [
	{
	"first": "Kirk",
	"middle": [],
	"last": "Roberts",
	"suffix": ""
	},
	{
	"first": "Michael",
	"middle": [
	"A"
	],
	"last": "Roach",
	"suffix": ""
	},
	{
	"first": "Joseph",
	"middle": [],
	"last": "Johnson",
	"suffix": ""
	},
	{
	"first": "Josh",
	"middle": [],
	"last": "Guthrie",
	"suffix": ""
	},
	{
	"first": "Sanda",
	"middle": [
	"M"
	],
	"last": "Harabagiu",
	"suffix": ""
	}
	],
	"year": 2012,
	"venue": "Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12)",
	"volume": "",
	"issue": "",
	"pages": "3806--3813",
	"other_ids": {},
	"num": null,
	"urls": [],
	"raw_text": "Kirk Roberts, Michael A. Roach, Joseph Johnson, Josh Guthrie, and Sanda M. Harabagiu. 2012. Em- paTweet: Annotating and detecting emotions on Twitter. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12), pages 3806-3813, Istanbul, Turkey. Eu- ropean Language Resources Association (ELRA).",
	"links": null
	},
	"BIBREF25": {
	"ref_id": "b25",
	"title": "Mining of opinions on covid-19 large-scale social restrictions in indonesia: public sentiment and emotion analysis on online media",
	"authors": [
	{
	"first": "Andi Muhammad Tri",
	"middle": [],
	"last": "Sakti",
	"suffix": ""
	},
	{
	"first": "Emma",
	"middle": [],
	"last": "Mohamad",
	"suffix": ""
	},
	{
	"first": "Arina Anis",
	"middle": [],
	"last": "Azlan",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "Journal of medical Internet research",
	"volume": "23",
	"issue": "8",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.2196/28249"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Andi Muhammad Tri Sakti, Emma Mohamad, and Arina Anis Azlan. 2021. Mining of opinions on covid-19 large-scale social restrictions in indone- sia: public sentiment and emotion analysis on on- line media. Journal of medical Internet research, 23(8):e28249.",
	"links": null
	},
	"BIBREF26": {
	"ref_id": "b26",
	"title": "Attention is all you need",
	"authors": [
	{
	"first": "Ashish",
	"middle": [],
	"last": "Vaswani",
	"suffix": ""
	},
	{
	"first": "Noam",
	"middle": [],
	"last": "Shazeer",
	"suffix": ""
	},
	{
	"first": "Niki",
	"middle": [],
	"last": "Parmar",
	"suffix": ""
	},
	{
	"first": "Jakob",
	"middle": [],
	"last": "Uszkoreit",
	"suffix": ""
	},
	{
	"first": "Llion",
	"middle": [],
	"last": "Jones",
	"suffix": ""
	},
	{
	"first": "Aidan",
	"middle": [
	"N"
	],
	"last": "Gomez",
	"suffix": ""
	},
	{
	"first": "\u0141ukasz",
	"middle": [],
	"last": "Kaiser",
	"suffix": ""
	},
	{
	"first": "Illia",
	"middle": [],
	"last": "Polosukhin",
	"suffix": ""
	}
	],
	"year": 2017,
	"venue": "Advances in neural information processing systems",
	"volume": "",
	"issue": "",
	"pages": "5998--6008",
	"other_ids": {},
	"num": null,
	"urls": [],
	"raw_text": "Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, \u0141ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in neural information pro- cessing systems, pages 5998-6008.",
	"links": null
	},
	"BIBREF27": {
	"ref_id": "b27",
	"title": "A study of sentiment analysis using deep learning techniques on thai twitter data",
	"authors": [
	{
	"first": "Peerapon",
	"middle": [],
	"last": "Vateekul",
	"suffix": ""
	},
	{
	"first": "Thanabhat",
	"middle": [],
	"last": "Koomsubha",
	"suffix": ""
	}
	],
	"year": 2016,
	"venue": "13th International joint conference on computer science and software engineering (JCSSE)",
	"volume": "",
	"issue": "",
	"pages": "1--6",
	"other_ids": {
	"DOI": [
	"10.1109/JCSSE.2016.7748849"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Peerapon Vateekul and Thanabhat Koomsubha. 2016. A study of sentiment analysis using deep learning techniques on thai twitter data. In 2016 13th Inter- national joint conference on computer science and software engineering (JCSSE), pages 1-6. IEEE.",
