sequoiaandrade
commited on
Commit
•
3ec547b
1
Parent(s):
498f016
updated read me
Browse files
README.md
CHANGED
@@ -6,11 +6,12 @@ tags:
|
|
6 |
- sentence-similarity
|
7 |
---
|
8 |
|
9 |
-
#
|
|
|
10 |
|
11 |
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
12 |
|
13 |
-
|
14 |
|
15 |
## Usage (Sentence-Transformers)
|
16 |
|
@@ -35,10 +36,12 @@ print(embeddings)
|
|
35 |
|
36 |
## Evaluation Results
|
37 |
|
38 |
-
|
39 |
-
|
40 |
-
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
|
41 |
|
|
|
|
|
|
|
|
|
42 |
|
43 |
## Training
|
44 |
The model was trained with the parameters:
|
@@ -87,4 +90,4 @@ SentenceTransformer(
|
|
87 |
|
88 |
## Citing & Authors
|
89 |
|
90 |
-
|
|
|
6 |
- sentence-similarity
|
7 |
---
|
8 |
|
9 |
+
# Manager for Intelligent Knowledge Access (MIKA)
|
10 |
+
# Custom Information Retrieval Model
|
11 |
|
12 |
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
13 |
|
14 |
+
The model is custom trained on engineering documents for asymmetric infromation retrieval. It is intended to be used to identify engineering documents relevant to a query for use in design time. For example, a repository can be queried to find support for requirements or learn more about a specific type of failure.
|
15 |
|
16 |
## Usage (Sentence-Transformers)
|
17 |
|
|
|
36 |
|
37 |
## Evaluation Results
|
38 |
|
39 |
+
This model was evaluated on three queries using precision at k for k=10,20, and 30. Mean average precision (MAP) was also calculated. The model was baselines against the pre-trained SBERT.
|
|
|
|
|
40 |
|
41 |
+
|IR Method | MAP |
|
42 |
+
|----------|-----|
|
43 |
+
|Pre-trained sBERT| 0.648|
|
44 |
+
|Fine-tuned sBERT| 0.807|
|
45 |
|
46 |
## Training
|
47 |
The model was trained with the parameters:
|
|
|
90 |
|
91 |
## Citing & Authors
|
92 |
|
93 |
+
Walsh, HS, & Andrade, SR. "Semantic Search With Sentence-BERT for Design Information Retrieval." Proceedings of the ASME 2022 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. Volume 2: 42nd Computers and Information in Engineering Conference (CIE). St. Louis, Missouri, USA. August 14–17, 2022. V002T02A066. ASME. https://doi.org/10.1115/DETC2022-89557
|