Top-k queries over web applications

Daniel Deutch, Tova Milo, Neoklis Polyzotis

Research output: Contribution to journalArticlepeer-review

Abstract

The core logic of web applications that suggest some particular service, such as online shopping, e-commerce etc., is typically captured by Business Processes (BPs). Among all the (maybe infinitely many) possible execution flows of a BP, analysts are often interested in identifying flows that are "most important", according to some weight metric. The goal of the present paper is to provide efficient algorithms for top-k query evaluation over the possible executions of Business Processes, under some given weight function. Unique difficulties in top-k analysis in this settings stem from (1) the fact that the number of possible execution flows of a given BP is typically very large, or even infinite in presence of recursion and (2) that the weights (e.g., likelihood, monetary cost, etc.) induced by actions performed during the execution (e.g., product purchase) may be inter-dependent (due to probabilistic dependencies, combined discount deals etc.). We exemplify these difficulties, and overcome them to provide efficient algorithms for query evaluation where possible. We also describe in details an application prototype that we have developed for recommending optimal navigation in an online shopping web site that is based on our model and algorithms.

Original languageEnglish
Pages (from-to)519-542
Number of pages24
JournalVLDB Journal
Volume22
Issue number4
DOIs
StatePublished - Aug 2013

Keywords

  • Business processes
  • Top-k query processing
  • Web applications

Fingerprint

Dive into the research topics of 'Top-k queries over web applications'. Together they form a unique fingerprint.

Cite this