Matthias THURNBAUER

Master's thesis

Analysis of Word Embeddings

Abstract:
The influence of the window size on semantic and thematic similarities between words was analyzed for the algorithms Word2Vec and Glove. Furthermore, a disambiguation method was proposed using a density based clustering algorithm called DBSCAN. The report also contains performance measurements regarding the training times of these static word embedding algorithms showing the influence of the token …more
Abstract:
The influence of the window size on semantic and thematic similarities between words was analyzed for the algorithms Word2Vec and Glove. Furthermore, a disambiguation method was proposed using a density based clustering algorithm called DBSCAN. The report also contains performance measurements regarding the training times of these static word embedding algorithms showing the influence of the token …more
 

Keywords

Word Embeddings DBSCAN Cosine Similarity Semantic Thematic Word2Vec GloVe
 
Language used: Czech
Date on which the thesis was submitted / produced: 6. 2. 2023

Thesis defence

  • Supervisor: prof. Dr. Andreas Fischer, doc. Dr.rer.nat. Jan Valdman

Citation record

Cite this text
The right form of listing the thesis as a source quoted

THURNBAUER, Matthias. Analysis of Word Embeddings. České Budějovice, 2023. diplomová práce (Mgr.). JIHOČESKÁ UNIVERZITA V ČESKÝCH BUDĚJOVICÍCH. Přírodovědecká fakulta

Full text of thesis

Contents of on-line thesis archive
Published in Theses:
  • Soubory jsou nedostupné.
Other ways of accessing the text
Institution archiving the thesis and making it accessible: JIHOČESKÁ UNIVERZITA V ČESKÝCH BUDĚJOVICÍCH, Přírodovědecká fakulta
http://wstag.jcu.cz/StagPortletsJSR168/CleanUrl?urlid=prohlizeni-prace-detail&praceIdno=70517 Reference to the local database file of the institution

UNIVERSITY OF SOUTH BOHEMIA IN ČESKÉ BUDĚJOVICE

Faculty of Science

Master programme / field:
Artificial Intelligence and Data Science / Artificial Intelligence and Data Science

Theses on a related topic

List of theses with an identical keyword.
  • No theses on a related topic available.