Recommending Energy-Efficient Data Mining Services with Data as Contextual Factors
Zainab Al-Zanbouri, Chen Ding · 2019
Energy-efficiency is considered an important aspect when consuming web services. For data mining services, input dataset can have an impact on the energy consumption level as well as other QoS values of a service. For instance, a service not good at dealing with high-dimensional data may consume more energy than another service offering the same functionality. Therefore, when recommending such services, it is beneficial to consider dataset information in the recommendation process. However, in the past research on service recommender systems, usually, only the historical data on how services are consumed by different users is taken as the main source for making recommendations. Sometimes, contextual factors such as time and location (when and where the services are consumed) are also included. In this paper, we propose to consider dataset information as contextual factors in the recommendation model. We apply contextual modeling approach to incorporate them into our service recommender system. We explore different ways of representing dataset information as contextual factors and investigate their impacts on recommendation accuracy. Experimental results show that our introduced approach outperforms the baseline model by 11.7-22.6% on various evaluation metrics measuring recommendation accuracy.