Trust, but Verify: Predicting Contribution Quality for Knowledge Base Construction and Curation

From Wikipedia Quality
Revision as of 23:38, 3 July 2018 by Librarian (talk | contribs) (New scientific work)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Trust, but Verify: Predicting Contribution Quality for Knowledge Base Construction and Curation
Authors
Chunhow Tan
Eugene Agichtein
Panos G. Ipeirotis
Evgeniy Gabrilovich
Publication date
2014
ISBN
978-145032351-2
DOI
10.1145/2556195.2556227
Links

Trust, but Verify: Predicting Contribution Quality for Knowledge Base Construction and Curation - scientific work about Wikipedia quality published in 2014, written by Chunhow Tan, Eugene Agichtein, Panos G. Ipeirotis and Evgeniy Gabrilovich.

Overview

The largest publicly available knowledge repositories, such as Wikipedia and Freebase, owe their existence and growth to volunteer contributors around the globe. While the majority of contributions are correct, errors can still creep in, due to editors' carelessness, misunderstanding of the schema, malice, or even lack of accepted ground truth. If left undetected, inaccuracies often degrade the experience of users and the performance of applications that rely on these knowledge repositories. Authors present a new method, CQUAL, for automatically predicting the quality of contributions submitted to a knowledge base. Significantly expanding upon previous work, their method holistically exploits a variety of signals, including the user's domains of expertise as reflected in her prior contribution history, and the historical accuracy rates of different types of facts. In a large-scale human evaluation, their method exhibits precision of 91% at 80% recall. Their model verifies whether a contribution is correct immediately after it is submitted, significantly alleviating the need for post-submission human reviewing.

Embed

Wikipedia Quality

Tan, Chunhow; Agichtein, Eugene; Ipeirotis, Panos G.; Gabrilovich, Evgeniy. (2014). "[[Trust, but Verify: Predicting Contribution Quality for Knowledge Base Construction and Curation]]". Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Volume 8359 LNCS, 2014, pp. 14-28. ISBN: 978-145032351-2. DOI: 10.1145/2556195.2556227.

English Wikipedia

{{cite journal |last1=Tan |first1=Chunhow |last2=Agichtein |first2=Eugene |last3=Ipeirotis |first3=Panos G. |last4=Gabrilovich |first4=Evgeniy |title=Trust, but Verify: Predicting Contribution Quality for Knowledge Base Construction and Curation |date=2014 |isbn=978-145032351-2 |doi=10.1145/2556195.2556227 |url=https://wikipediaquality.com/wiki/Trust,_but_Verify:_Predicting_Contribution_Quality_for_Knowledge_Base_Construction_and_Curation |journal=Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Volume 8359 LNCS, 2014, pp. 14-28}}

HTML

Tan, Chunhow; Agichtein, Eugene; Ipeirotis, Panos G.; Gabrilovich, Evgeniy. (2014). &quot;<a href="https://wikipediaquality.com/wiki/Trust,_but_Verify:_Predicting_Contribution_Quality_for_Knowledge_Base_Construction_and_Curation">Trust, but Verify: Predicting Contribution Quality for Knowledge Base Construction and Curation</a>&quot;. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Volume 8359 LNCS, 2014, pp. 14-28. ISBN: 978-145032351-2. DOI: 10.1145/2556195.2556227.