AI-ML

Designed an algorithm to automatically identify high quality content from Q/A sites (Yahoo Answers).

Analysed the usage of twitter hashtags associated with the natural disaster to identify the different phases of the natural disaster.

Designed a link-prediction algorithm to predict the university of authors based on the location of co-authors, number of co-authored papers, most recent paper, and the areas of research of the authors.

Implemented unigram and trigram probablistic language models and provided a detailed comparative analysis of the same based on perplexity score and qualitative analysis.

Implemented an algorithm to rank tweets related to news articles by generating a query and rank based on Lucene’s Scoring function.

Created a full-stack search portal for institute senate meeting minutes which takes .pdf/.docx documents, automatically identifies the subsections and paragraphs in the document, and provides intelligent search on them using Elasticsearch backend.

Implemented a sequence-to-sequence model to translate the text from one language to another. Compared the performance of various sampling strategies used for decoding text using BLEU score.