​

MENTORED BY: DR. MANOJ AGARWAL {PRINCIPAL RESEARCH SCIENTIST (RESEARCH AND AI), HYDERABAD, INDIA}

• My responsibility included collecting DBLP dataset with abstract, title, citation network, authors.

• It was a Sequence Models and statistical language modeling based project aiming at building a system for EXPERT FINDING problem using Attention Encoder-Decoder (RNN comprising of bidirectional-LSTMs) with CopyNet layer for keyphrase generation over the text of papers from DBLP-dataset and train FastText model on the text to get word embeddings of the keyphrases,i.e.create a Computer Science Dictionary to convert extracted keywords to embedding.

• Later NMSLIB(An efficient similarity search library and a toolkit for evaluation of k-NN methods for generic non-metric spaces) is used to get the nearest neighbor evaluation with Cosine Similarity as Distance Metrics.

• Now, the end goal of dividing papers into topics based on ACM-Library is achieved so when a query is parsed, papers belonging to its nearest neighbor topics are considered for the Ranking Algorithm to get the Experts in that Domain.