Documents representation for comparable corpora clustering: A preliminary study

Ma, Shutian; Zhang, Chengzhi

Permalink

https://hdl.handle.net/2142/96740 Copy

Description

Title

Documents representation for comparable corpora clustering: A preliminary study

Author(s)

• Ma, Shutian
• Zhang, Chengzhi

Issue Date

2017

Keyword(s)

• Comparable corpora clustering
• Document representation method

Abstract

With increasing globalization, digital libraries tend to provide multilingual documents access. There have been lots of available text information covering the same or similar topic written in multiple languages, namely comparable corpora. To better organize such information with clustering technique, we have explored three document representation methods, Latent Semantic Indexing (LSI), Latent Dirichlet Allocation (LDA) and Doc2Vec (D2V) in task of comparable corpora clustering before. Previously used comparable corpora are in small size of hundred magnitude. In this poster, we use the comparable corpora of regular amount. Methods are found to perform differently when representing dimension sizes are different. Clustering results are investigated according to different representation methods. Choices of the best method for comparable corpora clustering are also discussed.

Publisher

iSchools

Series/Report Name or Number

iConference 2017 Proceedings

Type of Resource

text

Language

en

Permalink

http://hdl.handle.net/2142/96740

Copyright and License Information

Copyright 2017 Shutian Ma and Chengzhi Zhang

Owning Collections