Der Login über E-Mail und Passwort wird in Kürze abgeschaltet. Für Externe steht ab sofort der Login über ORCID zur Verfügung.
The login via e-mail and password will be retired in the near future. External uses can login via ORCID from now on.
 

Analysis of Automatic Annotation Suggestions for Hard Discourse-Level Tasks in Expert Domains

dc.contributor.author Schulz, Claudia
dc.contributor.author Meyer, Christian M.
dc.contributor.author Kiesewetter, Jan
dc.contributor.author Sailer, Michael
dc.contributor.author Bauer, Elisabeth
dc.contributor.author Fischer, Martin R.
dc.contributor.author Fischer, Frank
dc.contributor.author Gurevych, Iryna
dc.date.accessioned 2019-05-28T12:49:45Z
dc.date.available 2019-05-28T12:49:45Z
dc.date.created 2019-07
dc.date.issued 2019-05-28
dc.description Many complex discourse-level tasks can aid domain experts in their work but require costly expert annotations for data creation. To speed up and ease annotations, we investigate the viability of automatically generated annotation suggestions for such tasks. As an example, we choose a task that is particularly hard for both humans and machines: the segmentation and classification of epistemic activities in diagnostic reasoning texts. We create and publish a new dataset covering two domains and carefully analyse the suggested annotations. We find that suggestions have positive effects on annotation speed and performance, while not introducing noteworthy biases. Envisioning suggestion models that improve with newly annotated texts, we contrast methods for continuous model adjustment and suggest the most effective setup for suggestions in future expert tasks. en_US
dc.identifier.uri https://tudatalib.ulb.tu-darmstadt.de/handle/tudatalib/2001
dc.language.iso de en_US
dc.rights.licenseIn Copyright (https://rightsstatements.org/vocab/InC/1.0/)
dc.subject Diagnostic Reasoning en_US
dc.subject Annotation Suggestion en_US
dc.subject Interactive Machine Learning en_US
dc.subject FAMULUS en_US
dc.subject.ddc 000 Informatik, Informationswissenschaft, allgemeine Werke en_US
dc.title Analysis of Automatic Annotation Suggestions for Hard Discourse-Level Tasks in Expert Domains en_US
dc.type Text en_US
dcterms.accessRights openAccess
person.identifier.orcid 0000-0002-2569-7065
person.identifier.orcid 0000-0002-8673-7665
person.identifier.orcid #PLACEHOLDER_PARENT_METADATA_VALUE#
person.identifier.orcid #PLACEHOLDER_PARENT_METADATA_VALUE#
person.identifier.orcid #PLACEHOLDER_PARENT_METADATA_VALUE#
person.identifier.orcid #PLACEHOLDER_PARENT_METADATA_VALUE#
person.identifier.orcid 0000-0003-2419-6629
person.identifier.orcid 0000-0003-2187-7621
tuda.tubiblio 113668 en_US
tuda.tubiblio 109926

Files

Original bundle

Now showing 1 - 1 of 1
NameDescriptionSizeFormat
famulus-data.zip6.76 MBZIP-Archivdateien Download

Version History

Now showing 1 - 1 of 1
VersionDateSummary
1*
2019-05-28 14:49:45
* Selected version