README.md

metadata

license: mit
dataset_info:
  features:
    - name: 'Unnamed: 0'
      dtype: int64
    - name: claim_input
      dtype: string
    - name: claim_idx
      dtype: int64
    - name: applicationNumber
      dtype: int64
    - name: applicationTypeCategory
      dtype: string
    - name: relatedDocumentData
      dtype: float64
    - name: patentClassification
      dtype: string
    - name: applicantCitedExaminerReferenceIndicatorCount
      dtype: float64
    - name: filingDate
      dtype: string
    - name: publicationDate
      dtype: string
    - name: claimNumberArrayDocument
      dtype: float64
    - name: abstract
      dtype: string
    - name: percentile
      dtype: float64
    - name: claim_label_101
      dtype: int64
    - name: claim_label_102
      dtype: int64
    - name: claim_label_103
      dtype: int64
    - name: claim_label_112
      dtype: int64
    - name: relatedApplicationNumber
      dtype: string
    - name: max_score_x
      dtype: float64
    - name: mean_score
      dtype: float64
    - name: max_citations
      dtype: float64
    - name: max_other_citations
      dtype: float64
    - name: max_article_citations
      dtype: float64
    - name: max_score_y
      dtype: float64
    - name: component
      dtype: int64
    - name: is_closed
      dtype: int64
    - name: is_open
      dtype: int64
    - name: is_half
      dtype: int64
    - name: similarity_product
      dtype: float64
    - name: transitional_phrase
      dtype: string
    - name: app_claim_id
      dtype: string
    - name: bert_scores_102
      dtype: float64
    - name: claim_label_combined
      dtype: int64
    - name: bert_scores_101
      dtype: float64
    - name: combined_pred_scores
      dtype: float64
    - name: claim_label_101_adjusted
      dtype: int64
    - name: bert_score_102_app_feats_no_hinge
      dtype: float64
    - name: bert_score_101_app_feats
      dtype: float64
    - name: bert_score_102_app_feats_w_hinge
      dtype: float64
    - name: bert_score_102_no_app_feats
      dtype: float64
    - name: dataset
      dtype: string
    - name: lexical_diversity
      dtype: float64
    - name: patent_class
      dtype: float64
    - name: foreignPriority
      dtype: bool
  splits:
    - name: train
      num_bytes: 2152439051
      num_examples: 1485693
    - name: validation
      num_bytes: 402312853
      num_examples: 278215
    - name: test
      num_bytes: 267789008
      num_examples: 185477
  download_size: 501745093
  dataset_size: 2822540912
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

PatentAP

A dataset for the task of Patent Approval Prediction, which is proposed in the paper "Beyond Scaling: Predicting Patent Approval with Domain-specific Fine-grained Claim Dependency Graph".

Dataset Details

More details will be added soon!