Google Summer of Code 2024: Contributor Proposal - Bringing Together LLMs and RDF Knowledge Graphs
- Mentors
- Johannes F, Bruno Messias, Hannes Hartmann, Akshay NS, Mahdi Hedayat Mahmoudi
- Organization
- DBpedia
- Technologies
- c, python, java, sparql
- Topics
- machine learning, information retrieval, Large Language Models, Vector Databases
This project seeks to improve the capabilities of Large Language Models in interfacing RDF Data and RDF Knowledge Graphs. This is to be achieved through the creation of a vector database based on LLM representations of the ontologies present in the RDF Knowledge Graphs. This database would then be used to search for ontologies relevant to the user input and allow the LLM to make use of the information present in DBpedia.
Deliverables are to include: the source code for development of the vector database, additional source code for integrating this database in the DPedia LLM plugin(s) (which would include user input processing (such as Named Entity Recognition) and database match ranking), as well as benchmarking/test results and all related documentation.