Tree-based Mining of Semantically Related Words in Tamil Biomedicine
11 Pages Posted: 27 Feb 2020
Date Written: February 27, 2020
Unsupervised mining of unstructured data is a vast field open for a number of research ideas. This paves way for identifying methods with minimum computational complexities but maximum consequential yields. One such method is to mine data purely based on Spanning Tree traversal. In this work, we concentrate on identifying semantically related and not similar entities from a wide collection of Tamil Siddha medicinal data. The proposed work converts the unstructured terms into a single graph with an adjacency matrix, dissects them into small closely knit subunits and constructs a maximum spanning tree from which word association information is mined. The breaking down into subunits involves two types of cliques, the biggest maximal cliques and vertex-oriented cliques, both giving agreeable results on tree traversal for any given context.
Keywords: Maximum Spanning Tree, Semantic Relation Mining, Graph theory, Clique Analysis, Tamil Biomedicine
Suggested Citation: Suggested Citation