A Hybrid Approach to Web Service Recommendation Based on QoS-Aware Rating and Ranking

Mingming Chen,Yutao Ma

DOI: https://doi.org/10.48550/arXiv.1501.04298

2015-01-18

Abstract:As the number of Web services with the same or similar functions increases steadily on the Internet, nowadays more and more service consumers pay great attention to the non-functional properties of Web services, also known as quality of service (QoS), when finding and selecting appropriate Web services. For most of the QoS-aware Web service recommendation systems, the list of recommended Web services is generally obtained based on a rating-oriented prediction approach, aiming at predicting the potential ratings that an active user may assign to the unrated services as accurately as possible. However, in some application scenarios, high accuracy of rating prediction may not necessarily lead to a satisfactory recommendation result. In this paper, we propose a ranking-oriented hybrid approach by combining the item-based collaborative filtering and latent factor models to address the problem of Web services ranking. In particular, the similarity between two Web services is measured in terms of the correlation coefficient between their rankings instead of between the traditional QoS ratings. Besides, we also improve the measure NDCG (Normalized Discounted Cumulative Gain) for evaluating the accuracy of the top K recommendations returned in ranked order. Comprehensive experiments on the QoS data set composed of real-world Web services are conducted to test our approach, and the experimental results demonstrate that our approach outperforms other competing approaches.

Information Retrieval,Software Engineering

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is that in Web service recommendation, traditional methods based on rating prediction may not be able to provide satisfactory recommendation results. Especially in some application scenarios, users are more concerned about the ranking of Web services rather than specific QoS values. Therefore, the author proposes a hybrid method that combines item - based collaborative filtering and latent factor model, aiming to achieve personalized Web service recommendation through QoS ranking prediction.

Specifically, the paper proposes the following key points:

1. **Ranking - oriented learning techniques**: Considering the application requirements of Web service ranking, especially in terms of response time, the paper introduces ranking - oriented learning techniques into the traditional item - based collaborative filtering method. The purpose is to find the most similar Web services according to ranking similarity rather than rating similarity.

2. **Improved evaluation metrics**: In order to verify the effectiveness of the proposed method, the paper also improves the NDCG (Normalized Discounted Cumulative Gain) evaluation metric, which is used to measure the accuracy of the top K recommended items returned.

3. **Ranking - oriented hybrid recommendation method**: The paper proposes a new ranking - oriented hybrid method. This method integrates baseline estimation, item - based collaborative filtering recommendation algorithm and latent factor model to form a unified model in order to achieve better prediction performance.

4. **Experimental verification**: Through experiments on well - known datasets, it is proved that the proposed method is superior to other competing methods in evaluation metrics.

In summary, the main problem that this paper mainly solves is how to improve the accuracy of the recommendation system and user satisfaction in Web service recommendation by combining rating prediction and ranking prediction methods.