sigsegvsquad / wordlink-pharmasearch Goto Github PK

Thanks to the efforts of @OverPoweredDev and @prabhav-pandya , we have a few baseline approaches using the pretrained pubmed model and various sections of the documents. Further work in making new Approaches, possibly with changes in mapping (apart from something like cosine similarity and euclidean distance) or with preprocessing is crucial to the success of the Project

So far, the documents used in search_list_txt/ have been succesfully segregated into parts of Abstract, Introduction and all. Parts in the abstract have even been cleaned up so do the same for the other parts i.e. remove garbage text left from pdf scanning, random numbers left as aresult of converting tables to text.

Note: Do NOT change the Document Sections i.e. Abstract, Keywords, Introduction, Main, Conclusion, References, Acknowledgements, End