This repository provides the implementation of Large Language Models (LLM) based extractions of facts from conference proceedings front matter and conference websites.

The prompts used to extract the number of submitted and accepted papers, members of the organization committee and their roles, members of the programme committee with the corresponding track and role (PC/SPC), important dates (deadline) along with their activity and corresponding track, conference topics, etc. are found in prompt_templates.py.

scholarly-wikidata repo currently implements the following extractors.

• acceptance_ratio_processor
• conference_topic_processor.py
• important_dates_processor.py
• organization_committee_role_processor.py
• program_committee_processor.py

dblp_to_open_refine.ipynb provides the code needed to extract the papers and their corresponding authors from DBLP for a given conference series and prepare CSV files that can be imported to OpenRefine for entity linking / reconciliation to Wikidata and prepare the Quick Statements to be submitted to Wikidata.

The source data required to run the different extractors are distributed in Zenodo.