WORKSHOP CONTEXT

Over the last two decades, several trends of works have demonstrated how concept lattices formalize conceptual structures by coding any kind of dualities, and can be used to address a variety of problems in Databases, Data Analysis and Machine Learning. These include :

- association rules or data dependencies in Databases, searching frequent item sets, indexing documents for information retrieval...;

- exploring attributes in Data Analysis for binary as well as more complex symbolic data structures...;

- designing concept lattices-based machine Learning Systems for conceptual clustering and classifications...

These studies have generated diverse methods and specific tools along their developments, which are now well implemented, commonly used and have shown the usefulness of this structural approach for each domain. Consequently, it may be the right time to exchange and share the researchers' experiences of these areas, in order to increase lattice-based contributions to Knowledge Discovery in Databases.
 

WORKSHOP GOALS

The main purpose of this workshop is to bring together the members of the Databases, Data Analysis and Machine Learning communities that are working on such concept lattice structures, for considering the current state of art and identifying the strengths -and the potential weaknesses- of these approaches. The workshop will address the general methodologies and principles that are appropriate in the development of concept lattice based Data Mining methods
and tools.

Hence, we are seeking for papers devoted on the theory and practice of concept lattices for modelling conceptual structures and more specifically for KDD. The workshop will focus on the following questions:

- which tasks of KDD's processes can be improved with lattices?

- which sampling data size could reasonably deal with lattices?

- for which data (Boolean, symbolic, numerical, semistructured...)?

- are there "fast" algorithms to design lattices from data?

- how to revise and to incrementally build lattices?

- how to appropriately visualize the generated lattices?

- how to reduce computational complexity of lattices?

- how to interpret the designed lattice and its structure?

- are they "fast" algorithms to generate significant rules from lattices?

- how to reduce the lattice size (approximations, irreducible nodes...)?

These questions should also be demonstrated on real examples through applications of concept lattices for knowledge discovery (Biology, Genetics, Social Sciences, Banking, ...). Emphasis will be given on recent theory and algorithms of concept lattices for KDD, on systems based on concept lattices and their applications in Data Mining.
 
 

INTENDED AUDIENCE

This workshop is firstly aimed at bringing together members of different overlapping communities that share concept lattices as a kernel for developing Data Mining methods and tools:

- those coming from Data Bases, which use lattices for generating association rules or indexing data;

- those coming from Data Analysis, with an interest in formal concept analysis;

- those coming from Machine Learning, developing lattice-based machine learning systems.

This will give an opportunity to push further the discussion upon the potentialities -and limitations- of concept lattices for Knowledge Discovery in Databases accross these communities, but also to spread it furthermore among other trends of works in Conceptual Structures.
 
 

WORKSHOP FORMAT

The workshop is planned to include three kinds of session:

Invited talks: The workshop will feature a number of invited presentations by people who have contributed to these areas.

Technical papers: Contributors to the workshop will be requested to highlight only key contributions when presenting their work, in order to set aside the maximum possible time for discussion and interaction.

Panel/discussion sessions: Key issues will be debated by panellists and the workshop participants as a whole.
 
 

HOW TO CONTRIBUTE

We invite submissions of technical papers and short position papers (two pages). Authors of accepted technical and position papers will be invited to participate in the workshop.
Format requirements for submissions of technical papers are:
- maximum 12 double-spaced pages, excluding title page and bibliography
- all submissions should be made electronically if possible, by email attachment and preferably in Postscript format. Only if electronic submission is impossible should you send three hardcopies.
- all submissions must be sent to the primary workshop contact, Engelbert Mephu Nguifo, at the address below.
 
 

IMPORTANT DATES

8 May  2001       Submission of technical and position papers
8 June 2001       Notification of acceptance
1 July 2001       Camera-ready papers

Preliminary Workshop Program

Session 5 : General Discussion and Conclusion
5.30 pm - 5.50pm : General Discussion
5.50 pm - 6.00 pm : Conclusion
 
 

WORKSHOP ORGANISING COMMITEE