	"links": null
	},
	"BIBREF28": {
	"ref_id": "b28",
	"title": "Single-dose administration and the influence of the timing of the booster dose on immunogenicity and efficacy of chadox1 ncov-19 (azd1222) vaccine: a pooled analysis of four randomised trials",
	"authors": [
	{
	"first": "Merryn",
	"middle": [],
	"last": "Voysey",
	"suffix": ""
	},
	{
	"first": "Sue",
	"middle": [
	"Ann"
	],
	"last": "",
	"suffix": ""
	},
	{
	"first": "Costa",
	"middle": [],
	"last": "Clemens",
	"suffix": ""
	},
	{
	"first": "A",
	"middle": [],
	"last": "Shabir",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Madhi",
	"suffix": ""
	},
	{
	"first": "Y",
	"middle": [],
	"last": "Lily",
	"suffix": ""
	},
	{
	"first": "Pedro",
	"middle": [
	"M"
	],
	"last": "Weckx",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Folegatti",
	"suffix": ""
	},
	{
	"first": "Brian",
	"middle": [],
	"last": "Parvinder K Aley",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Angus",
	"suffix": ""
	},
	{
	"first": "L",
	"middle": [],
	"last": "Vicky",
	"suffix": ""
	},
	{
	"first": "Shaun",
	"middle": [
	"L"
	],
	"last": "Baillie",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Barnabas",
	"suffix": ""
	},
	{
	"first": "E",
	"middle": [],
	"last": "Qasim",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Bhorat",
	"suffix": ""
	}
	],
	"year": 2021,
	"venue": "The Lancet",
	"volume": "397",
	"issue": "",
	"pages": "881--891",
	"other_ids": {
	"DOI": [
	"10.1016/S0140-6736(21)00432-3"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Merryn Voysey, Sue Ann Costa Clemens, Shabir A Madhi, Lily Y Weckx, Pedro M Folegatti, Parvin- der K Aley, Brian Angus, Vicky L Baillie, Shaun L Barnabas, Qasim E Bhorat, et al. 2021. Single-dose administration and the influence of the timing of the booster dose on immunogenicity and efficacy of chadox1 ncov-19 (azd1222) vaccine: a pooled analysis of four randomised trials. The Lancet, 397(10277):881-891.",
	"links": null
	},
	"BIBREF29": {
	"ref_id": "b29",
	"title": "Harnessing twitter\" big data\" for automatic emotion identification",
	"authors": [
	{
	"first": "Wenbo",
	"middle": [],
	"last": "Wang",
	"suffix": ""
	},
	{
	"first": "Lu",
	"middle": [],
	"last": "Chen",
	"suffix": ""
	},
	{
	"first": "Krishnaprasad",
	"middle": [],
	"last": "Thirunarayan",
	"suffix": ""
	},
	{
	"first": "Amit P",
	"middle": [],
	"last": "Sheth",
	"suffix": ""
	}
	],
	"year": 2012,
	"venue": "2012 International Conference on Privacy, Security, Risk and Trust and 2012 International Confernece on Social Computing",
	"volume": "",
	"issue": "",
	"pages": "587--592",
	"other_ids": {
	"DOI": [
	"10.1109/SocialCom-PASSAT.2012.119"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Wenbo Wang, Lu Chen, Krishnaprasad Thirunarayan, and Amit P Sheth. 2012. Harnessing twitter\" big data\" for automatic emotion identification. In 2012 International Conference on Privacy, Security, Risk and Trust and 2012 International Confernece on So- cial Computing, pages 587-592. IEEE.",
	"links": null
	},
	"BIBREF30": {
	"ref_id": "b30",
	"title": "Transformers: State-of-the-art natural language processing",
	"authors": [
	{
	"first": "Thomas",
	"middle": [],
	"last": "Wolf",
	"suffix": ""
	},
	{
	"first": "Lysandre",
	"middle": [],
	"last": "Debut",
	"suffix": ""
	},
	{
	"first": "Victor",
	"middle": [],
	"last": "Sanh",
	"suffix": ""
	},
	{
	"first": "Julien",
	"middle": [],
	"last": "Chaumond",
	"suffix": ""
	},
	{
	"first": "Clement",
