中文摘要：

针对传统的金融预测系统仅仅依靠股票价格和市场指数等定量数据而不能很好地满足实时性和高准确性的问题，提出一种基于加权关联规则和文本挖掘的新闻传播 Agent 实现方法。首先，利用中文知识与信息处理系统将每个新闻标题分离得到每个中文单词；然后，利用加权关联规则算法检测频繁出现在同一条新闻标题中的多个术语，并提取名词、动词和复合语；最后，根据新闻供给市场第一个交易日股票交易金融价格指数为提取的关键字分配权重，并根据新闻标题的权重值判断其对股票价格的影响程度。新闻标题特征数据库上的实验验证了该方法在金融新闻标题的实时信息发布应用中的可行性，实验结果表明，相比其他几种预测方法，该方法取得了更高的预测准确率和召回率。

英文摘要：

Traditional financial prediction systems cannot well satisfy both real-time property and high accuracy because they rely on quantitative data of stock prices and market indexes only.For which,we propose the weighted association rules and text mining-based Agent realisation of news spreading.First,it employs Chinese knowledge and information processing system to divide every news headline into single Chinese characters.Then,it uses WAR algorithm to detect multiple terminologies frequently appearing in same news headlines,and extracts noun,verb and complex languages as well.Finally,it assigns weights to the extracted keywords according to the first day’s financial price index of stock transactions in news supplying market,and estimates the influence degree of weighted values of news headlines on stock prices. The effectiveness of the proposed method in application of real-time information delivery of financial news headlines has been verified by the experiments on news headlines characteristic database.Experimental results show that the proposed method achieves higher accuracy rate and recall rate in prediction than several other prediction methods.