Datasets:

Modalities:
Text
Formats:
json
Languages:
English
ArXiv:
Libraries:
Datasets
pandas
License:
espejelomar commited on
Commit
bdfd116
1 Parent(s): d282209

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +25 -10
README.md CHANGED
@@ -7,6 +7,7 @@ pretty_name: SPECTER
7
  ---
8
 
9
  # Dataset Card for "ESPECTER"
 
10
  ## Table of Contents
11
  - [Dataset Description](#dataset-description)
12
  - [Dataset Summary](#dataset-summary)
@@ -30,7 +31,9 @@ pretty_name: SPECTER
30
  - [Licensing Information](#licensing-information)
31
  - [Citation Information](#citation-information)
32
  - [Contributions](#contributions)
 
33
  ## Dataset Description
 
34
  - **Homepage:** [https://github.com/allenai/specter](https://github.com/allenai/specter)
35
  - **Repository:** [More Information Needed](https://github.com/allenai/specter/blob/master/README.md)
36
  - **Paper:** [More Information Needed](https://arxiv.org/pdf/2004.07180.pdf)
@@ -38,24 +41,39 @@ pretty_name: SPECTER
38
  - **Size of downloaded dataset files:**
39
  - **Size of the generated dataset:**
40
  - **Total amount of disk used:** 38.3 MB
 
41
  ### Dataset Summary
42
- SPECTER: Document-level Representation Learning using Citation-informed Transformers. A new method to generate document-level embedding of scientific documents based on pretraining a Transformer language model on a powerful signal of document-level relatedness: the citation graph. Unlike existing pretrained language models, SPECTER can be easily applied to downstream applications without task-specific fine-tuning.
 
 
 
 
 
 
43
  ### Supported Tasks and Leaderboards
 
44
  [More Information Needed](https://github.com/allenai/specter)
 
45
  ### Languages
 
46
  [More Information Needed](https://github.com/allenai/specter)
 
47
  ## Dataset Structure
48
- Specter requires two main files as input to embed the document. A text file with ids of the documents you want to embed and a json metadata file consisting of the title and abstract information. Sample files are provided in the `data/` directory to get you started. Input data format is according to:
 
 
 
 
 
49
  metadata.json format:
50
 
 
 
51
  {
52
  'doc_id': {'title': 'representation learning of scientific documents',
53
  'abstract': 'we propose a new model for representing abstracts'},
54
  }
55
-
56
- ### Data Instances
57
- ### Data Splits
58
- ## Dataset Creation
59
 
60
  ### Curation Rationale
61
 
@@ -119,6 +137,7 @@ metadata.json format:
119
  year={2020}
120
  }
121
 
 
122
  SciDocs benchmark
123
 
124
  SciDocs evaluation framework consists of a suite of evaluation tasks designed for document-level tasks.
@@ -127,11 +146,7 @@ Link to SciDocs:
127
 
128
  - [https://github.com/allenai/scidocs](https://github.com/allenai/scidocs)
129
 
130
- ```
131
-
132
 
133
  ### Contributions
134
 
135
  Thanks to [@armancohan](https://github.com/armancohan), [@sergeyf](https://github.com/sergeyf), [@haroldrubio](https://github.com/haroldrubio), [@jinamshah](https://github.com/jinamshah)
136
-
137
- ---
 
7
  ---
8
 
9
  # Dataset Card for "ESPECTER"
10
+
11
  ## Table of Contents
12
  - [Dataset Description](#dataset-description)
13
  - [Dataset Summary](#dataset-summary)
 
31
  - [Licensing Information](#licensing-information)
32
  - [Citation Information](#citation-information)
33
  - [Contributions](#contributions)
34
+
35
  ## Dataset Description
36
+
37
  - **Homepage:** [https://github.com/allenai/specter](https://github.com/allenai/specter)
38
  - **Repository:** [More Information Needed](https://github.com/allenai/specter/blob/master/README.md)
39
  - **Paper:** [More Information Needed](https://arxiv.org/pdf/2004.07180.pdf)
 
41
  - **Size of downloaded dataset files:**
42
  - **Size of the generated dataset:**
43
  - **Total amount of disk used:** 38.3 MB
44
+
45
  ### Dataset Summary
46
+
47
+ SPECTER: Document-level Representation Learning using Citation-informed Transformers.
48
+ A new method to generate document-level embedding of scientific documents based on
49
+ pretraining a Transformer language model on a powerful signal of document-level relatedness: the citation graph.
50
+ Unlike existing pretrained language models, SPECTER can be easily applied to
51
+ downstream applications without task-specific fine-tuning.
52
+
53
  ### Supported Tasks and Leaderboards
54
+
55
  [More Information Needed](https://github.com/allenai/specter)
56
+
57
  ### Languages
58
+
59
  [More Information Needed](https://github.com/allenai/specter)
60
+
61
  ## Dataset Structure
62
+ Specter requires two main files as input to embed the document.
63
+ A text file with ids of the documents you want to embed and a json metadata file
64
+ consisting of the title and abstract information.
65
+ Sample files are provided in the `data/` directory to get you started.
66
+ Input data format is according to:
67
+
68
  metadata.json format:
69
 
70
+ ```
71
+
72
  {
73
  'doc_id': {'title': 'representation learning of scientific documents',
74
  'abstract': 'we propose a new model for representing abstracts'},
75
  }
76
+ ```
 
 
 
77
 
78
  ### Curation Rationale
79
 
 
137
  year={2020}
138
  }
139
 
140
+ ```
141
  SciDocs benchmark
142
 
143
  SciDocs evaluation framework consists of a suite of evaluation tasks designed for document-level tasks.
 
146
 
147
  - [https://github.com/allenai/scidocs](https://github.com/allenai/scidocs)
148
 
 
 
149
 
150
  ### Contributions
151
 
152
  Thanks to [@armancohan](https://github.com/armancohan), [@sergeyf](https://github.com/sergeyf), [@haroldrubio](https://github.com/haroldrubio), [@jinamshah](https://github.com/jinamshah)