department of informatics

Fuzzy Classification Query Language (fCQL)

Description: 

We present the framework of the fuzzy Classification Query Language (fCQL) for data mining in information systems. This framework is applicable for very large relational databases and has been implemented using appropriate fuzzy classification, linguistic variables and fuzzy sets at the database schema level. The fuzzy Classification Query Language provides easy-to-use functionality for data extraction similar to the conventional Structured Query Language (SQL). We propose the fCQL for using as data mining tool in information systems due to its integration with conventional relational databases, its power for flexible data analysis and improvement of information presentation and report generation. We are looking forward for your feedback regarding the presented approach and we hope that you will find it useful for solving your tasks.


Project Head: Prof. Dr. Andreas Meier

Introduction

The vast expansion of the Internet gives rise to a significant growth of the number of huge online databases. Therefore, data mining with its goal of reducing the complexity and the number of computations in very large databases attracts more and more attention from the information industry. At the stage of the typical data classification in information systems, there appear some types of uncertainty, for instance, when the boundaries of a class of objects are not sharply defined. In this case, the most common, useful and widely accepted solution is the introduction of fuzzy sets. Fuzzy sets provide mathematical meanings to the natural language statements and become an effective solution for dealing with uncertainty. In particular, only one property or measure seldom defines a business process or the quality of a service. In most practical classification situations, more than one attribute or criteria must be considered simultaneously. There is no simple procedure for combining the different criteria into one general performance measure because the criteria are measured with different scales, the relative significance of different criteria differs and for some criteria the objective is maximization, but for others it is minimization or a specific target. The approach of fuzzy set theory with its membership functions is widely used to form a realistic description of the evaluation. Different criteria with separate scales and optimization objectives can be combined into a joint response measure - the aggregated value of the membership.

A number of different schemes and tools for the implementation of fuzzy sets in database management systems have been proposed in recent years, such as fuzzy querying, fuzzy extension of the structured query language (SQL), fuzzy object oriented database schemes, etc. However, the reality shows that despite of a large importance of data mining, fuzzy methods are not widely used in relational database systems in practice. Why is this the case? The main reason is that most of the proposed fuzzy methods for data mining require to change the conventional relational database structure or to add special features to the database management tools, for example, to modify the functionality of the conventional SQL. Most of the database system owners and users do not want to switch to fuzzy database structures and, therefore, cannot apply fuzzy querying (compare following figure).

fCQL Toolkit

We propose a framework based on the fuzzy Classification Query Language (fCQL) for data mining and data warehouse management. The main benefit of this framework is that there is no need to modify the functionality of the conventional relational databases. All manipulations can be done as an extension of the database schema, applying fuzzy data classification and fCQL which is based on the conventional SQL. Therefore, all benefits of using fuzzy sets and fuzzy classification in data mining, like user-friendly data presentation in the report generation phase, introducing linguistic variables and fuzzy values, easy-to-use facilities for querying the extended database schema, abstraction of the data using classes, etc., will become available for the users of the conventional relational databases

Benefits

The main benefits of using fCQL are the following:

  1. The fCQL framework allows human-oriented queries with the help of linguistic variables. Those queries are therefore much more intuitive, implying an easier querying process and a better results retrieval.

  2. fCQL by providing more information than traditional classification tools improves business decisions.

  3. By operating on a linguistic level, fCQL encapsulates the complexity of the domain. Experts can define and adapt the parameters of the fuzzy classification without interfering with the end users.

  4. Being an additional layer above the traditional relational database management systems (RDBMS), fCQL is independent of the underlying database systems and thus can operate with every RDBMS.

