Recognition of Personal Objects in Netnews Oral Reviews

摘要

Personal objects in Netnews oral reviews are important component of public opinion. Recognition of personal objects is the base of sentiment analysis for oral reviews. This paper analyzes the characteristics of personal objects in Netnews oral reviews. Based on these characteristics, we present an efficient recognition approach. The approach firstly evaluates reliability of one Chinese character as part of personal object to get reliable and discriminating recognition clues; Secondly, through an improved algorithm of frequent pattern mining, we obtain candidate objects from processing windows using the clues; Lastly, make use of a series of rules to optimize the results. The experimental results show the approach can integrally and efficiently identify personal objects in Netnews oral reviews.