README.md

metadata

license: cc-by-4.0
task_categories:
  - text-classification
language:
  - en
tags:
  - SDQP
  - scholarly
  - citation_count_prediction
  - review_score_prediction
dataset_info:
  features:
    - name: paperhash
      dtype: string
    - name: s2_corpus_id
      dtype: string
    - name: arxiv_id
      dtype: string
    - name: title
      dtype: string
    - name: abstract
      dtype: string
    - name: authors
      sequence:
        - name: name
          dtype: string
        - name: affiliation
          struct:
            - name: laboratory
              dtype: string
            - name: institution
              dtype: string
            - name: location
              dtype: string
    - name: summary
      dtype: string
    - name: field_of_study
      sequence: string
    - name: venue
      dtype: string
    - name: publication_date
      dtype: string
    - name: n_references
      dtype: int32
    - name: n_citations
      dtype: int32
    - name: n_influential_citations
      dtype: int32
    - name: introduction
      dtype: string
    - name: background
      dtype: string
    - name: methodology
      dtype: string
    - name: experiments_results
      dtype: string
    - name: conclusion
      dtype: string
    - name: full_text
      dtype: string
    - name: decision
      dtype: bool
    - name: decision_text
      dtype: string
    - name: reviews
      sequence:
        - name: review_id
          dtype: string
        - name: review
          struct:
            - name: title
              dtype: string
            - name: paper_summary
              dtype: string
            - name: main_review
              dtype: string
            - name: strength_weakness
              dtype: string
            - name: questions
              dtype: string
            - name: limitations
              dtype: string
            - name: review_summary
              dtype: string
        - name: score
          dtype: float32
        - name: confidence
          dtype: float32
        - name: novelty
          dtype: float32
        - name: correctness
          dtype: float32
        - name: clarity
          dtype: float32
        - name: impact
          dtype: float32
        - name: reproducibility
          dtype: float32
        - name: ethics
          dtype: string
    - name: comments
      sequence:
        - name: title
          dtype: string
        - name: comment
          dtype: string
    - name: references
      sequence:
        - name: paperhash
          dtype: string
        - name: title
          dtype: string
        - name: abstract
          dtype: string
        - name: authors
          sequence:
            - name: name
              dtype: string
            - name: affiliation
              struct:
                - name: laboratory
                  dtype: string
                - name: institution
                  dtype: string
                - name: location
                  dtype: string
        - name: arxiv_id
          dtype: string
        - name: s2_corpus_id
          dtype: string
        - name: intents
          sequence: string
        - name: isInfluential
          dtype: bool
    - name: hypothesis
      dtype: string
    - name: month_since_publication
      dtype: int32
    - name: avg_citations_per_month
      dtype: float32
  splits:
    - name: train
      num_bytes: 160996851
      num_examples: 2060
    - name: validation
      num_bytes: 758528471
      num_examples: 9013
    - name: test
      num_bytes: 853299931
      num_examples: 9014
  download_size: 872309865
  dataset_size: 1772825253
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

Datasets related to the task of Scholarly Document Quality Prediction (SDQP). Each sample is an academic paper for which either the citation count or the review score can be predicted (depending on availability). The information that is potentially available for each sample can be found below.

ACL-OCL Extended

A dataset for citation count prediction only, based on the ACL-OCL dataset. Extended with updated citation counts, references and annotated research hypothesis

OpenReview

A dataset for review score and citation count prediction, obtained by parsing OpenReview. Due to licensing the dataset comes in 3 formats:

1. openreview-public: Contains full information on all OpenReview submissions that are accompanied by a license.
2. openreview-full-light: The full dataset excluding the parsed pdfs of the submitted papers.
3. openreview-full: A script to obtain the full dataset with submissions.

Citation

If you use the dataset in your work, please cite:

The data model for the papers:

Paper Data Model

{
# ID's
"paperhash": str,
"arxiv_id": str | None,
"s2_corpus_id": str | None,

# Basic Info
"title":str,
"authors": list[Author],
"abstract": str | None,
"summary": str | None,
"publication_date": str | None,

# OpenReview Metadata
"field_of_study": list[str] | str | None,
"venue": str | None,

# s2 Metadata
"n_references": int | None,
"n_citations": int | None,
"n_influential_citations": int | None,
"open_access": bool | None,
"external_ids": dict | None,
"pdf_url": str | None,

# Content
"parsed_pdf": dict | None,
"parsed_latex": dict | None,
"structured_content": dict[str, Section],

# Review Data
"openreview": bool,
"decision": bool | None,
"decision_text": str | None,
"reviews": list[Review] | None,
"comments": list[Comment] | None,

# References
"references": list[Reference] | None,
"bibref2section": dict,
"bibref2paperhash": dict,

# Hypothesis
"hypothesis": dict | None
}

Author Data Model

{
"name":str,
"affiliation": {
  "laboratory": str | dict | None,
  "institution": str | dict | None, 
  "location": str | dict | None
  }
}

Reference Data Model

{
"paperhash": str,
"title": str,
"abstract": str = "",
"authors": list[Author],

# IDs
"arxiv_id": str | None,
"s2_corpus_id": str | None,
"external_ids": dict| None,

# Reference specific info
"intents": list[str] | None = None,
"isInfluential": bool | None = None
}

Comment Data Model

{
"title": str,
"comment": str
}

Section Data Model

{
"name": str,
"sec_num": str,
"classification": str,
"text": str,
"subsections": list[Section]
}

Review Data Model

{
"review_id": str,
"review": {
  "title": str | None,
  "paper_summary": str | None,
  "main_review": str | None,
  "strength_weakness": str | None,
  "questions": str | None,
  "limitations": str | None,
  "review_summary": str | None
}
"score": float | None,
"confidence": float | None,
"novelty": float | None,
"correctness": float | None,
"clarity": float | None,
"impact": float | None,
"reproducibility": float | None,
"ethics": str | None
}