A Study On The Space-time Elements Of Chinese Opinion

Along with Web2.0 technology revolution, the Internet has an explosive growth in opinioned texts, which contain a large number of the user's emotions. Opinion mining techniques emerge and focus on the opinion model which is composed of topic, holder, claim and sentiment for a long time. It has achieved fruitful results.Based on the new applications and requirements, in this paper it studies on the space-time elements of Chinese opinion. It first solves the problem of lacking comprehensive semantic knowledge in Opinion Mining. It utilizes Wikipedia to construct Concept Dictionary and Wikipedia Dictionary whose entries are explained by Wikipedia categories. While Concept Dictionary is for identifying the meaning of the topic, Wikipedia Dictionary is used to compute semantic relatedness of any two texts. The results show the effectiveness of the algorithm and successful application to the identification of opinion topic.In this paper, a fine-grained approach is proposed for Chinese sentence opinion analysis and reveals the characteristics of sentimental orientation from the structure and linguistic phenomenon of Chinese sentences. After constructing a sentimental dictionary automatically using HowNet, a system is implemented for the first Chinese Opinion Analysis Evaluation (COAE 2008). The official evaluation results show its superiority.In this paper, it proposes a concept of Opinion Important Factor which is composed of Time Important Factor and Source Important Factor. It analyzes formulae of Time Important Factor in the different applications and divides Source Important Factor into two parts, the influence of source and the relatedness of source and domain.Finally, this paper backs to the applications of space-time elements of opinion. It explores mining methods and evaluation methods for two typical applications whose topics are "mining the most popular products in a certain period of time "and "mining opinion trend" respectively.