Abstract: Current Web service clustering methods mainly focus on the semantic information of service function description,but lack of consideration of service collaboration.To further improve service clustering quality,a service clustering method integrating functional semantics and service collaboration was proposed.The service function feature vectors with high-quality function semantic expression were generated by fusing tag vectors and service description text vectors.Then the service collaboration network was established.A weighted GraphSAGE model was designed to aggregate the features of neighborhood service nodes on the service collaboration network.Service representation vector was obtained by integrating the collaboration relationship into service function feature vector.Finally,the K-means + + algorithm was used for clustering.Experiments showed that the service function vector generated by this method significantly improved the clustering quality by comparing with the traditional models.Moreover,the integration of collaborative features further improved the effect of service clustering.

Key words: Web services, service cluster, functional semantics, service collaboration

摘要： 当前Web服务聚类方法主要关注服务功能描述语义信息,缺乏对服务之间协作关系的考量。为进一步提高服务聚类质量,提出一种融合功能语义与服务协作关系的服务聚类方法。首先通过融合标签向量和服务描述文本向量生成高质量功能语义表达的服务功能特征向量;然后建立服务协作网络,构建加权GraphSAGE模型,在服务协作网络上实现邻域服务结点的特征聚合,获得融入协作关系的服务表征向量;最后采用K-means++算法进行聚类。实验表明,相比其他常用模型,所提方法所生成的服务功能向量提高了聚类质量,协作特征的融入进一步提升了聚类效果。

关键词: Web服务, 服务聚类, 功能语义, 服务协作