CFP

Previous | item28

                       FIRST CALL FOR PAPERS

                         KDD Challenge 2000
     2000 International Workshop of KDD Challenge on Real-world Data
            Tuesday, April 18, 2000, Keihanna Plaza, Kyoto, Japan

                        in conjunction with:
The Fourth Pacific-Asia Conference on Knowledge Discovery and Data Mining
                            (PAKDD2000)
                         (April 18-20, 2000)

http://www.slab.dnj.ynu.ac.jp/challenge2000/


OBJECTIVES

This workshop intends to bring together researchers and practitioners
interested in KDD applied to real-world data.  Four data sets acquired
from real-world domains are available to all participants, who are
welcome to use on those data all available data preparation, discovery
and presentation techniques.  The workshop will provide the
opportunity to discuss the methods employed and results reached by the
participants.

Most of real-world databases involve irregularities in data format and
quality. Some data may have certain structures which can not be easily
converted to the acceptable formats for the standard KDD methods.
Many real-world data contain noise and missing values. Accordingly,
KDD on real-world data needs wide variety of techniques involving data
selection, preprocessing, format transformation, data mining,
interpretation and evaluation. All required techniques should be
interactively applied and well-tuned to efficiently discover knowledge
embedded in the data. The wide and concrete discussion on the
practical issues of KDD on real-world data is hardly seen in ordinary
presentations of technical conferences.

DATA SETS

We encourage to submit papers challenging to the discovery and mining of
knowledge on any regularity hidden in the following common example data
set. These data sets will be available to all prospective PAKDD2000
conference participants. Meaning of data as well as sample exploration
tasks are also provided.

Data are available and explained at this web site:

http://www.slab.dnj.ynu.ac.jp/challenge2000/

1. Diagnosis data on meningoencephalitis: The data contain both symbolic
and numerical features with relatively few missing values. The size of
the data is relatively small.

2. Bacteriological examination data: The data mainly contain symbolic
features with many missing values.

3. Treatment history data of patients under collagen diseases: The data
have time sequence of events together with the features of
patients. They involve both symbolic and numerical features containing
many missing values.

4. Mutagenicity data: The data contain 3 numerical features including
mutagenic activity together with the structural formulae of 230 nitro
compounds.  The SDF format is used to represent the data, whose
explanation is given separately. There appear no missing values.

The latter example data sets have more complexity of their structure and
lower quality. The authors should describe the approach adopted to any
of these data sets, and provide the discovered knowledge in
comprehensive format in their paper. The appropriateness of the
resultant knowledge is evaluated by domain experts. Upon the
evaluations, the performance of the approach utilized by each
participants are compared and discussed. Finally, the future issues for
KDD on real-world data are investigated.

PAPER SUBMISSION

Submitted papers should be in English and not exceed 6 single-spaced
pages. Use 10pt font, the printed area being 22.5cm (9in) height and
16.3cm (6.5in) width in the center of each page; pages not numbered.
Papers will be reviewed by at least 3 program committee members for
their technical quality, advantage, significance, and relevance to the
workshop. Accepted papers will be reviewed by domain experts on the
supervisory board. Results of their evaluation on discovered knowledge
will be presented during the workshop.  Please send abstract (no more
than 250 words in ASCII text) of the paper via email including title of
the paper, authors' names and 3-4 relevant keywords by March 8, 2000.

We strongly encourage electronic submissions of a full paper in plain
Postscript or PDF file. The title page must include the name and email
address of the contact author. (Alternatively, four hard-copies of your
paper may be sent to the workshop organizer at the address given below.)
Your paper should reach us no later than March 11, 2000.

CONTACT ADDRESS (WORKSHOP PROGRAM COMMITTEE CHAIR)

Prof. Einoshin Suzuki
Electrical and Computer Engineering,
Yokohama National University
79-5, Tokiwadai, Hodogaya, Yokohama, 240-8501, Japan
phone: +81-45-339-4148
fax:   +81.45-338-1157 (Fax of the division)
e-mail: challenge2000@slab.dnj.ynu.ac.jp

IMPORTANT DATES

Abstract Submission Deadline:    March  8, 2000
Paper Submission Deadline:       March 11, 2000
Notification of Acceptance:      March 29, 2000
Camera-ready Copy Due:           April  8, 2000
Workshop Date:                   April 18, 2000

WORKSHOP PROGRAM COMMITTEE

Takashi Okada           (Kwansei Gakuin University)
Einoshin Suzuki         (Yokohama National University)
Terano Takao            (Tsukuba University)
Shusaku Tsumoto         (Shimane Medical University)
Takashi Washio          (Osaka University)
Takahira Yamaguchi      (Shizuoka University)
Jan M Zytkow            (University of North Carolina)

WORKSHOP SUPERVISORY BOARD

Makoto Hayashi           (National Institute of Health Sciences)
Katsuhiko Takabayashi    (Chiba University)
Shusaku Tsumoto          (Shimane Medical University)

Previous | item28