Word embeddings or distributed representations of words are being used in various applications like machine translation, sentiment analysis, topic identification etc. Quality of word embeddings and performance of their applications depends on several factors like training method, corpus size and relevance etc. In this study we compare performance of a dozen of pretrained word embedding models on lyrics sentiment analysis and movie review polarity tasks. According to our results, Twitter Tweets is the best on lyrics sentiment analysis, whereas Google News and Common Crawl are the top performers on movie polarity analysis. Glove trained models slightly outrun those trained with Skipgram. Also, factors like topic relevance and size of corpus significantly impact the quality of the models. When medium or large-sized text sets are available, obtaining word embeddings from same training dataset is usually the best choice.

Quality of Word Embeddings on Sentiment Analysis Tasks / Çano, Erion; Morisio, Maurizio. - ELETTRONICO. - 10260:(2017), pp. 332-338. (Intervento presentato al convegno NLDB 2017 22nd International Conference on Natural Language & Information Systems tenutosi a Liege, Belgium nel 21 - 23 June 2017) [10.1007/978-3-319-59569-6_42].

Quality of Word Embeddings on Sentiment Analysis Tasks

Çano, Erion;Morisio, Maurizio
2017

Abstract

Word embeddings or distributed representations of words are being used in various applications like machine translation, sentiment analysis, topic identification etc. Quality of word embeddings and performance of their applications depends on several factors like training method, corpus size and relevance etc. In this study we compare performance of a dozen of pretrained word embedding models on lyrics sentiment analysis and movie review polarity tasks. According to our results, Twitter Tweets is the best on lyrics sentiment analysis, whereas Google News and Common Crawl are the top performers on movie polarity analysis. Glove trained models slightly outrun those trained with Skipgram. Also, factors like topic relevance and size of corpus significantly impact the quality of the models. When medium or large-sized text sets are available, obtaining word embeddings from same training dataset is usually the best choice.
2017
978-3-319-59569-6
File in questo prodotto:
File Dimensione Formato  
ErionCanoWordEmbNLDB2017.pdf

accesso aperto

Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 461.39 kB
Formato Adobe PDF
461.39 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2668229
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo