Exploring task-based query expansion at the TREC-COVID track

Thomas Schoegje; Chris Kamphuis; Koen Dercksen; Djoerd Hiemstra; Toine Pieters; Arjen de Vries

doi:10.48550/arXiv.2010.12674

Exploring task-based query expansion at the TREC-COVID track

Thomas Schoegje, Chris Kamphuis, Koen Dercksen, Djoerd Hiemstra, Toine Pieters, Arjen de Vries

Research output: Working paper › Preprint › Academic

Abstract

We explore how to generate effective queries based on search tasks. Our approach has three main steps: 1) identify search tasks based on research goals, 2) manually classify search queries according to those tasks, and 3) compare three methods to improve search rankings based on the task context. The most promising approach is based on expanding the user's query terms using task terms, which slightly improved the NDCG@20 scores over a BM25 baseline. Further improvements might be gained if we can identify more specific search tasks.

Original language	English
Publisher	arXiv
Pages	1-16
DOIs	https://doi.org/10.48550/arXiv.2010.12674
Publication status	Published - 23 Oct 2020

Bibliographical note

Update version 2: Improved title Update version 3: corrected terminology hyponym -> hypernym in two instances Documents our participation to the TREC-COVID track. Contains 16 pages, 0 figures

Keywords

cs.IR

Access to Document

10.48550/arXiv.2010.12674

2010.12674v3Submitted manuscript, 121 KB

Cite this

@techreport{ec1a51928a4d45d98e64d7741bcaaba6,

title = "Exploring task-based query expansion at the TREC-COVID track",

abstract = " We explore how to generate effective queries based on search tasks. Our approach has three main steps: 1) identify search tasks based on research goals, 2) manually classify search queries according to those tasks, and 3) compare three methods to improve search rankings based on the task context. The most promising approach is based on expanding the user's query terms using task terms, which slightly improved the NDCG@20 scores over a BM25 baseline. Further improvements might be gained if we can identify more specific search tasks. ",

keywords = "cs.IR",

author = "Thomas Schoegje and Chris Kamphuis and Koen Dercksen and Djoerd Hiemstra and Toine Pieters and Vries, {Arjen de}",

note = "Update version 2: Improved title Update version 3: corrected terminology hyponym -> hypernym in two instances Documents our participation to the TREC-COVID track. Contains 16 pages, 0 figures",

year = "2020",

month = oct,

day = "23",

doi = "10.48550/arXiv.2010.12674",

language = "English",

pages = "1--16",

publisher = "arXiv",

type = "WorkingPaper",

institution = "arXiv",

}

TY - UNPB

T1 - Exploring task-based query expansion at the TREC-COVID track

AU - Schoegje, Thomas

AU - Kamphuis, Chris

AU - Dercksen, Koen

AU - Hiemstra, Djoerd

AU - Pieters, Toine

AU - Vries, Arjen de

N1 - Update version 2: Improved title Update version 3: corrected terminology hyponym -> hypernym in two instances Documents our participation to the TREC-COVID track. Contains 16 pages, 0 figures

PY - 2020/10/23

Y1 - 2020/10/23

N2 - We explore how to generate effective queries based on search tasks. Our approach has three main steps: 1) identify search tasks based on research goals, 2) manually classify search queries according to those tasks, and 3) compare three methods to improve search rankings based on the task context. The most promising approach is based on expanding the user's query terms using task terms, which slightly improved the NDCG@20 scores over a BM25 baseline. Further improvements might be gained if we can identify more specific search tasks.

AB - We explore how to generate effective queries based on search tasks. Our approach has three main steps: 1) identify search tasks based on research goals, 2) manually classify search queries according to those tasks, and 3) compare three methods to improve search rankings based on the task context. The most promising approach is based on expanding the user's query terms using task terms, which slightly improved the NDCG@20 scores over a BM25 baseline. Further improvements might be gained if we can identify more specific search tasks.

KW - cs.IR

U2 - 10.48550/arXiv.2010.12674

DO - 10.48550/arXiv.2010.12674

M3 - Preprint

SP - 1

EP - 16

BT - Exploring task-based query expansion at the TREC-COVID track

PB - arXiv

ER -

Exploring task-based query expansion at the TREC-COVID track

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this