Toward Knowledge as a Service (KaaS): Predicting Popularity of Knowledge Services Leveraging Graph Neural Networks

• Published in: IEEE transactions on services computing 2023-01, Vol.16 (1), p.642-655
• Main Authors: Lin, Haozhe, Fan, Yushun, Zhang, Jia, Bai, Bing, Xu, Zhenghua, Lukasiewicz, Thomas
• Format: Article
• Language: English
• Subjects: Encyclopedias; Evolution; graph convolutional networks; Graph neural networks; Hypertext systems; Internet; Knowledge; Knowledge as a service; Knowledge engineering; Neural networks; Online services; popularity prediction; Recurrent neural networks; Spatial data; Spatial dependencies; spatiotemporal prediction; Training
• ISSN: 1939-1374; 2372-0204
• DOI: 10.1109/TSC.2022.3145019

Knowledge services are becoming a rising star in the family of XaaS (Everything as a Service). In recent years, people are more willing to search for answers and share their knowledge directly over the Internet, which drives the knowledge service ecosystem prosperous and quickly evolve. In this article, we aim to predict the popularity of knowledge services, which will benefit the downstream industries that provide Knowledge as a Service (KaaS). Toward such a task, the spatial interactions (e.g., hyperlinks in Wikipedia ) and temporal observations (e.g., page views) provide crucial information. However, it is difficult to utilize this information due to: (i) complicated and different usage observations, (ii) intricate and evolutionary spatial interactions, and (iii) small world trait of the network. To tackle such issues, we propose Evolutionary Graph Convolutional Recurrent Neural Networks (E-GCRNNs) to simultaneously model both temporal and spatial dependencies of knowledge services from their evolving networks. Specifically, an elementary unit (called E-GCGRU) is designed to dynamically perceive the evolutionary spatial dependencies, aggregate spatial information of knowledge services, and model the temporal patterns by considering the records of one sequence and its neighbors simultaneously. Additionally, a localized mini-batch training scheme is developed, which allows the E-GCRNNs to work on large-scale knowledge services networks and reduce the prediction bias caused by the small world trait. Extensive experiments on real-world datasets have demonstrated that the proposed E-GCRNNs outperform baselines in terms of prediction accuracy, especially with the prediction range being longer, while remaining computationally efficient.