Noise-free attribute-oriented induction

Hsiao Wei Hu, Yen Liang Chen, Jia Yu Hong

Research output: Contribution to journalArticlepeer-review

Abstract

Attribute-oriented induction (AOI) was originally developed to facilitate the mining of generalized knowledge in relational databases. Input data for the AOI method comprises a relational table and a concept tree for each attribute. The output is a small relation that contains a number of generalized tuples which summarize the general characteristics of the relational table. Ideally, the generalized tuples shown in the induction table represent the patterns of information that appear in the table. However, if the input data contains a large amount of noise, the generalized tuples may contain too little information to be useful. Existing research into AOI has yet to focus on the elimination of noise. To fill this gap, we developed two noise-free AOI algorithms that filter out noise to enhance the specificity of AOI results.

Original languageEnglish
Pages (from-to)333-349
Number of pages17
JournalInformation Sciences
Volume568
DOIs
StatePublished - Aug 2021

Keywords

  • Attribute-Oriented Induction
  • Concept Hierarchy
  • Data mining
  • Noise

Fingerprint

Dive into the research topics of 'Noise-free attribute-oriented induction'. Together they form a unique fingerprint.

Cite this