	"middle": [],
	"last": "Delangue",
	"suffix": ""
	},
	{
	"first": "Anthony",
	"middle": [],
	"last": "Moi",
	"suffix": ""
	},
	{
	"first": "Pierric",
	"middle": [],
	"last": "Cistac",
	"suffix": ""
	},
	{
	"first": "Tim",
	"middle": [],
	"last": "Rault",
	"suffix": ""
	},
	{
	"first": "Remi",
	"middle": [],
	"last": "Louf",
	"suffix": ""
	},
	{
	"first": "Morgan",
	"middle": [],
	"last": "Funtowicz",
	"suffix": ""
	},
	{
	"first": "Joe",
	"middle": [],
	"last": "Davison",
	"suffix": ""
	},
	{
	"first": "Sam",
	"middle": [],
	"last": "Shleifer",
	"suffix": ""
	},
	{
	"first": "Clara",
	"middle": [],
	"last": "Patrick Von Platen",
	"suffix": ""
	},
	{
	"first": "Yacine",
	"middle": [],
	"last": "Ma",
	"suffix": ""
	},
	{
	"first": "Julien",
	"middle": [],
	"last": "Jernite",
	"suffix": ""
	},
	{
	"first": "Canwen",
	"middle": [],
	"last": "Plu",
	"suffix": ""
	},
	{
	"first": "Teven",
	"middle": [
	"Le"
	],
	"last": "Xu",
	"suffix": ""
	},
	{
	"first": "Sylvain",
	"middle": [],
	"last": "Scao",
	"suffix": ""
	},
	{
	"first": "Mariama",
	"middle": [],
	"last": "Gugger",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Drame",
	"suffix": ""
	}
	],
	"year": 2020,
	"venue": "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
	"volume": "",
	"issue": "",
	"pages": "38--45",
	"other_ids": {
	"DOI": [
	"10.18653/v1/2020.emnlp-demos.6"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier- ric Cistac, Tim Rault, Remi Louf, Morgan Funtow- icz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoest, and Alexander Rush. 2020. Trans- formers: State-of-the-art natural language processing. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 38-45, Online. Association for Computational Linguistics.",
	"links": null
	},
	"BIBREF31": {
	"ref_id": "b31",
	"title": "Google's neural machine translation system: Bridging the gap between human and machine translation",
	"authors": [
	{
	"first": "Yonghui",
	"middle": [],
	"last": "Wu",
	"suffix": ""
	},
	{
	"first": "Mike",
	"middle": [],
	"last": "Schuster",
	"suffix": ""
	},
	{
	"first": "Zhifeng",
	"middle": [],
	"last": "Chen",
	"suffix": ""
	},
	{
	"first": "V",
	"middle": [],
	"last": "Quoc",
	"suffix": ""
	},
	{
	"first": "Mohammad",
	"middle": [],
	"last": "Le",
	"suffix": ""
	},
	{
	"first": "Wolfgang",
	"middle": [],
	"last": "Norouzi",
	"suffix": ""
	},
	{
	"first": "Maxim",
	"middle": [],
	"last": "Macherey",
	"suffix": ""
	},
	{
	"first": "Yuan",
	"middle": [],
	"last": "Krikun",
	"suffix": ""
	},
	{
	"first": "Qin",
	"middle": [],
	"last": "Cao",
	"suffix": ""
	},
	{
	"first": "Klaus",
	"middle": [],
	"last": "Gao",
	"suffix": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Macherey",
	"suffix": ""
	}
	],
	"year": 2016,
	"venue": "",
	"volume": "",
	"issue": "",
	"pages": "",
	"other_ids": {
	"DOI": [
	"10.48550/arXiv.1609.08144"
	],
	"arXiv": [
	"arXiv:1609.08144"
	]
	},
	"num": null,
	"urls": [],
	"raw_text": "Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, et al. 2016. Google's neural machine translation system: Bridging the gap between human and machine trans- lation. arXiv preprint arXiv:1609.08144.",
	"links": null
	}
	},
	"ref_entries": {
	"FIGREF0": {
	"text": "Box plots of distributions of predicted emotion scores for writers and readers. Whiskers are draw at 1.5 \u00d7 inter-quartile range from the first and third quartiles. Mean predicted scores are indicated by black points.",
	"type_str": "figure",
	"uris": null,
	"num": null
	},
	"FIGREF1": {
	"text": "Q-Q plots of predicted writer and reader emotions.",
	"type_str": "figure",
	"uris": null,
	"num": null
	},
