New M.E. Thesis Submitted from CSE Student

EFFICIENCY IMPROVEMENT IN RMN BY USING MONTE CARLO METHOD bY Mamta Singla,cse

Abstract
Various probabilistic models such as Hidden Markov Model (HMM), Maximum Entropy Markov Models (MEMMs), Conditional Markov Models (CMMs) are the prominent models for information extraction. In research, latest probabilistic model for information extraction is Relational Markov Model (RMM), which can be seen as a generalization of undirected graphical models such as Conditional Random Field (CRFs) that allow for collective classification of a set of arbitrarily related entities by integrating information from features of individual entities as well as the relation between them. All these models consists of the Markov Property i.e. decisions about the state at a particular position in the sequence can depend only on a small local structure,which is the limitation in many tasks since natural language processing is contain a great deal of Non-local structure. A general method for solving this problem is to relax the requirement of exact inference, substituting approximate inference algorithms instead, thereby permitting tractable inference in models with non-local structure. One such algorithm is Monte Carlo algorithm. This dissertation is to improve the performance of RMN for Collective Information Extraction because local template- relational Markov Network accuracy is still less significant than CRF’s. Monte Carlo Method can be used for improving the local template – RMN’s than CRF’s for Collective Information Extraction.


Leisure Readings :