Scalable top-k spatio-temporal term querying

Abstract

With the rapidly increasing deployment of Internetconnected, location-aware mobile devices, very large and increasing amounts of geo-tagged and timestamped user-generated content, such as microblog posts, are being generated. We present indexing, update, and query processing techniques that are capable of providing the top-k terms seen in posts in a userspecified spatio-temporal range. The techniques enable interactive response times in the millisecond range in a realistic setting where the arrival rate of posts exceeds today's average tweet arrival rate by a factor of 4-10. The techniques adaptively maintain the most frequent items at various spatial and temporal granularities. They extend existing frequent item counting techniques to maintain exact counts rather than approximations. An extensive empirical study with a large collection of geo-tagged tweets shows that the proposed techniques enable online aggregation and query processing at scale in realistic settings.

On main-memory flushing in microblogs data management systems

“…Due to its popularity and high application needs, managing real-time microblogs has attracted several research efforts in industry and academia. However, the main focus was on either indexing (e.g., [28,29]), querying (e.g., keyword search [5,6,16,28] or location-based search [4,19,24]), analysis (e.g., event and trend detection [11,22], news and topic extraction [14,23], or semantic and sentiment analysis [2]), or query lan guages [18,21]. In all this work, it is assumed that queries are all answered fr om in-memory contents.…”

“…In that case, a query COining to any of the nine keywords kwl to kw9 will be fully answered from in-memory contents, which significantly increases the system memory hit ratio. kFlushing enables existing algorithms for top-k microblog search queries (e.g., [5,6,16,24,19,28]) to reach to their full potential and significantly increasing their memory hit ratio. different terminologies, e.g., buffer management in database management systems (DBMSs) [9], anti-caching in main memory databases [8,15,30], and load shedding in data stream management systems (DSMSs) [1, 12,13].…”

“…Such important usage of microblogs has motivated researchers to spend maj or efforts to efficiently support search queries for large numbers of microblogs. Search queries on microblogs include keyword search queries "Find microblogs that contain certain key word(s)" [5,6,16], location search queries "Find microblogs that are posted within a certain location" [19,24], and user timeline search queries "Find microblogs that are posted by a certain user" [28] . Due to the large number of returned results for any of these search queries, all proposed techniques agree This research is capitally supported by NSF grants I1 S-0952977, I1 S-1218168, IIS-1525953, CNS-1512877, Amazon Web Service grant, and the University of Minnesota Doctoral Dissertation Fellowship.…”

GARNET: A holistic system approach for trending queries in microblogs

“…All such applications rely on the ability to understand what people are talking about in their microblogs, and use this information as an indication of the importance of news, events, and/or people interests. As a result, numerous recent research efforts have focused on supporting frequent and trending queries on microblogs with the form: "Find top-k frequent/trending keywords within the last T time units within location L" [4], [24]. Unfortunately, such efforts have a narrow scope by: (a) supporting either frequent queries as in [24] or trending queries as in [4].…”

Selectivity estimation on streaming spatio-textual data using local correlations

“…There are many important variants in the literature with different focuses such as the spatial keyword ranking query (e.g., [25]) and collective spatial keyword search (e.g., [17]). Recently, many works have been dedicated to make sense of streaming spatio-textual objects, especially the microblogs with geo-locations, including localized event detection [1], twitter advertising [12], geocorrelated information trends detection [4], frequent spatiotemporal term queries [21], spatio-textual object filtering [5,16], etc. However, none of the existing work investigates the problem of selectivity estimation on streaming spatiotextual data.…”