Databases are becoming increasingly more complex over the years, as entities can easily be «tagged» with different types of auxiliary information, such as keywords and spatial locations. For example, webpages contain keywords and they may also be associated to locations, photographs in photo-sharing services such as Flickr are assigned descriptive tags and spatial locations, persons in social networks have profile entries (keywords) and addresses. This multi-source enrichment of objects by descriptive information allows for more complex queries and analysis tasks on the data. We are interested in new search operations for spatio-textual objects. In [1], we have developed an efficient algorithm for spatio-textual similarity joins (find pairs of objects which are spatially close to each other and their textual descriptions have high overlap). Since the results of such a query could be too many, in [2], we studied the problem of ranking pairs of nearby objects based on how similar their textual descriptions are. More recently, we modeled and studied the evaluation of a spatio-textual skyline (STS) query [3]; the objective is to find objects that are relevant to a set of query keywords and do not dominate each other with respect to their distances to a set of interesting points on a map. We also studied several models for location-based keyword query suggestion and autocompletion [4,6,8]; the objective is to recommend semantically similar queries to an initial keyword search, such that the results of the suggested queries are spatially close to the query location. In [5], we studied the problem of finding objects of a certain class (e.g., houses) on a map, which have objects in their vicinity that are relevant to some given keywords (e.g., “gym”, “park”). In [7], we defined and studied the finding of object combinations that satisfy some spatial distance constraints and are they relevant to some keywords.
[1] P. Bouros, S. Ge, and N. Mamoulis, «Spatio-Textual Similarity Joins,» Proceedings of the VLDB Endowment (PVLDB), 6(1): 1-12, November 2012.
[2] S. Qi, P. Bouros, and N. Mamoulis, «Efficient Top-k Spatial Distance Joins,» Proceedings of the 13th International Symposium on Spatial and Temporal Databases (SSTD), pp. 1-18, Munich, Germany, August 2013.
[3] J. Shi, D. Wu, and N. Mamoulis, «Textually Relevant Spatial Skylines,» IEEE Transactions on Knowledge and Data Engineering (TKDE), 28(1): 224-237, January 2016.
[4] S. Qi, D. Wu, and N. Mamoulis, «Location Aware Keyword Query Suggestion Based on Document Proximity, » IEEE Transactions on Knowledge and Data Engineering (TKDE), 28(1): 82-97, January 2016.
[5] C. Doulkeridis, A. Vlachou, D. Mpestas, and N. Mamoulis, «Parallel and Distributed Processing of Spatial Preference Queries using Keywords,» Proceedings of the 20th International Conference on Extending Database Technology (EDBT), pp. 318-329, Venice, Italy, March 2017.
[6] Z. Huang and N. Mamoulis, «Location-Aware Query Recommendation for Search Engines at Scale,» Proceedings of the 15th International Symposium on Spatial and Temporal Databases (SSTD), pp. 203-220, Arlington VA, August 2017.
[7] Y. Fang, R. Cheng, G. Cong, N. Mamoulis, and Y. Li, «On Spatial Pattern Matching,» Proceedings of the 34th IEEE International Conference on Data Engineering (ICDE), Paris, France, April 2018.
[8] Z. Huang, Y. Qian, and N. Mamoulis, «Location-aware query reformulation for search engines,» GeoInformatica, 22(4): 869-893, October 2018.