Towards efficient multi-feature queries in heterogeneous environments

Abstract

Applications like multimedia databases or enterprisewide information management systems have to meet the challenge of efficiently retrieving best matching objects from vast collections of data. We present a new algorithm Stream-Combine for processing multi-feature queries on heterogeneous data sources. Stream-Combine is seljadapting to different data distributions and to the specific kind of the combining function. Furthermore we present a new retrieval strategy that will essentially speed up the output of relevant objects.

UPRE: User Preference Based Search System

“…From the random access only approaches U. Güntzer et al [8] proposed stream-combine algorithm as the first one. Much better and supported by good formal model was No Random Access algorithm presented by Fagin et al [6].…”

“…Their implementations are based on several theoretical researches [1,5,7,8] -so far we have programmed 7 various heuristics.…”

Top-k Query Evaluation with Probabilistic Guarantees

“…Note that, unlike TA-random, the algorithm requires remembering all nondiscarded candidates in memory. We will come back to these implementation details in Section 4; neither the original work [19,23] nor the follow-up research [1,9,29] discuss concrete bookkeeping data structures despite their strong performance impact.…”

Efficient Aggregation of Ranked Inputs

“…On the other hand, in this paper, we focus on top-k queries in the case where the atomic scores in each source can be accessed only in sorted order; i.e., it is not possible to know the score of an object in source S i , before all objects better than x in S i have been seen there. This case has received increasing interest [8,12,9,10] for several reasons. First, in many applications, random accesses to scores are impossible [6].…”

“…Nevertheless, as shown in [8,12,9,10], NRA algorithms can have significant performance differences in terms of (i) accesses, (ii) computational cost, and (iii) memory requirements. The number of accesses is a significant cost factor, especially for middleware applications which charge by the amount of information transferred from the various (distributed) sources.…”