【摘要】 分析了句子中确定信息的特征,将其归结为不同层次的模式.分别对模式进行了形式化,对相同层次的模式赋予不同的优先级.利用这些确定信息,提出一个不经过词性消歧的直接语法分析算法:首先分析确定的信息,减少搜索空间,获得更多确定信息,并不断利用已有的信息进行深入的分析,最终导出正确的分析结果.实验显示:该算法精确率和召回率分别达到92.90%和91.47%.更多还原

【Abstract】 The determinate information in Chinese sentences was analyzed and was induced to different level models.The models were formulized and different priorities were set for same level models.Using the determinate information,a new syntax parsing algorithm that needed no POS disambiguation was proposed.The algorithm parsed the determinate information,and reduced search space to get more determinate information.And by using new determinate information conducted deeper parsing,the right parsing result was finally obtained.Experiment shows that the precision and recall rate of the proposed method can achieve 92.9% and 91.47 respectively.更多还原