NTCIR-7 Patent Mining Task CALL FOR PARTICIPATION

Update

• The formal run of the Patent Mining Task will start soon. Details are available at the following page.
  http://www.ls.info.hiroshima-cu.ac.jp/~nanba/ntcir-7/release/formalrun/ (2008/5/3)

Task Overview

Background

The need for academic researchers to retrieve patents and research papers is increasing, because applying for patents is now considered to be an important research activity. In fact, the development of an IR system of research papers and patents for academic researchers is enshrined in " Intellectual Property Strategic Program 2006" by Intellectual Property Strategy Headquarters Japan, published in June 2006.

Searching both research papers and patents is also required for examiners in government patent offices, and for searchers in the intellectual property divisions of private companies. Their particular purpose is to carry out an invalidity search on existing patents or research papers that can invalidate the patents of rival companies or patents under application in a patent office. Therefore, the NTCIR-7 Patent Mining Task aims to develop several techniques for retrieving and classifying research papers and patents.

The task at NTCIR-7

• The purpose is the categorization of research papers (abstracts) written in Japanese or in English into the International Patent Classification (IPC), which is one of patent classification systems.
  * Two tasks (English and Japanese) will be conducted.
• Each group submits a ranked list of the IPC codes for each target research paper.
• Recall and precision are used to evaluate automatic classification methods.

All the participants who attended the previous NTCIR patent classification task can apply their systems to this task with minor modification (F-term -> IPC). The participants who attended the previous NTCIR patent retrieval task can also attend this task with minor modification. (Please see the following steps.)

1. Extract queries from a given abstract, and retrieve related patents using search engines for NTCIR-5 and 6.
2. Extract IPC codes from retrieved patents in step 1.
3. Rank IPC codes by their frequency.

Objective of NTCIR-7 Patent Mining Task

Roadmap

Technical Trend Analysis using Research Papers and Patents

The purpose of this task is to create technical trend maps, as shown in Figure 1, from a set of research papers and patents, which a particular IPC code is assigned. In Figure 1, red and blue character strings indicate research papers and patents, respectively.

Technical Trend Mining Using Citation Relationships between Research Papers and Patents

Recently, many researchers in scientmetrics focus on the technical trend analysis using patents and research papers in terms of measuring the releavnce of basic science to industry. In their works, the citations between patents and research papers are used to analyse their effects of science to technology. In general, if basic research in a domain strongly affects a technology, then it is assumed that patents in that specific domain will cite many papers. Therefore, the researchers measure the relevance of basic research to an industry by counting the number of cited research papers in patents in a given domain. Please refer to the following paper about the detail of the studies in this domain.

Mayer, M. Does Science Push Technology? Patents Citing Scientific Literature. Research Policy. Vol. 29, 409-434 (2000)

On the other hand, it is said that the citations of related research papers in patents are not enough. Therefore, the following steps are required for more reliable technical trend analysis using citations.

1. Categorization of research papers into a patent classification system:
   To categorize research papers into the IPC. (NTCIR-7)
2. Detection of related research papers:
   To detect research papers that should be cited in patents from a set of research papers, which the same IPC codes are assigned in the previous step.
3. Technical trend mining based on the citations between research papers and patents:
   To analyze technical trend using the citation relationships between research papers and patents. The following tasks are assumed: "Analyzing transition of technical trend", "Detecting important techniques in a specific domain", and "Detecting research papers that can invalidate existing claims".

Data used

• Unexamined Japanese patent applications published in 1993-2002
• USPTO patent data published in 1993-2002
• Patent Abstracts of Japan (English translations for Japio patent abstracts) in 1993-2002
• NTCIR-1, 2 CLIR task Test Collection (Abstracts of research papers)

Application Procedure

Application form is available at this page.

Schedule

Preparation	Call For Participation	-2007.11.15
	Data release	As needed.
Dry run	Topic release	2008.01.10
	Submission deadline	2007.02.10
	Evaluation release	2007.03.05
	Analysis of the evaluation	2008.03
Formal run	Topic release	2008.05.01 2008.05.14
	Submission deadline	2008.05.31 2008.06.13
	Evaluation release	2008.06.10 2008.06.23
	Analysis of the evaluation	2008.07
Preparation for meeting	Paper for the Proceedings Due	2008.10.01
	Camera-ready Paper for the Proceedings Due	2008.12.16-19
	Workshop Meeting at Tokyo	2008.12.16-19

  Changes in schedule will be announced via this page and the mailing list for participants.

Organizers

• Hidetsugu Nanba (Hiroshima City University)
• Taiichi Hashimoto(Tokyo Institute of Technology)
• Atsushi Fujii (University of Tsukuba)
• Makoto Iwayama (Tokyo Institute of Technology / Hitachi Ltd.)

Contact

Last modified on May 3, 2008