Cluster-Based Subscription Matching for Geo-Textual Data Streams

Geo-textual data that contain spatial, textual, and temporal information are being generated at a very high rate. These geo-textual data cover a wide range of topics. Users may be interested in receiving local popular topics from geo-textual messages. We study the cluster-based subscription matching (CSM) problem. Given a stream of geo-textual messages, we maintain up-to-date clustering results based on a threshold-based online clustering algorithm. Based on the clustering result, we feed subscribers with their preferred geo-textual message clusters according to their specified keywords and location. Moreover, we summarize each cluster by selecting a set of representative messages. The CSM problem considers spatial proximity, textual relevance, and message freshness during the clustering, cluster feeding, and summarization processes. To solve the CSM problem, we propose a novel solution to cluster, feed, and summarize a stream of geo-textual messages efficiently. We evaluate the efficiency of our solution on two real-world datasets and the experimental results demonstrate that our solution is capable of high efficiency compared with baselines.