	"FIGREF2": {
	"text": "Average score for each emotion and the number of new vaccinations, people vaccinated, and people fully vaccinated for several days in December 2021. Vaccination data missing due to gaps in the dataset is represented with broken lines.(a) With vaccinations. (b) With people vaccinated. (c) With people fully vaccinated.",
	"type_str": "figure",
	"uris": null,
	"num": null
	},
	"FIGREF3": {
	"text": "Correlation analysis of emotion scores and vaccination statistics. Emotion scores are scatter plotted against vaccination numbers with best fit lines. No significant correlations are found for emotions with people vaccinated and people fully vaccinated. unison model. IEEE transactions on affective computing, 11(3):433-446. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171-4186, Minneapolis, Minnesota. Association for Computational Linguistics.",
	"type_str": "figure",
	"uris": null,
	"num": null
	},
	"TABREF1": {
	"type_str": "table",
	"content": "<table><tr><td>Average # of to-kens per text Five most com-mon POS (de-scending)</td><td>Vaccine-related Tweets 51 ADP, AUX, NOUN, PUNCT, VERB</td><td>WRIME 23 ADP, PUNCT, SYM, NOUN, VERB</td></tr></table>",
	"text": "Number of Tweets per keyword in the vaccine Tweet dataset. Keywords are translated from Japanese. Some Tweets contain multiple keywords.",
	"html": null,
	"num": null
	},
	"TABREF2": {
	"type_str": "table",
	"content": "<table/>",
	"text": "Average number of tokens and the five most common parts of speech using UniDic. 1",
	"html": null,
	"num": null
	},
	"TABREF3": {
	"type_str": "table",
	"content": "<table><tr><td>Writer</td><td>0</td><td>3</td><td>0</td><td>1</td><td>3</td><td>0</td><td>0</td><td>0</td></tr><tr><td>Reader</td><td>0</td><td>2.33</td><td>0</td><td>2.33</td><td>0.33</td><td>1</td><td>0.67</td><td>0</td></tr></table>",
	"text": "The tire of my car was flat. I heard that it might be mischief.) Annotator Joy Sadness Anticipation Surprise Anger Fear Disgust Trust",
	"html": null,
	"num": null
	},
	"TABREF4": {
	"type_str": "table",
	"content": "<table><tr><td>Model BoW+LinReg fastText+SVM 1.141 Joy 0.889 BERT (ours) 0.658 BoW+LinReg 0.351 fastText+SVM 0.374 BERT (ours) 0.192 BoW+LinReg 0.615 fastText+SVM 0.758 BERT (ours) 0.425</td><td>Sadness Anticipation Surprise Anger Fear Disgust Trust Overall Writer Emotions 0.830 0.849 0.617 0.605 0.536 0.822 0.399 0.692 0.973 1.066 0.801 0.754 0.628 1.032 0.394 0.849 0.688 0.746 0.542 0.486 0.462 0.664 0.400 0.581 Reader Emotions 0.270 0.344 0.172 0.049 0.201 0.175 0.037 0.200 0.297 0.422 0.177 0.047 0.269 0.221 0.090 0.237 0.178 0.211 0.139 0.032 0.147 0.123 0.029 0.131 Both Emotions 0.550 0.597 0.394 0.327 0.368 0.499 0.218 0.446 0.635 0.744 0.489 0.400 0.449 0.626 0.242 0.543 0.433 0.479 0.341 0.259 0.304 0.394 0.214 0.356</td></tr></table>",
	"text": "Sample training data point. While WRIME has annotations for three separate readers, we create only one set by averaging the reader scores per emotion.",
	"html": null,
	"num": null
	},
	"TABREF5": {
	"type_str": "table",
	"content": "<table/>",
	"text": "",
	"html": null,
	"num": null
	},
	"TABREF6": {
	"type_str": "table",
	"content": "<table><tr><td>Annotator Writer Reader</td><td>Joy Sadness Anticipation Surprise Anger Fear Disgust Trust 0.019 1.206 -0.069 0.785 2.424 0.070 2.224 -0.106 0.049 0.648 0.120 0.657 1.573 0.369 2.139 -0.060</td></tr><tr><td>Tweet</td><td>\u4e45\u3057\u3076\u308a\u306e2\u9023\u4f11\u3002\u660e\u65e5\u5e73\u65e5\u4f11\u307f\uff01\uff01(Haven't had back-to-back holidays in a while. Tomorrow is a weekday break!!)</td></tr><tr><td>Annotator Writer Reader</td><td>Joy Sadness Anticipation Surprise Anger Fear Disgust Trust 2.989 0.171 0.469 0.107 0.069 0.126 0.827 2.111 2.396 0.049 1.567 0.241 -0.011 0.022 -0.005 0.023</td></tr></table>",
