ParallelDots Research & Development

Content Based Document Recommender using Deep Learning

Can a recommendation engine actually recommend documents not just by by plain searching but understanding the language of the query and the indexed documents. In this paper, we try to create a recommendation engine which gives the most relevant healthcare publications (BioASQ2016 dataset) by understanding the language of query. We use a modification of DSSM networks which uses word embeddings instead of character ngrams for this purpose.