A Survey on Filter Techniques for Feature Selection in Text Mining

Abstract

A large portion of a document is usually covered by irrelevant features. Instead of identifying actual context of the document, such features increase dimensions in the representation model and computational complexity of underlying algorithm, and hence adversely affect the performance. It necessitates a requirement of relevant feature selection in the given feature space. In this context, feature selection plays a key role in removing irrelevant features from the original feature space. Feature selection methods are broadly categorized into three groups: filter, wrapper, and embedded. Filter methods are widely used in text mining because of their simplicity, computational complexity, and efficiency. In this article, we provide a brief survey of filter feature selection methods along with some of the recent developments in this area.

Distance Variance Score: An Efficient Feature Selection Method in Text Classification

“…Feature selection is one of the important tasks in text classification due to the high dimensionality of feature space and the existence of indiscriminative features [1]. Feature selection methods are able to reduce the high-dimensional indiscriminative feature space into low-dimensional discriminative feature subspace [2]. For example, in (2), feature "beautiful" and feature "morning" may be discriminative features and then a six-dimensional DTM can be reduced into a two-dimensional one.…”

“…This paper focuses on filter feature selection and all the methods introduced in this paper are filter feature selection methods. Filter methods are chosen because they are more computationally efficient than wrapper methods [2]. A lot of feature selection methods have been proposed.…”

“…to measure similarity between and and ( − ) 2 to measure similarity between and . For a discriminative feature, ( − ) 2 tends to small, which indicates that each feature selected by LS contains many 0 values. Thus, LS is easily affected by 0 values in DTM and the sub-DTMs constructed by features selected by LS are also sparse.…”

“…and the second is "Good morning, you guys!". DTM of two documents is shown in (2). 1 = [1, 1, 0, 0, 0, 1] and 1 [1] = 11 = 1 show that the first feature "are" exists once in the first text document ("You are beautiful!").…”

RnkHEU: A Hybrid Feature Selection Method for Predicting Students’ Performance

“…Embedded methods are some feature selection methods included in machine learning algorithms, such as feature selection method based on information gain used in decision tree classification algorithm. Feature selection has achieved good results in many fields, such as gene array analysis [11], intrusion detection [12], and text mining [13].…”