Abstract: With the number increasing of Web services, recommending and selecting the personalized Web services for consumers has become one of the most important challenges in the field of service computing. This paper studies the approach for personalized Web services recommendation and proposes a model-based and memory-based quality of service (QoS) prediction approach for Web services. In the proposed approach, the consumers’ expectation, rating and the QoS information are quantified. And based on the objective QoS data and the subjective rating, the quality of services is predicted by clustering, mapping and aggregation. The result is a list of recommended services for selection. In addition, this paper designs a prediction framework, and realizes the information collecting and processing and personalized Web services recommendation through the framework. The experimental results demonstrate that compared with most other service recommendation approaches, the proposed approach increases the accuracy of recommendation results.

Key words: Web service, clustering, QoS prediction, personalization, recommendation

摘要： 随着网络上Web服务的不断增加，Web服务的个性化选择和推荐成为服务计算领域最重要的挑战之一。对个性化Web服务推荐方法进行了研究，提出了基于模型和基于内存混合的Web服务推荐方法。该方法基于客观连续的服务质量（quality of service，QoS）数据和主观离散的评价数据，采用聚类、映射、聚合等算法预测服务的质量，并对用户的期望、评分和服务的QoS信息进行了量化描述。此外，设计了Web服务推荐框架，实现了信息的采集与处理、Web服务的个性化推荐。实验结果表明，与主流的推荐算法相比，所提方法在多种评分误差的评价指标上都取得了更好的结果。

关键词: Web服务, 聚类, 服务质量预测, 个性化, 推荐