Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Text Processing

Heterogeneous Information Based Financial Event Detection

Author LiuLiBo
Tutor ChenQingCai
School Harbin Institute of Technology
Course Computer Science and Technology
Keywords financial ontology event detection heterogeneous information time alignment text mining
CLC TP391.1
Type Master's thesis
Year 2010
Downloads 43
Quotes 0
Download Dissertation

With the continuous development of computer science and technology, Internet has become an indispensable information source in our daily life. Network information is mainly qualitative reference for users with their own characteristics. In the financial field, particularly, structured data have always been the main target to process. These quantitative data are very suitable for computer to analysis. However, due to the characteristics of the financial field, the accurate analysis of unstructured information, especially which related to the real-time tracking and detection implied in the important financial events, will also have a crucial impact on the structured data forecast. Then base on the financial ontology construction, we will research on the financial event detection methods with heterogeneous information. The research content mainly consists of the following aspects:1) Heterogeneous information preprocessing, elaborate the construction of financial ontology, the process of stock data and financial webpage information preprocessing.2) Heterogeneous information association analysis, process the time information of the heterogeneous information, then use time alignment method to associate them, take the stock data to guide the process of financial webpage.3) Financial event detection, base on the associated heterogeneous information, take the text classification and clustering methods in use to analyze the financial webpage, then get the financial events that influence the trends of stock data.In our thesis, we take the closing price of the stocks from year 2005 to year 2008 as the structured data, and Sina financial webpages between year 2005 and year 2008 as the unstructured data in use to do our research. Protégétool model and OWL language are used to construct the financial domain ontology.Finally, the evaluation standards of the financial news webpage classification, clustering and financial event detection have been described in details. Base on large amount of contrast experiments, validate the methods used in our thesis have a good feasibility, which can be used in the application such as analysis the influence that financial events acting on the trends of financial product price.

Related Dissertations
More Dissertations