↓Skip to main content

Projects

Table of Contents

Focus #

My Master’s research ignited a passion for developing and implementing novel algorithms that leverage knowledge graphs to enhance information retrieval and semantic understanding. This experience solidified my interest in pursuing a PhD in Knowledge Graphs and High Performance Computing, with the goal of contributing to advancements in areas like Natural Language Processing, Recommender Systems, and Bioinformatics. Through my projects, I gained expertise in graph algorithms, knowledge graph construction, and machine learning, and a deep understanding of the challenges and opportunities in applying knowledge graphs to diverse domains.

Semantic PageRank #

Enhancing Web Page Ranking with Knowledge Graphs Master’s Thesis - M.Tech (IIT Jodhpur)

  • Developed Semantic PageRank, a novel algorithm that addresses the limitations of traditional PageRank by integrating insights from a knowledge graph to improve the accuracy and relevance of web page ranking, particularly for academic search.
  • Constructed a knowledge graph from a corpus of academic papers, capturing entities (authors, papers, topics) and their relationships (citations, co-authorship, etc.) to create a rich semantic representation of the domain.
  • Enhanced traditional PageRank by incorporating semantic similarity measures derived from the knowledge graph. Specifically, entity linking and relation extraction were employed to disambiguate query intent and identify web pages with strong semantic relevance to the user’s query.
  • Enhanced traditional PageRank by incorporating semantic similarity measures based on path-finding algorithms and node embeddings within the knowledge graph.
  • Conducted extensive experiments, comparing Semantic PageRank’s performance to traditional PageRank, demonstrating significant improvements, especially for semantically complex queries.

Recommender System Using Knowledge Graphs #

Movie Recommender System Leveraging Knowledge Graphs for Enhanced Semantic and Contextual Recommendations Minor Project - M.Tech (IIT Jodhpur)

  • Developed a knowledge graph-powered movie recommender system that leverages semantic understanding and contextual information to deliver personalized recommendations.
  • Constructed a comprehensive knowledge graph integrating movie data (genres, actors, directors, etc.), user profiles, and interaction history (ratings, reviews, etc.) to capture rich semantic relationships and context.
  • Implemented hybrid recommendation algorithms combining knowledge graph-based techniques (e.g., personalized PageRank, path ranking algorithms) with traditional collaborative filtering (matrix factorization) and content-based filtering.
  • Employed semantic similarity measures based on node embeddings and graph features to infer user preferences and provide more nuanced recommendations.
  • Evaluated the system’s performance through extensive testing, demonstrating the added value of incorporating knowledge graph-based techniques for improved recommendation accuracy and user satisfaction.