针对VSM不能揭示文档中特征词间的潜在语义关系，相似度计算准确性较低的问题，结合本体模型的结构特点，从语义重合度、语义距离以及本体结构等因素综合考虑概念间的相似度计算，提出了一种基于领域本体的文档向量空间模型。该模型通过构建概念间的语义相似度矩阵对特征词权值进行调整，建立包含语义关系的标准(学生)答案的向量空间模型，并用“VSM模型+余弦值”算法评估学生答案和标准答案的相似度。实验表明，与传统方法相比，该方法提高了评测效果及准确率。%In view of the problems that VSM couldn't reveal the latent semantic relations between the key words in a document,and have a low accuracy in document similarity calculation,combined the structural feature of ontology model,considering the concept similarity from the semantic contact ratio,semantic distance and the ontology structure,a document vector space model based on domain ontology is proposed. The model adjusts the weight of feature words by building concept semantic similarity matrix,constructing the standard answer and the student answer VSM contains semantic relation, and making use of the “VSM+cosine” algorithm to assess similarity of the student answer and the standard answer. Compared with the traditional method,experiments show that this method improves the scoring results and accuracy.
展开▼