	"text": "Tweet\u4e00\u6642\u9593\u5f85\u3063\u3066\u3082\u3053\u306a\u3044\uff01\uff01\u30ac\u30b9\u4f1a\u793e\u306a\u3093\u306a\u306e\uff01\uff01\u6025\u3044\u3067\u5e30\u3063\u3066\u304d\u305f \u306e\u306b?\u6642\u9593\u5b88\u3089\u306a\u3044\u3072\u3068\u304d\u3089\u3044\uff01\uff01\uff01\u306a\u3093\u304b\u98df\u3079\u7269\u3067\u3082\u6301\u3063\u3066\u304d\u305f\u3089 \u8a31\u305d\u3046\u3002(I waited for an hour but no one came!! What's up with this gas company!! Even though I hurried back? I hate people who can't keep track of time!!! Maybe I'll forgive them though if they bring me food.)",
	"html": null,
	"num": null
	},
	"TABREF7": {
	"type_str": "table",
	"content": "<table/>",
	"text": "Inferred emotion scores for data points from the WRIME test set. English translations are provided in parentheses after each Japanese Tweet. The two most intense emotions per annotator for each Tweet are in bold.",
	"html": null,
	"num": null
	},
	"TABREF9": {
	"type_str": "table",
	"content": "<table><tr><td>: Kolmogorov-Smirnov test results for emotions between Tweets referencing Moderna and Pfizer. Sig-nificant KS statistics and p-values are in bold.</td></tr></table>",
	"text": "",
	"html": null,
	"num": null
	},
	"TABREF10": {
	"type_str": "table",
	"content": "<table><tr><td>Annotator Writer Reader</td><td>Joy Sadness Anticipation Surprise Anger Fear Disgust Trust 0.180 0.675 0.214 1.250 2.138 0448 2.077 0.259 -0.030 0.648 0.104 0.762 1.147 0.575 1.848 0.090</td></tr><tr><td>Tweet</td><td>\u7686\u69d8\u3001\u304a\u75b2\u308c\u69d8\u3067\u3059\u3002\u4eca\u65e5\u306f\u3088\u3046\u3084\u304f\u30ef\u30af\u30c1\u30f3\u63a5\u7a2e\u3057\u307e\u3057\u305f\u3002\u8155\u304c</td></tr><tr><td/><td>\u3061\u3087\u3063\u3068\u75db\u3044\u3067\u3059\u304c\u3001\u305d\u308c\u4ee5\u5916\u306b\u554f\u984c\u306f\u3042\u308a\u307e\u305b\u3093\u3002\u795e\u5bae\u7403\u5834\u306b\u3082</td></tr><tr><td/><td>\u3061\u3087\u3053\u3063\u3068\u9854\u3092\u51fa\u3057\u3066\u30ce\u30e0\u3055\u3093\u3092\u5072\u3093\u3067\u304d\u307e\u3057\u305f\u3002\u7686\u69d8\u3001\u8eab\u306e\u5b89\u5168\u78ba</td></tr><tr><td/><td>\u4fdd\u3068\u4f53\u8abf\u7ba1\u7406\u306b\u6c17\u3092\u3064\u3051\u3066\u3001\u304a\u904e\u3054\u3057\u4e0b\u3055\u3044\u3002\u4eca\u65e5\u3082\u4e00\u65e5\u304a\u75b2\u308c\u69d8\u3067</td></tr><tr><td colspan=\"2\">\u3057\u305f\u3002(Annotator Joy Sadness Anticipation Surprise Anger Fear Disgust Trust Writer 2.020 0.428 1.321 0.594 0.075 0.145 0.014 2.069 Reader 1.582 0.100 0.827 -0.089 -0.038 0.239 0.046 0.419</td></tr></table>",
	"text": "There are close contacts with the Omicron strain among spectators at the Emperor's Cup on December 16, 2021.) Thank you everyone for your hard work. Today, I finally got vaccinated. My arm hurts a little bit, but other than that there are no problems. I also swung by Jing\u016b Stadium for a bit to think about Mr. Nomu. Everyone, please look after your safety and health. Thank you for working hard today.)",
	"html": null,
	"num": null
	},
	"TABREF11": {
	"type_str": "table",
	"content": "<table/>",
	"text": "Inferred emotion scores for data points from the constructed vaccine-related Tweet dataset. English translations are provided in parentheses after each Japanese Tweet. Links are removed. The two most intense emotions per annotator for each Tweet are in bold.",
	"html": null,
	"num": null
	}
	}
	}
	}