面向新闻事件的故事树构建方法

为解决当前方法在追踪新闻事件时容易遗漏相关新闻以及所构建的事件发展脉络不连贯、整体性较差的问题,提出一种面向新闻事件的故事树构建方法。对相关新闻进行动态追踪时根据新闻关键词的词频-逆类别频率(TF-ICF)和热度对事件关键词进行反馈更新,采用有主干和分支的故事树结构展现事件发展脉络,利用新闻热词定位事件发展的关键节点,生成故事的主干结构。实验结果表明,该方法所构建的事件脉络在可读性、完整性上表现良好。

Story tree construction approach for news events

To solve the problem that the current methods are easy to miss relevant news and that the event venation constructed using current methods is of inconsistency and poor overallity when tracking news events,a story tree construction approach for news events was proposed.When dynamically tracking related news,the event keywords were updated according to the term frequency-inverse category frequency(TF-ICF)and heat of the news keywords,and event development venation was organized into a story tree with trunk and branch.The news hot words were used to locate the key nodes of the event development,and the backbone structure of the story was generated.Experimental results show that the event venation constructed using this approach performs well in readability and integrity.