摘要
定义了骨架语片的概念。用互信息量作为衡量两个词语间相关程度的参考值,借助依存关系、基本语法将满足相关度阈值的两个词组合成骨架语片。用骨架语片做特征项,用空间向量模型表示文本语义,用语片的出现频度做语片权重,用余弦法计算文本间语义相似度。应用于试卷主观题自动评分,实验证实这种方法结果具有令人满意的正确度。
Defining the concept of skeleton semantic clip in the paper. Comparing relevancy between two words using mutual information. Structuring two words accord with some value of mutual information through semantic dependence and basic syntax. Computing the semantic similarity of sentences by the method of cosine, eigenvalue come from the skeleton semantic clip, and the semantics of sentence expressed the vector space model. The application of the method is the auto gradeing system of subjective test questions in examination. The method is validated by some use case. The result is satisfying.
出处
《计算机与数字工程》
2007年第10期24-25,34,共3页
Computer & Digital Engineering
关键词
空间向量模型
相关度
骨架语片
互信息
相似度
vector space model,relevancy,skeleton semantic clip,mutual information,semantic similarity of sentences
