Transductive Learning for Short-Text Classification Problems Using Latent Semantic Indexing

Abstract

This paper presents work that uses Transductive Latent Semantic Indexing (LSI) for text classification. In addition to relying on labeled training data, we improve classification accuracy by incorporating the set of test examples in the classification process. Rather than performing LSI's singular value decomposition (SVD) process solely on the training data, we instead use an expanded term-by-document matrix that includes both the labeled data as well as any available test examples. We report the performance of LSI on data sets both with and without the inclusion of the test examples, and we show that tailoring the SVD process to the test examples can be even more useful than adding additional training data. This method can be especially useful to combat possible inclusion of unrelated data in the original corpus, and to compensate for limited amounts of data. Additionally, we evaluate the vocabulary of the training and test sets and present the results of a series of experiments to illustrate how the test set is used in an advantageous way.

Short Text Classification Based on Latent Topic Modeling and Word Embedding

“…However, it is unrealistic to query the search engine for each document for the great amount of documents, which is expensive. Another way is to utilize online data repositories, such as Wikipedia [16] and ODP [17], as external knowledge sources [10,11]. This method is more useful and practical.…”

A Method for Chinese Short Text Classification Considering Effective Feature Expansion

“…Therefore, traditional text classification methods cannot be applied to short-text classification better. An effective way of short-text classification is to use some extra information to assist classification [1][2]. Some methods has already achieved certain effect, one of which is based on hyponymy relation and is proposed by Sheng Wang [3] and another one is Agent and Patient Relation Acquisition for Short-text Classification which is proposed by Dingbang Wei [4].…”

On Clustering and Evaluation of Narrow Domain Short-Test Corpora

“…On the one hand, the particular case of short text classification has been investigated in the past in various research projects from both, a categorization and a clustering perspective. In [153], for instance, it is described a method for improving the classification of short text strings by using a combination of labeled training data, plus a secondary corpus of unlabeled but related longer documents. It is shown that such unlabeled background problems [154].…”

“…In [153], for instance, it is described a method for improving the classification of short text strings by using a combination of labeled training data, plus a secondary corpus of unlabeled but related longer documents. It is shown that such unlabeled background problems [154]. LSI is a technique that allows to find a low-rank approximation to a original term-document matrix (describing the occurrences of terms in documents).…”

“…News, document titles, abstracts, FAQs, chats, etc., are some examples of the high volume of short texts available on Internet. Therefore, there exists sufficient interest from the computational linguistic community to analyse the behaviour of classifiers when using short-text corpora [153,54,154,111,21,11,97]. If the short texts belong to the same domain (e.g.…”