Towards Hybrid NER: A Study of Content and Crowdsourcing-Related Performance Factors thumbnail
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Towards Hybrid NER: A Study of Content and Crowdsourcing-Related Performance Factors

Published on Jul 15, 20151495 Views

This paper explores the factors that influence the human component in hybrid approaches to named entity recognition (NER) in microblogs, which combine state-of-the-art automatic techniques with hum

Related categories

Chapter list

Towards hybrid NER: a study of content and crowdsourcing-related performance factors 00:00
Motivation00:23
Overview: Named entity recognition in tweets00:55
Aims of this work 02:33
Research hypotheses 03:07
Task design04:13
Platform: Wordsmith05:02
Experiment06:22
Datasets08:29
Gold standard: entity definition 09:14
Gold standard: entity type mapping 09:58
Results10:37
H1.1 Number of entities 10:48
H1.2 Micropost length 11:18
H1.3 Entity types 11:35
H2.1 Skipped tweets: number of entities 12:24
H2.1 Skipped tweets: Micropost length 12:46
H2.1 Skipped tweets: Entity types 13:13
H2.2 Avg. accurate annotation time (secs)13:50
H2.3 Accuracy of annotation 15:36
Discussion: Difficult cases16:06
Discussions: Implicit entities 18:55
Summary20:36
Thank you22:13