Open HridayM25 opened 9 months ago
This sounds great. Am interested to work on the issue. A small clarification, the agenda of this issue is like if a user want to asks a query related to the giver code, the LLM must answer it, using KG. If am not wrong.
Yes, this is also something interesting that can be looked at (retrieval). What I had thought of was to have some kind of knowledge graphs between the individual files/ clusters of files. This can be done to explain any dependencies between these files and exploit them for other purposes.
Contributors can look into incorporating knowledge graphs based on the clustered or non-clustered files to provide further insight in the structure of the code. Retrieval of code can also be a subsequent feature after this.