Using news analysis and search trends to forecast stock volatility
| Autor: | Jyun-DaHuang, 黃俊達 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 Recently, Internet has become the main source to collect financial information. In them, online financial news carries information about the firm's qualitative information which influences stock volatility. In the past, many studies used sentiment analysis on financial news to forecast stock price volatility. However, news written in Chinese meet several language-processing issues. For example, Chinese cannot use a blank character to separate words which cause extracted features may be wrong. In addition, as a lot of people used search engine to find out financial information, search trends may be used to analyse the current economic trends. In this research, a method uses search suggestions to improve the quality of Chinese word segmentation was proposed. The proposed method aims to select suitable features from news articles and to improve news sentiment analysis on financial news in order to forecasting stock price volatility. According to experimental results, using search suggestions could improve Chinese word segmentation thus increasing forecast accuracy of news sentiment analysis. About search behaviour, search trends analysis that could be found relevance between search terms and stock prices by pattern matching. From a compared results, used the search trends analysis could get higher forecast accuracy than news sentiment analysis. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|