Proportionality on Spatial Data with Context

More often than not, spatial objects are associated with some context, in the form of text, descriptive tags (e.g., points interest, flickr photos), or linked entities semantic graphs Yago2, DBpedia). Hence, location-based retrieval should be extended to consider not only locations but also context objects, especially when retrieved too many and query result is overwhelming. In this article, we study problem selecting a subset result, which most representative. We argue that similar nearby proportionally represented selection. Proportionality dictates pairwise comparison all hence bears high cost. propose novel algorithms greatly reduce cost proportional object selection practice. addition, pre-processing, pruning, approximate computation techniques their combination reduces computational even further. theoretically analyze approximation quality our approaches. Extensive empirical studies on real datasets show effective efficient. A user evaluation verifies more preferable random based diversification.