Evolving Rules for Document Classification

Abstract

Abstract. We describe a novel method for using Genetic Programming to create compact classification rules based on combinations of N-Grams (character strings). Genetic programs acquire fitness by producing rules that are effective classifiers in terms of precision and recall when evaluated against a set of training documents. We describe a set of functions and terminals and provide results from a classification task using the Reuters 21578 dataset. We also suggest that because the induced rules are meaningful to a human analyst they may have a number of other uses beyond classification and provide a basis for text mining applications.

SPAM Detection: Naïve Bayesian Classification and RPN Expression-Based LGP Approaches Compared

“…Hirsch et al [23] implemented a GP system to evolve rules based on n-grams (strings n character long) for document classification. GP has been used by Shengen et al [24] to generate new features used as inputs to Support Vector Machines (SVM) and GP classifiers to detect link webspam.…”

A Review of evolutionary Algorithms for Data Mining

“…Specific kinds of classification problems may well be more effectively solved by EAs using rule representations "tailored" to the target kind of problem. For instance, (Hirsch et al 2005) propose a rule representation tailored to document classification (i.e., a text mining problem), where strings of characters -in general fragments of words, rather than full words -are combined via Boolean operators to form classification rules.…”

User Behavior Analysis of the Open-Ended Document Classification System

“…T and B are respectively defined as T = {t 1 , t 2 In the inference process, the MCRDR-Classifier evaluates each rule node of the knowledge base (KB). If a case is selected from the case list (CL), the system evaluates rules from the root node and the inference result is provided by the last rule satisfied in a pathway.…”

“…Although ML method produced accurate classifiers, there are a number of drawbacks as compared to a rule-based one [2]. Some limitations of the ML approach come from its assumptions about the document classification problem.…”