以单词为网络节点,由自然语言描述中单词的同现频率确定单词间的相关度并作为边的权值,构建自然语言描述集合的加权单词网络模型。利用Newman快速算法对加权单词网络模型进行社团划分,得到单词聚类结果;根据单词聚类结果与服务之间的映射关系实现服务聚类。实验结果与手工分类结果的对比表明,平均查准率达74.7%以上。
This paper described an algorithm which created WWN of service descriptions. The nodes was words, and WWN got the edge-weight through the word's co-appearance. Newman fast algorithm could detect community structure in WWN and return the clusters of words. Using the relationship between words and services, achieved service-clustering. At last, an experiment on a collection of 1 007 Web services shows the high precision of 74.7%