摘要
Web服务系统中大量无使用记录的服务和不断发布的新创建服务被称为冷启动服务。为了帮助服务组合开发者了解冷启动服务的特性,提高冷启动服务的关注率与使用率,从而增强服务系统的元素多样性和系统鲁棒性,该文提出了一种冷启动服务协作关系挖掘与预测方法。该方法利用服务描述重构和功能主题分析为每个服务建立功能属性向量。对非冷启动服务,基于其历史协作关系和功能属性向量为其建立协作属性向量。通过对冷启动服务功能属性向量与非冷启动服务协作属性向量进行相似性比较,实现冷启动服务组合协作关系预测。真实数据集上的实验结果证明该方法在预测效果上显著强于当前最优方法。
Services that have never been used and newly released services in Web service systems are called cold start services.A cold start service collaboration relationship mining and predicting method is developed to help service composition developers identify the characteristics of cold start services,increase attention to and usage of cold start services,and enhance the element diversity and robustness of service systems. The method first establishes a functional vector for each service using service description reconstruction and a functional topic analysis.Next,the method builds a collaborative vector for each non-cold start service based on its historical collaboration record and functional vector.Finally,the method compares the functional vectors of cold start services with the collaborative vectors of non-cold start services to predict the collaboration relationships for cold start services. Tests on real-world data show that this method more effectively predicts the relationships than state-of-the-art methods.
作者
郝予实
范玉顺
HAO Yushi;FAN Yushun(Department of Automation.Tsinghua University,Beijing 100084.China)
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2019年第11期917-924,共8页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金资助项目(61673230)
国家高技术船舶科研项目(17GC26102.01)
关键词
服务系统
冷启动服务
服务组合
协作关系
主题模型
service systems
cold start services
service composition
collaboration relationships
topic models