WordNet2Vec: Corpora agnostic word vectorization method

NEUROCOMPUTING

卷:326

Article

Bartusiak, Roman¹  Augustyniak, Lukasz¹  Kajdanowicz, Tomasz¹  Kazienko, Przemyslaw¹  Piasecki, Maciej¹ 

[1] Wroclaw Univ Sci & Technol, Dept Computat Intelligence, Wroclaw, Poland

关键词: Natural language structuring;    WordNet;    WordNet2Vec;    Vectorization;    Network transformation;    Sentiment analysis;    Transfer learning;    Big data;    Complex networks;   

DOI  :  10.1016/j.neucom.2017.01.121

来源: Elsevier

【 摘 要 】

The complex nature of big data resources requires new structuring methods, especially for textual content. WordNet is a good knowledge source for the comprehensive abstraction of natural language as it offers good implementation for many languages. Since WordNet embeds natural language in the form of a complex network, a transformation mechanism, WordNet2Vec, is proposed in this paper. This creates vectors for each word from WordNet. These vectors encapsulate a general position - the role of a given word related to all other words in the given natural language. Any list or set of such vectors contains knowledge about the context of its components within the whole language. This type of word representation can be easily applied to many analytic tasks such as classification or clustering. The usefulness of the WordNet2Vec method is demonstrated in sentiment analysis including the classification of an Amazon opinion text dataset with transfer learning. (C) 2017 Elsevier B.V. All rights reserved.

【 授权许可】

Free