| Summary: | 碩士 === 國立政治大學 === 會計研究所 === 101 === Analyzing financial information and textual information in footnote is able to help users categorize unstructured text information so that make it easy to read. This research randomly samples 40 companies from U.S. listed semiconductor firms in 2012. We use TFIDF to analyze textual information in footnote and combine it with 5 dimensions from financial information, using SAX Praser to abstract financial information and textual information in footnote. Finally apply it by JAVA under the eclispse platform in practical. Via this application, we try helping investors reduce the investing risks, and make more profits. According to this research, the accuracy of correct categorization of textual information in footnote reaches 86%, and generic link is able to effectively establish the relationship between financial information and textual information in footnote. Thus, investors can guickly find out related textual information in footnote from financial subjects they try to understand, so that investors can lower their risks and make faster and correcter decisions.
|
|---|
