The manual classification of a large amount of textual materials are very costly in time and personnel. For this reason, a lot of research has been devoted to the problem of automatic classification and work on the subject dates from 1960. A lot of text classification software has appeared. For some tasks, automatic classifiers perform almost as well as humans, but for others, the gap is still large. These systems are directly related to machine learning. It aims to achieve tasks normally affordable only by humans. There are generally two types of learning: learning “by heart,” which consists of storing information as is, and learning generalization, where we learn from examples. In this chapter, the authors address the classification concept in detail and how to solve different classification problems using different machine learning techniques.
TopIntroduction
The manual classification of a large amount of textual materials are very costly in time and personnel for this reason a lot of research has been devoted to the problem of automatic classification and work on the subject dates from 1960.
These days a lot of text classification software have appeared For some tasks, automatic classifiers perform almost as well as humans, but for others, the gap is still large, these systems are directly related to machine learning (machine learning) it aims to achieve tasks normally affordable only by human there are generally two types of learning: learning “by heart” which consists of storing information as is, and learning generalization where we learn from examples a model we will recognize new examples (Ko, 2012).
In this chapter we will address the classification concept in detail how to solve different classification problems were appealed to different machine learning kind. The text classification is a generic task of assigning one or more categories from a predefined list or not a document by finding a functional link between a set of texts and a set of categories (tags, classes) according to criteria. They apply to many human activities and particularly suited to the problems of automated decision-making.
For example: we are faced with a set of target text and the goal is to make a computer application capable of autonomously determine in which category classified each text, based on statistical data. Habitually the categories refer to the text subjects, but for particular applications, they can take other forms (Nigam, 2000).
Figure 1. Classification of texts (Sriram, 2010)
Classification can be found in several application domains as the language identification, recognition of writing, Categorization of multimedia document, spam detection, speech recognition, aid to medical diagnosis.
TopClassification Dual Class And Multi Class
The Bi-Classes Classification
The bi-classes classification is a kind of problem examples for which the classification system answers the question: Does the text belongs to the category “C” or to its complementary class (Wu, 2004). some tasks Typical of binary classification are: determining whether a document is it plagiarize or not, Medical tests to determine if a patient has certain disease or not, quality control in the factories decide whether a new product is good enough to be sold, or whether it should be eliminated.
III 3-2-The Multi-Class Classification
The multi-class classification is the problem of the classification of examples in more than two classes or given example can be associated with one or more classes or no class is the most general case of the classification, which can be used in multiple domain: such as handwriting recognition which is part of the marking speech, speech recognition. The classification methods Binary can be generalized in many ways to manage multiple classes there are two approaches to perform this generalization. (Wu, 2004).
- •
Approach One-Against-All (OAA): Allows the direct use of standard binary classifiers to encode and form the output label and assume that for each class, there is only one (single) separation between this class and all other classes that is to say each class is opposed to all other classes. The mth classifier separates the data from the M class all data remains latter takes one sign and another class -1.
- •
One Against One (OAO):Is a more expressive alternative that assumes the existence of a separator between each two classes this method is called a one against this is to say that each class is discriminated from another or one must calculate M (M-1) / 2 classifier (Polat, 2009).
TopWhy Automatic Document Classification?
The major consequences that pose a manual processing (by a human) classification of textual documents are: