Chan-Sheng Kuo

Mining association rules from quantitative data

Data-mining is the process of extracting desirable knowledge or interesting patterns from existing databases for specific purposes. Most conventional data-mining algorithms identify the relationships among transactions using binary values, however, transactions with quantitative values are commonly seen in real-world applications. This paper thus proposes a new data-mining algorithm for extracting interesting knowledge from transactions stored as quantitative values. The proposed algorithm integrates fuzzy set concepts and the apriori mining algorithm to find interesting fuzzy association rules in given transaction data sets. Experiments with student grades at I-Shou University were also made to verify the performance of the proposed algorithm.

A fuzzy AprioriTid mining algorithm with reduced computational time

Abstract Due to the increasing use of very large databases and data warehouses, mining useful information and helpful knowledge from transactions is evolving into an important research area. Most of conventional data mining algorithms identify the relation among transactions with binary values. Transactions with quantitative values are, however, commonly seen in real world applications. In the past, we proposed a fuzzy mining algorithm based on the Apriori approach to explore interesting knowledge from the transactions with quantitative values. This paper proposes another new fuzzy mining algorithm based on the AprioriTid approach to find fuzzy association rules from given quantitative transactions. Each item uses only the linguistic term with the maximum cardinality in later mining processes, thus making the number of fuzzy regions to be processed the same as that of the original items. The algorithm therefore focuses on the most important linguistic terms for reduced time complexity. Experimental results from the data in a supermarket of a department store show the feasibility of the proposed mining algorithm.

A fuzzy data mining algorithm for quantitative values

This paper attempts to propose a new data-mining algorithm to enhance the capability of exploring interesting knowledge from transactions with quantitative values. The proposed algorithm integrates the fuzzy set concepts and the a priori mining algorithm to find interesting fuzzy association rules from given transaction data. Experiments on student grades in I-Shou University are also made to verify the performance of the proposed algorithm.

Mining fuzzy sequential patterns from quantitative data

Data mining is the process of extracting desirable knowledge or interesting patterns from existing databases for specific purposes. Most of the conventional data mining algorithms can identify the relationships among transactions with binary values. Temporal transactions with quantitative values are, however, commonly seen in real-world applications. This paper thus attempts to propose a new data mining algorithm, which takes advantage of fuzzy set theory to enhance the capability of exploring interesting sequential patterns from databases with quantitative values. The proposed algorithm integrates the concepts of fuzzy sets and the AprioriAll algorithm to find interesting sequential patterns and fuzzy association rules from transaction data.

Applying genetic programming technique in classification trees

Classification problems are often encountered in many applications. In the past, classification trees were often generated by decision-tree methods and commonly used to solve classification problems. In this paper, we have proposed an algorithm based on genetic programming to search for an appropriate classification tree according to some criteria. The classification tree obtained can be transferred into a rule set, which can then be fed into a knowledge base to support decision making and facilitate daily operations. Two new genetic operators, elimination and merge, are designed in the proposed approach to remove redundancy and subsumption, thus producing more accurate and concise decision rules than that without using them. Experimental results from the credit card data also show the feasibility of the proposed algorithm.

AN IMPROVED KNOWLEDGE-ACQUISITION STRATEGY BASED ON GENETIC PROGRAMMING

Knowledge acquisition can deal with the task of extracting desirable or useful knowledge from data sets for a practical application. In this paper, we have modified our previous gp-based learning strategy to search for an appropriate classification tree. The proposed approach consists of three phases: knowledge creation, knowledge evolution, and knowledge output. In the creation phase, a set of classification trees are randomly generated to form an initial knowledge population. In the evolution phase, the genetic programming technique is used to generate a good classification tree. In the output phase, the final derived classification tree is transferred as a rule set, then outputted to the knowledge base to facilitate the inference of new data. One new genetic operator, separation, is designed in this proposed approach to remove contradiction, thus producing more accurate classification rules. Experimental results from the diagnosis of breast cancers also show the feasibility of the proposed algorithm.

A Knowledge-Evolution Strategy Based on Genetic Programming

Knowledge evolution is an important issue in knowledge management since enterprises face keen competition and need to keep the latest knowledge with time in an organization. In this paper, we proposed a GP-based knowledge-evolution framework to search for a good integrated classification tree with different evolving time points. The proposed approach can learn the evolving knowledge, integrating original and new knowledge, to deal properly with the organizational need for updating the latest knowledge as time goes on in a dynamic environment. In addition, we developed the initial population, consisting of four proportions, to accomplish suitable diversity and thus raise the search range as well as next learning efficiency in the evolutionary process.

Integrating Multiple Knowledge Sources by Genetic Programming

In this paper, we have proposed a GP-based knowledge-integration framework that automatically combines multiple rule sets into one integrated knowledge base. The proposed framework consists of three phases: knowledge collection and translation, knowledge integration, and knowledge output. Two new genetic operators, abridgement and compromise, are designed in the proposed approach. Experimental results from diagnosis of breast cancer also show the feasibility of the proposed algorithm.