  5. No modification or migration of the underlying databases, which are in practice very large, is needed.

  6. If business perspectives change for marketing, sales or supply chain management, only the database schema has to be adapted

Publications

  • Nicolas Werro: Fuzzy Classication of  Online Customers, Dissertation, University of Fribourg, 2008.
    Download dissertation (PDF, 2'357 KB)
  • Andreas Meier, Günter Schindler and Nicolas Werro: Fuzzy Classification on Relational Databases, Accepted for publication in J. Galindo (Ed.), Handbook of Research on Fuzzy Information Processing in Databases. Hershey: Information Science Reference, 2008.
    Order at Amazon
  • Andreas Meier and Nicolas Werro: A Fuzzy Classification Model for Online Customers, Informatica - International Journal of Computing and Informatics, 31:175-182, 2007.
    Download paper (PDF, 164 KB)
  • Darius Zumstein, Nicolas Werro and Andreas Meier: Fuzzy Portfolio Analysis for Strategic Customer Relationship Management, Internal Working Paper no 07-01, University of Fribourg, Switzerland, Januar 2007.
  • Henrik Stormer, Nicolas Werro and Daniel Risch: An Experiment on Recommender Systems for SME Online Shops, Proceedings of the 7th International Working For E-Business Conference, We-B 2006, Melbourne, Australia, November 2006.
  • Nicolas Werro, Henrik Stormer and Marco Savini: eSarine - Le Magasin Electronique pour PME, Proceedings of the 'Congrès International Francophone en Entrepreneuriat et PME', CIFEPME 2006, Fribourg, Switzerland, October 2006.
    Download paper (PDF, 210 KB)
  • Nicolas Werro, Henrik Stormer and Andreas Meier: A Hierarchical Fuzzy Classification of Online Customers, Proceedings of the IEEE International Conference on e-Business Engineering, ICEBE 2006, Shanghai, China, October 2006.
    Download at IEEE
  • Andreas Meier and Nicolas Werro: Extending a Webshop with a Fuzzy Classification Model for Online Customers, Proceedings of the IADIS International Conference, e-Society 2006, Dublin, Ireland, July 2006. This paper has been selected for an Outstanding Paper Award by the IADIS International Conference Committee.
  • Henrik Stormer, Nicolas Werro and Daniel Risch: Recommending Products by the mean of a Fuzzy Classification, Proceedings of the European Conference on Collaborative Electronic Commerce Technology and Research, CollECTeR 2006, Basel, Switzerland, June 2006.
  • Nicolas Werro, Henrik Stormer and Andreas Meier: Personalized Discount - A Fuzzy Logic Approach, Proceedings of the 5th IFIP International Conference on eBusiness, eCommerce and eGovernment, I3E 2005, Poznan, Poland, October 2005. (ISBN 0-387-28753-1)
    Download at Springer
  • Andreas Meier, Nicolas Werro, Martin Albrecht and Miltiadis Sarakinos: Using a Fuzzy Classification Query Language for Customer Relationship Management, Proceedings of the 31st International Conference on Very Large Data Bases, VLDB 2005, Trondheim, Norway, August 2005.
    Download paper (PDF, 353 KB)
  • Nicolas Werro, Andreas Meier, Christian Mezger and Günter Schindler: Concept and Implementation of a Fuzzy Classification Query Language, Proceedings of the International Conference on Data Mining, DMIN'05, World Congress in Applied Computing, Las Vegas, USA, June 2005. (ISBN 1-932415-79-3)
  • Nicolas Werro, Henrik Stormer, Daniel Frauchiger and Andreas Meier: eSarine - A Struts-based Webshop for Small and Medium-sized Enterprises, Proceedings of the EMISA Conference - Information Systems in E-Business and E-Government, Luxembourg, October 2004. (ISBN 3-88579-385-7)
  • Daniel Frauchiger, Andreas Meier, Henrik Stormer und Nicolas Werro: Zur Entwicklung des Struts-basierten Webshops eSarine, HMD - Praxis der Wirtschaftsinformatik, dpunkt Verlag, Vol. 238, August 2004. (ISBN 3-89864-291-7)
  • Andreas Meier, Christian Mezger, Nicolas Werro und Günter Schindler: Zur unscharfen Klassifikation von Datenbanken mit fCQL, Proceedings of the GI-Workshop LLWA - Lehren, Lernen, Wissen, Adaptivität, Karlsruhe, Germany, October 2003.
    Download paper (PDF, 56 KB)

Back to top