DOCUMENT SUGGESTION DURING CONVERSATION USING KEYWORD EXTRACTION AND CLUSTERING

Abstract

This paper states the problem statement of keyword extraction from conversations, based on which document gets
suggested. With the help of idea described in here user can access the relevant document based on keyword from each
short conversation fragment, which can be recommended to users. Sometimes, even a short fragment includes different
types of words, which are potentially related to different topics. However, errors can be introducing into system or
conversation by using automatic speech recognition system. To extract keyword from the output of ASR technique we
use one algorithm. It is based on the sub modular functions which gives range of different words and reduces the noise.
Then use a method to obtain a multiple topic from this keyword set only for to access the one relevant recommended
document.