面向网络股评观点的垂直搜索引擎设计与实现

摘要

The stock market is a barometer of the national macroeconomic environment,the views of users on the stock market to a large extent reflect the views of the stock market,but also affect the stock market.Therefore,through the emotional network text polarity analysis technology and search engine technology to mining stock network view is one of the hot topic of current financial information and cross discipline.This paper designed a fusion vertical search full-text search and opinion mining for network stock review engine system structure,put forward a network oriented theme stock data acquisition point harvest algorithm and crawler structure,and established a multi granularity unsupervised emotion network stock review fuzzy calculation polarity classification method,to achieve the analysis of online stock comment.The test results show that the classification accuracy,the efficiency of data acquisition and the response time of search engine can meet the requirements of practical application.%股票市场是国家宏观经济环境的晴雨表.网民对股市评论观点在很大程度上反映了股市行情,也影响着股市涨跌.因此,通过网络文本情感极性分析技术和搜索引擎技术来挖掘网络股评观点是当前信息和金融学科的交叉研究课题热点之一.设计了一种融合全文搜索和观点挖掘的面向网络股评的垂直搜索引擎系统结构,提出了一种面向网络股评主题数据采集的定点收割算法和爬虫结构,并建立了一种网络股评的多粒度模糊计算的无监督情感极性分类方法,能实现股评观点的在线分析.通过对实现的垂直搜索引擎的测试表明,其在情感极性分类精度、爬虫数据采集效率、搜索引擎响应时间等性能指标上都能满足实际应用要求.