| In recent years, internet information expand rapidly. There are more and moreremarks with personal feeling. Analysis of these texts is of great practical significance,So how to effectively extract and accurately filter information on the Internet, and howto analyze the sentiment orientation analysis of text have become research hot spots ofcurrent internet intelligent information processing. This involves natural languageprocessing, information retrieval, information extraction, machine learning and so on.In this paper, it firstly introduces the research background and theoreticalfoundation, then a complete process from the extraction of internet information to theanalysis of sentiment orientation is studied in detail, including internet informationextraction, text filtering, text sentiment orientation analysis. In information extraction, itmainly analyzes the structure of internet information, applies the DOM tree and HTMLparser to process the Web data, and introduces the template technology to solve theinformation extraction of a diversity of Web pages. In text information filtering section,it takes a step forward to improve the algorithm based on the word’s semantic similarityto calculate the semantic similarity between sentences; finally, the algorithm to calculatethe semantic similarity between documents is proposed, and then information filtering isrealized by the semantic similarity value between a document and the topic. Following，it researches the algorithm to calculate the semantic orientation of a document，whichincludes the calculation of the semantic orientation of phrase, sentence and chapter.Thus, a analysis system of sentiment orientation on Internet information is designed andimplement.Finally, a analysis system of sentiment orientation on Internet information isdesigned and implemented to verify the algorithms in the thesis. The experiment resultsshow the feasibility of information extraction, the validity of information filtering andthe accuracy of sentiment orientation analysis. |
