Change search
ReferencesLink to record
Permanent link

Direct link
Monitoring Term Drift Based on SemanticConsistency in an Evolving Vector Field
University of Borås, Faculty of Librarianship, Information, Education and IT.ORCID iD: 0000-0002-1539-8256
University of Borås, Faculty of Librarianship, Information, Education and IT. (DIGRES)
Show others and affiliations
2015 (English)In: Proceedings of IJCNN-15, 2015Conference paper (Refereed)
Abstract [en]

Based on the Aristotelian concept of potentialityvs. actuality allowing for the study of energy and dynamics inlanguage, we propose a field approach to lexical analysis. Fallingback on the distributional hypothesis to statistically model wordmeaning, we used evolving fields as a metaphor to express timedependentchanges in a vector space model by a combinationof random indexing and evolving self-organizing maps (ESOM).To monitor semantic drifts within the observation period, anexperiment was carried out on the term space of a collection of12.8 million Amazon book reviews. For evaluation, the semanticconsistency of ESOM term clusters was compared with theirrespective neighbourhoods in WordNet, and contrasted withdistances among term vectors by random indexing. We found thatat 0.05 level of significance, the terms in the clusters showed a highlevel of semantic consistency. Tracking the drift of distributionalpatterns in the term space across time periods, we found thatconsistency decreased, but not at a statistically significant level.Our method is highly scalable, with interpretations in philosophy.

Place, publisher, year, edition, pages
Keyword [en]
evolving semantics, index terms, vector field, self-organizing maps, knowledge organization, digital preservation
National Category
Other Computer and Information Science
Research subject
Library and Information Science
URN: urn:nbn:se:hb:diva-8527OAI: diva2:894316
International Joint Conference on Neural Networks
EU, FP7, Seventh Framework Programme, FP7-601138
Available from: 2016-01-14 Created: 2016-01-14 Last updated: 2016-02-18Bibliographically approved

Open Access in DiVA

fulltext(120 kB)11 downloads
File information
File name FULLTEXT01.pdfFile size 120 kBChecksum SHA-512
Type fulltextMimetype application/pdf

Other links

Search in DiVA

By author/editor
Wittek, PeterDarányi, Sándor
By organisation
Faculty of Librarianship, Information, Education and IT
Other Computer and Information Science

Search outside of DiVA

GoogleGoogle Scholar
Total: 11 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 255 hits
ReferencesLink to record
Permanent link

Direct link