| Annual report is a main channel for investors and other stakeholders to obtainimportant management information from listed companies. In recent years, all kindsof textual description and explanation are more and more. The thickness of the annualreport is increasing, in which narrative content increases in a very rapid speed.Research proves that descriptive information in annual reports of listed companies hasgreat decision relevance. How to extract valuable information from the large amountof text is a problem worthy of study.The GEM as a new thing in China, provides a good financing platform for thehigh-tech enterprises, and greatly supports China’s science development andtechnology innovation. Because of characteristics of GEM itself, it brings high returnsfor investors and the same as great risk. What type of risk a Gem company disclosesin its annual report and how the way it discloses, have important influence oninvestors’ understanding of the risk.This paper will apply information extraction (IE) method based on the domainontology to the risk content of gem companies’ annual reports. Firstly, hackle the paststudies. Secondly, get risk domain knowledge by analyzing the risk text-features ofGEM companies’ annual reports and reading a lot of related materials about risk areas,and design risk domain ontology. Thirdly, use ontology building tool protégé to buildthe risk domain ontology, and store the ontology OWL files in the local disk. Fourthly,make IE rules, and extract risk types, quantitative information, and measureinformation disclosed in the annual reports of sample companies by java programming.Fifthly, evaluate system performance by calculating accuracy and recall rate, andprove that the designed information extraction method has good performance. In thispaper, we combined accounting subject with computer technology and naturallanguage processing technology, providing an efficient basic data processing tool forthe study of traditional accounting. |
PDF Full Text Request