Efficient fuzzy ranking queries in uncertain databases

  • Authors:

  • Chuanfei Xu;Yanqiu Wang;Yu Gu;Shukuan Lin;Ge Yu

  • Affiliations:

  • Institute of Computer Systems, College of Information Science and Engineering, Northeastern University, Shenyang, China 110819;College of Information Science and Engineering, Northeastern University, Shenyang, China 110819;College of Information Science and Engineering, Northeastern University, Shenyang, China 110819;College of Information Science and Engineering, Northeastern University, Shenyang, China 110819;College of Information Science and Engineering, Northeastern University, Shenyang, China 110819

  • Venue:
  • Applied Intelligence
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Recently, uncertain data have received dramatic attention along with technical advances on geographical tracking, sensor network and RFID etc. Also, ranking queries over uncertain data has become a research focus of uncertain data management. With dramatically growing applications of fuzzy set theory, lots of queries involving fuzzy conditions appear nowadays. These fuzzy conditions are widely applied for querying over uncertain data. For instance, in the weather monitoring system, weather data are inherent uncertainty due to some measurement errors. Weather data depicting heavy rain are desired, where "heavy" is ambiguous in the fuzzy query. However, fuzzy queries cannot ensure returning expected results from uncertain databases.In this paper, we study a novel kind of ranking queries, Fuzzy Ranking queries (FRanking queries) which extend the traditional notion of ranking queries. FRanking queries are able to handle fuzzy queries submitted by users and return k results which are the most likely to satisfy fuzzy queries in uncertain databases. Due to fuzzy query conditions, the ranks of tuples cannot be evaluated by existing ranking functions. We propose Fuzzy Ranking Function to calculate tuples' ranks in uncertain databases for both attribute-level and tuple-level uncertainty models. Our ranking function take both the uncertainty and fuzzy semantics into account. FRanking queries are formally defined based on Fuzzy Ranking Function. In the processing of answering FRanking queries, we present a pruning method which safely prunes unnecessary tuples to reduce the search space. To further improve the efficiency, we design an efficient algorithm, namely Incremental Membership Algorithm (IMA) which efficiently answers FRanking queries by evaluating the ranks of incremental tuples under each threshold for the fuzzy set. We demonstrate the effectiveness and efficiency of our methods through the theoretical analysis and experiments with synthetic and real datasets.