Please use this identifier to cite or link to this item:
|Title:||Using Multi-Objective Optimization for Word Sense Induction and Disambiguation|
|Keywords:||Computer Science & Engineering|
|Abstract:||Word Sense Induction is a task of automatically finding new senses to words. Hierarchi- cal Dirichlet Process, a topic modelling concept performs this clustering Task by inducing different senses to clusters. An existing implementation of Simulated Annealing Based Multi-objective Optimization Algorithm(AMOSA) performs Search Results Clustering and it initializes itself with random clusters as its starting solutions. It searches better solutions through Simulated Annealing mechanism. In Simulated Annealing after each iteration, new solutions obtained through mutation are compared with solutions at current iteration using multiple con icting Objective functions, compactness and separability. The motivation is to improve the results by initializing AMOSA with output of a Topic Modelling Based solution to Word Sense Induction problem, by guiding the clustering Task. Due to the availability of page ranks for documents to be clustered, two new Objective Functions Page Rank based Compactness and Page Rank based Separability are introduced to work along with the other Objective Functions of AMOSA.|
|Appears in Collections:||01. CSE|
Files in This Item:
|Using Multi-Objective Optimization for Word Sense Induction and Disambiguation.pdf||434.14 kB||Adobe PDF||View/Open Request a copy|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.