Zur Kurzanzeige

dc.contributor.authorPustulka, Elzbieta
dc.contributor.authorHanne, Thomas
dc.contributor.authorBlumer, Eliane
dc.contributor.authorFrieder, Manuel
dc.contributor.editorWong, Ka Chun
dc.date.accessioned2019-08-19T07:04:35Z
dc.date.available2019-08-19T07:04:35Z
dc.date.issued2018-08-27
dc.identifier.urihttp://hdl.handle.net/11654/27841
dc.identifier.urihttp://dx.doi.org/10.26041/fhnw-1765
dc.description.abstractWe are developing a multilingual sentiment analysis solution for a Swiss human resource company working in the gig sector. To examine the feasibility of using machine learning in this context, we carried out three sentiment assignment experiments. As test data we use 963 hand annotated comments made by workers and their employers. Our baseline, machine learning (ML) on Twitter, had an accuracy of 0.77 with the Matthews correlation coefficient (MCC) of 0.32. A hybrid solution, Semantria from Lexalytics, had an accuracy of 0.8 with MCC of 0.42, while a tenfold cross-validation on the gig data yielded the accuracy of 0.87, F1 score 0.91, and MCC 0.65. Our solution did not require language assignment or stemming and used standard ML software. This shows that with more training data and some feature engineering, an industrial strength solution to this problem should be possible.
dc.description.urihttp://www.iscbi.com
dc.language.isoen_US
dc.relation.ispartof6th International Symposium on Computational and Business Intelligence (ISCBI 2018)
dc.accessRightsAnonymous
dc.subjectsentiment analysis, machine learning application, natural language processing, gig economy
dc.titleMultilingual Sentiment Analysis for a Swiss Gig
dc.type04 - Beitrag Sammelband oder Konferenzschrift
dc.spatialBasel
dc.event.start2018-08-27
dc.event.end2018-08-29
dc.audienceScience
fhnw.publicationStatePre-print in printing
fhnw.ReviewTypeAnonymous ex ante peer review of a complete publication
fhnw.InventedHereYes
fhnw.PublishedSwitzerlandYes
fhnw.IsStudentsWorkno


Dateien zu dieser Ressource

Thumbnail

Der Eintrag erscheint in:

Zur Kurzanzeige