Relational Concept Analysis: Mining Multi-relational Datasets for Assisted Class Model - LIRMM - Laboratoire d’Informatique, de Robotique et de Microélectronique de Montpellier Access content directly
Other Publications Year : 2014

Relational Concept Analysis: Mining Multi-relational Datasets for Assisted Class Model

Abstract

Formal Concept Analysis, is a well established framework for extracting an ordered set of concepts from a dataset, called a Formal Context, composed of entities described by characteristics. This data analysis framework is currently applied to support various tasks, including information retrieval, data mining, or ontology alignment. It also has many applications in software engineering such as software understanding, extracting or maintaining class models, or software reengineering. Relational Concept Analysis (RCA) is an extension of the FCA framework to multi-relational datasets, namely datasets composed of several categories of entities described by both characteristics and inter-entities links. RCA generates a set of concept lattices, precisely one for each category of entity. The concepts are connected via "relational attributes" that are abstractions of the initial links and traverse the lattice frontier. The concept lattice set is a particular view on the dataset, which reveals implication rules involving relationships, as well as relevant connections between classified groups of entities. In this talk, we introduce RCA and we explain its strengths and limits. Then we develop an exploratory approach for assisting a domain expert in class model evolution, more precisely for the class model building and for the follow-up of the class model abstraction level.
Fichier principal
Vignette du fichier
Huchard.pdf (2.15 Mo) Télécharger le fichier
Origin Publisher files allowed on an open archive
Loading...

Dates and versions

lirmm-01075525 , version 1 (17-10-2014)

Identifiers

  • HAL Id : lirmm-01075525 , version 1

Cite

Marianne Huchard. Relational Concept Analysis: Mining Multi-relational Datasets for Assisted Class Model. 2014. ⟨lirmm-01075525⟩
324 View
232 Download

Share

Gmail Mastodon Facebook X LinkedIn More