Sentiment Lexicon Generation Based On Stock Price

In recent years, with the development of social medias, more and more people begin to engage themselves to the Internet. Internet has attracted them to write down their comments about everything on their blogs. This information is very important to human beings. Companies want to use this information to bootstrap their abilities to compete with their enemies. The majority of people seek to find out the trend of our society. The information on the Internet is increased with a very quick speed. However, the ability to read and analyze this info is quite low comparing to the storage of the information. Thus, we need an algorithm, which could automatically extract useful things from information.The main work of this paper is summarized as follows.First, we incorporate the metadata of the article to the PLSA model, which leads to a novel algorithm called Trend-PLSA model. This model could automatically generate useful information from the Internet source and the stock price. We construct a financial lexicon.Second, we rank theses words appearing in the lexicon to give people a better understanding of financial lexicons.Third, we develop a system, which is based on lucene to fulfill all the job described above.The main technologies we used here is natural language processing and stock analysis. Those two techs are guarantee for our experiment.