融合服务聚类与协作相似度的服务替换方法

当前服务替换方法多关注于替换服务与失效服务在功能和质量层面的匹配，缺乏对两者协作关系的考量。针对上述问题，提出一种融合服务聚类与协作关系的服务替换方法。首先，利用服务之间的组合关系构建服务协作图谱，基于node2vec进行序列采样并生成协作向量，进而计算服务之间的协作相似度。然后，通过引入服务簇缩减替换时的服务查找空间，从功能和质量层面快速构建候选替换服务集合。最后，在候选替换服务集合中综合服务质量评分和协作相似度实现可替换服务的择优推荐。实验证明该方法可以有效提高服务替换的效率与合理性。

Service substitution method combining service clustering and collaboration similarity

The current service substitution methods mainly focus on the function and quality matching between the substitutive service and invalid one. These methods rarely considered service collaboration. To solve the above problem, this paper proposed a service substitution method integrating service clustering and collaboration. Firstly, this method constructed service collaboration map according to service composition relationships. It used node2vec to sample node sequences and generate collaboration vectors from the service collaboration map. This method computed service collaboration similarity based on these collaboration vectors. Then, this paper introduced service clustering to reduce the search space during service substitution. It generated the candidate substitutive service set from the functional and quality levels. Finally, this method integrated the service quality score and collaboration similarity to achieve the recommendation of substitutive services. Experiments show that the proposed method can effectively improve the efficiency and rationality of service substitution.