Multiple Graph-Based Approach to Recommend Domain-Dependent Documents
Nikita Nikita, A. Maheshwari, Hardik Tulsiani, Hiren Dhadhal, Utsav Chordia, Dipti P. Rana, Rupa G. Mehta · 2024
For finding relevant information in domain-dependent documents, the recommender system plays a significant role. The huge availability of data provides opportunities as well as challenges for extracting relevant information. Natural language writing, domain-specific vocabulary and expertise pose challenges in extracting information from the documents. In this work, judgments from the Supreme Court of India are considered for the case study of the recommender system in domain-dependent documents. The need to access similar documents pertains to the fact that the judiciary, following the common law system, gives equal importance to laws as well as verdicts in related cases. To access similar relevant information from the pool of documents, a recommender system is needed. This work proposes a recommender system that works on the principle of similarity with respect to keywords from the user query and laws cited in the documents obtained from multiple graphs accessed in a hierarchy. The results obtained are superior to the existing state-of-the-art techniques.