Listing Corporation Annual Report Text Knowledge Discovery Based On LDA Topic Model

The emergence of Data Mining solves the problem of finding useful information from a large number of data. Traditional Data Mining algorithm is difficult to deal with for some text, WEB pages, email, and other unstructured or semi-structured data. Text Mining can search and sort, information extraction, information filtering, natural language understanding and so on, which has a very good research significance and commercial value. Annual report is one of the most easily accessible enterprise information, It’s a very important way for the stakeholders to understand the profitability, operation ability and the future development of the company. Therefore, the annual report of the listing Corporation of the text mining has become an important means for people to understand the operation of the listing Corporation.On the basis of Loughran and Mcdonald(2011) proposed financial semantic thesaurus, semantic word was finished adjustment. On this basis, the annual report of listing Corporation is constructed, which is based on the theme of LDA.The results show that: By analyzing the relationship between the semantic key words and the related financial indicators in annual reports, and the increase of uncertainty words and Future Ltd’s management has a negative correlation, the results were verified by multi sample.