Information Extraction in FinanceThis book provides a full overview of past and present algorithms forinformation extraction, which have unfortunately been scattered among different research institutes. It thus gives a complete idea of the researchactivities in the field. It includes basic algorithms descriptions, which give the non-expert reader an idea of the most common techniques in this field, and references. Primarily intended for financial organizations and business analysts, this book provides an introduction to the algorithmic solutions to automatically extract the desired information from Internet news and obtain it in a well structured form. It places emphasis on the principles of the method rather than its numerical implementation, omitting the mathematical details that might otherwise obscure the text and trying to focus on the advantages and on the problems of each method. The authors also include many practical examples with complete references, algorithms for similar problems, which may be useful in the financial field, and basic techniques applied in other informationextraction fields which may be imported to the analysis of financial news. |
Contents
1 Financial information and investment decisions | 1 |
2 Financial tools | 11 |
3 Traditional approaches on qualitative information | 27 |
4 Natural language processing and information extraction | 37 |
5 LOLITA and IEexpert systems | 75 |
6 Conclusions | 115 |
Common terms and phrases
able according acquire acquisition action algorithms analysis annotations announce application approach architecture automatically called collection competition complex component concept considered consists contain decisions defined developed documents domain entity evaluation event example expert system extraction fact Figure financial operators further grammar human identified IE system important inference information extraction information retrieval input Italy kind knowledge lexical linked LOLITA matching meaning measures module neural networks nodes object organisation output parse particular patterns performed person position possible precision produce provides qualitative quantitative recall recognised reference relevant relevant information represent retrieval Reuters rules semantic network semantic structure sentence shares similar slot specific statistical summary takeover takes task techniques template topic tree University usually