Datasets:
pqai
/

Modalities:
Text
Formats:
text
Libraries:
Datasets
License:
PoC / README.md
maheshmaan's picture
Add dataset card (README)
ae8354d
metadata
license: mit

PoC (Patents with One Citation) dataset

This dataset is useful for training or evaluating models that predict patent-to-patent similarity, such as those used for patent searching.

It was developed and used for the training of an ML model that powers the PQAI search engine.

Details

The dataset contains 90,013 samples.

Each sample contains:

  • a subject patent (sp)
  • its only citation (cit)
  • its CPC code (cpc)
  • a list of 10 patents (sims) that are similar to sp (in that they share the CPC code) and published before sp

Every line of the dataset is a JSON parsable string (.jsonl format), which upon parsing given an array of this format:

[pn, cit, cpc, [...sims]]

Task

Given the subject patent sp the task is to assign a similarity score to each patent [cit, ...sims]. Ideally, the score should be maximum for cit.

Metrics

It's a ranking task, so the following metrics make the most sense:

  • DCG/NDCG
  • Accuracy