Files in this item



application/pdf061_ready.pdf (515kB)
(no description provided)PDF


application/octet-stream061.epub (3MB)
(no description provided)Unknown


application/ (619kB)
(no description provided)Unknown


Title:A Crowdsourcing Approach for Finding Misidentifications of Bibliographic Records
Author(s):Morishima, Atsuyuki; Shiori, Tomita; Kawashima, Takanori; Harada, Takashi; Uda, Norihiko; Sato, Sho; Abematsu, Yukihiko
bibliographic records
human-powered join
Abstract:Because there is no perfect technique for automatic identification of bibliographic records, cleaning the identification results manually is indispensable. However, to recruit human resources for the task is often difficult. This paper discusses a microtask-based crowdsourcing approach to the problem. An important issue is to design a good strategy for generating tasks to be assigned to workers, maintaining the quality and reducing the number of tasks. In this study, we explore a design space defined by two criteria to reduce the number of assigned microtasks for finding misidentifications caused by automatic identification techniques. We compare four task-generation strategies using bibliographic records of the National Diet Library. One of the strategies reduced 55.7% of tasks from the baseline strategy and statistic analysis showed that the quality of its result is comparable to those of the other three strategies.
Issue Date:2014-03-01
Citation Info:Morishima, A., Shiori, T., Kawashima, T., Harada, T., Uda, N., Sato, S., & Abematsu, Y. (2014). A Crowdsourcing Approach for Finding Misidentifications of Bibliographic Records. In iConference 2014 Proceedings (p. 177 - 191). doi:10.9776/14061
Series/Report:iConference 2014 Proceedings
Publication Status:published or submitted for publication
Peer Reviewed:is peer reviewed
Rights Information:Copyright 2014 is held by the authors of individual items in the proceedings. Copyright permissions, when appropriate, must be obtained directly from the authors.
Date Available in IDEALS:2014-02-28

This item appears in the following Collection(s)

Item Statistics