INK-USC/USC-DS-RelationExtraction
A research system for distantly supervised relation extraction from text using knowledge bases and ML models.
Not currently ranked — collecting fresh signals.
star history
This repository implements sentence-level relation extraction using distant supervision, as described in the CoType WWW'17 paper. It uses knowledge bases like Freebase and DBpedia to heuristically label entity pairs in text, then trains ML models to identify relation types from sentence context. The pipeline includes entity detection via Stanford NER, entity linking to knowledge bases, and alignment of KB facts to sentences for training data generation.
Frequently asked
- What is INK-USC/USC-DS-RelationExtraction?
- A research system for distantly supervised relation extraction from text using knowledge bases and ML models.
- Is USC-DS-RelationExtraction open source?
- Yes — INK-USC/USC-DS-RelationExtraction is open source, released under the MIT license.
- What language is USC-DS-RelationExtraction written in?
- INK-USC/USC-DS-RelationExtraction is primarily written in C++.
- How popular is USC-DS-RelationExtraction?
- INK-USC/USC-DS-RelationExtraction has 420 stars on GitHub.
- Where can I find USC-DS-RelationExtraction?
- INK-USC/USC-DS-RelationExtraction is on GitHub at https://github.com/INK-USC/USC-DS-RelationExtraction.