ITSA ⋆ : An Effective Iterative Method for Short-Text Clustering Tasks

Abstract

Abstract. The current tendency for people to use very short documents, e.g. blogs, text-messaging, news and others, has produced an increasing interest in automatic processing techniques which are able to deal with documents with these characteristics. In this context, "short-text clustering" is a very important research field where new clustering algorithms have been recently proposed to deal with this difficult problem. In this work, ITSA , an iterative method based on the bio-inspired method PAntSA is proposed for this task. ITSA takes as input the results obtained by arbitrary clustering algorithms and refines them by iteratively using the PAntSA algorithm. The proposal shows an interesting improvement in the results obtained with different algorithms on several short-text collections. However, ITSA can not only be used as an effective improvement method. Using random initial clusterings, ITSA outperforms well-known clustering algorithms in most of the experimental instances.

An efficient Particle Swarm Optimization approach to cluster short texts

“…CLUDIPSO and CLUDIPSO ⋆ used the 11 Corpora are considered medium size if the number of documents is between 1000 and 10000.…”

“…The study of alternative methods to perform the initial sub-grouping and start the search process with higher quality initial subgroups could be an interesting extension. Another possible extension, is to use the results obtained with CLUDIPSO ⋆ as input to a recent (and effective) boosting method [11] and obtain thus, higher quality solutions.…”

“…These corpora are cataloged as medium size. 11 The complete versions of R8-Test and R8-Train corpora were considered in this work. Also, a larger version of JRC6 corpus named JRC-Full containing a larger amount of short documents (in fact, all the short texts of six categories) was considered.…”

Silhouette + attraction: A simple and effective method for text clustering

“…In particular, some recent bio-inspired proposals have gained increasing interest in short -text clustering. These approaches include algorithms based on Particle Swarm Optimization techniques (Cagnina et al 2008; Ingaramo et al 2009) and ant-behavior-based approaches (Errecalde, Ingaramo and Rosso 2010; Ingaramo, Errecalde and Rosso 2010 a, Errecalde and Rosso 2010 b).…”

“…Among the considerable number of ICVMs proposed up to now, the GS coefficient has shown good results as cluster validation method with respect to other well-known validity measures (Brun et al 2007). Furthermore, the silhouette coefficient has also shown its potential for determining the optimal number of groups in a clustering problem (Rousseeuw 1987; Tan et al 2005; Choi et al 2011), estimating how difficult a corpus is for an arbitrary clustering algorithm (Errecalde et al 2008), computing a target function to be optimized (Cagnina et al 2008; Ingaramo et al 2009), automatically determining a threshold for a similarity function (Bonato dos Santos et al 2011) and as a key component in other internal process of clustering algorithms (Aranganayagi and Thangavel 2007; Errecalde et al 2010; Ingaramo et al 2010a).…”