What is Discovery Science?

Einoshin Suzuki, Kyushu University, Japan
Jan. 3, 2011

My Definition

Discovery Science is a scientific discipline on any discovery process that is mainly approached by computer science. Highly relevant research fields are knowledge discovery in databases and scientific discovery, followed by machine learning and statistics. Data mining and data analysis may be considered as fruitful applications. Research in discovery science is more descriptive than predictive, focuses more on interestingness than accuracy, usually tends to focus on a part of the example space rather than the whole, and may be applied to both labeled and unlabeled data.

The Project

The name "Discovery Science" comes from the Discovery Science project which was funded by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan as a Grant-in-Aid for Scientific Research on Priority Area. It involved more than sixty scientists across various academic disciplines [1]. The principal investigator of the project was Professor Setsuo Arikawa of Kyushu University and it ran from April 1998 to March 2001. The primary goals of this project are to (1) develop new methodologies for knowledge discovery, (2) install network environments for knowledge discovery, and (3) establish Discovery Science as a new discipline of Computer Science/Artificial Intelligence Study. In order to achieve these goals, five groups were set up for studying the following respective research areas: (A) Logic for/of Knowledge Discovery, (B) Knowledge Discovery by Inference/Reasoning, (C) Knowledge Discovery Based on Computational Learning Theory, (D) Knowledge Discovery in Huge Database and Data Mining, and (E) Knowledge Discovery in Network Environments. These research areas and related topics can be regarded as a preliminary definition of Discovery Science by enumeration. Thus Discovery Science ranges over philosophy, logic, reasoning, computational learning and system developments. The final report of the project can be found in various references including [2].

The Conferences

International conference on Discovery Science started in 1998 and has been held every year. Around 1030 authors have contributed to the succeeds of the Discovery Science conference series during the years 1998-2010. [3] serves as an archive of the locations, the organizers, the invited speakers, and relevant information of the conference. Due to the constraints imposed by MEXT, the first three conferences were held in Japan, i.e., DS 1998 in Fukuoka, DS 1999 in Tokyo, and DS 2000 in Kyoto. Since 2001, the DS conference has been co-located with the ALT (Algorithmic Learning Theory) conference and shares important academic and social activities such as the invited lectures and the conference banquets. Many countries have hosted the conference: DS 2001 in Washington D.C.(USA), DS 2002 in Luebeck (Germany), DS 2003 in Sapporo (Japan), DS 2004 in Padova (Italy), DS 2005 in Singapore (Singapore), DS 2006 in Barcelona (Spain), DS 2007 in Sendai (Japan), DS 2008 in Budapest (Hungary), DS 2009 in Porto (Portugal), and DS 2010 in Canberra (Australia). In 2005, Carl Smith Award was founded in memory of Professor Carl Smith, who contributed to the success of the DS conference and who passed away in 2004. The award is given to an excellent student paper at each DS conference. In the coming years, DS 2011 and DS 2012 will be organized in Finland and in France, respectively.

The Scope

Currently, there exist various definitions of Discovery Science, which I find natural due to the wealth of the research topic and the long history of the conference series. With respect to my definition, Discovery Science encompasses various research fields such as machine learning, data mining, and knowledge discovery. Restricting it to discovery processes of humans is too narrow minded and against the reality and the history of the DS conferences. I, however, admit placing more emphasis on this aspect would help differentiating the Discovery Science conference from other machine learning/data mining conferences. Diversity should be considered as a wealth of an academic community and a strong driving force for its evolution. Undoubtedly the scope of Discovery Science is dynamic and active participations of excellent researchers are influential.


[1] Setsuo Arikawa: The Discovery Science Project in Japan. Discovery Science 2001: 1-2
[2] Setsuo Arikawa, Ayumi Shinohara: Progress in Discovery Science, Final Report of the Japanese Discovery Science Project Springer 2002
[3] Archives of International Conference on Discovery Science Series, http://www.i.kyushu-u.ac.jp/~suzuki/DSstc.html (as of Nov. 24, 2010).