Bridging Task Expressions and Search Queries

Chirag Shah, Ryen W. White

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

People engage in search episodes as they have a task or a problematic situation. Often this task is not clearly expressed by the information seeker, nor directly supported by the search system.People also list their tasks using tools such as to-do applications, and while many of these could be search tasks, there is a lack of that recognition or a possible bridge to a search system. In the work reported here, we aim to create that bridge by analyzing data on both sides. In task management, we examined 1,000 to-do tasks annotated by human assessors for their appropriateness for a search engine and created a simple process to learn that classification. In search, we analyzed millions of queries in a search engine log to understand how often queries represent tasks that people express in to-do lists. Our results show that (1) we can accurately predict which of the to-do tasks are appropriate as search queries; and (2) such tasks do indeed show up in search engines as a substantial segment. Together, these findings outline an opportunity to link explicitly expressed tasksto search queries and vice versa. This has implications for both task completion and query understanding.

Original languageEnglish (US)
Title of host publicationCHIIR 2021 - Proceedings of the 2021 Conference on Human Information Interaction and Retrieval
PublisherAssociation for Computing Machinery, Inc
Pages319-323
Number of pages5
ISBN (Electronic)9781450380553
DOIs
StatePublished - Mar 14 2021
Externally publishedYes
Event6th ACM SIGIR Conference on Human Information Interaction and Retrieval, CHIIR 2021 - Virtual, Online, Australia
Duration: Mar 14 2021Mar 19 2021

Publication series

NameCHIIR 2021 - Proceedings of the 2021 Conference on Human Information Interaction and Retrieval

Conference

Conference6th ACM SIGIR Conference on Human Information Interaction and Retrieval, CHIIR 2021
Country/TerritoryAustralia
CityVirtual, Online
Period3/14/213/19/21

All Science Journal Classification (ASJC) codes

  • Human-Computer Interaction
  • Information Systems

Keywords

  • human assessment
  • log analysis
  • queries
  • search tasks

Fingerprint

Dive into the research topics of 'Bridging Task Expressions and Search Queries'. Together they form a unique fingerprint.

Cite this