Approximate spatio-temporal top-k publish/subscribe

Abstract

Location-based publish/subscribe plays a significant role in mobile information disseminations. In this light, we propose and study a novel problem of processing location-based top-k subscriptions over spatio-temporal data streams. We define a new type of approximate location-based top-k subscription, Approximate Temporal Spatial-Keyword Top-k (ATSK) Subscription, that continuously feeds users with relevant spatio-temporal messages by considering textual similarity, spatial proximity, and information freshness. Different from existing location-based top-k subscriptions, Approximate Temporal Spatial-Keyword Top-k (ATSK) Subscription can automatically adjust the triggering condition by taking the triggering score of other subscriptions into account. The group filtering efficacy can be substantially improved by sacrificing the publishing result quality with a bounded guarantee. We conduct extensive experiments on two real datasets to demonstrate the performance of the developed solutions.

Pairwise Location-Aware Publish/Subscribe for Geo-Textual Data Streams

“…(2) Similarity-aware matching condition: In this category, a similarity score is considered to be the matching condition [4], [6], [14], [15], [48]. In particular, existing studies in this category have two types of matching condition: spatial matching condition and textual matching condition.…”

ETGP: Top-K Geography-Text P/S Approach without Threshold

“…Geography keyword query has been widely studied in literature [4]. These studies are based on Boolean matching.…”

“…Subsequently, geography-text P/S systems have attracted widespread attention [3]. Hence, the geography-text P/S has been used in lots of mobile network scenarios, such as social networks and geography-based product [4].…”

Optimizing Continuous kNN Queries over Large-Scale Spatial-Textual Data Streams

“…As a generic continuous query over spatial-textual data streams (CQST), CkQST can be evaluated with the evaluation framework of CQST, i.e. according to the features of queries, selecting an appropriate spatial index and a textual index, and exploiting them with some appropriate filtering strategies to process the incoming objects [1][2][3][4][5][6][7][8][9][10][11][12].…”

“…Second, for textual filtering, evaluating CkQST is essentially identifying queries whose keywords are completely contained in a given object. The inverted index is usually used to organize CQST [1,[6][7][8][9][10], but a large number of queries make the posting lists very long, and the fast-arriving objects are verified against massive posting lists multiple rounds in a short time, generating the bottleneck of textual filtering.…”