| Project/Area Number |
16300047
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | National Institute of Advanced Industrial Science and Technology |
|---|
Principal Investigator |
IZUMI Kiyoshi National Institute of Advanced Industrial Science and Technology, Digital Human Research Center, Senior Researcher, デジタルヒューマン研究センター, 主任研究員 (10356454)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
MATSUO Yutaka National Institute of Advanced Industrial Science and Technology, Information Technology Research Institute, Researcher, 情報技術研究部門, 研究員 (30358014)
|
|---|
| Project Period (FY) |
2004 – 2006
|
| Project Status |
Completed (Fiscal Year 2006)
|
| Budget Amount *help |
¥12,500,000 (Direct Cost: ¥12,500,000)
Fiscal Year 2006: ¥2,700,000 (Direct Cost: ¥2,700,000)
Fiscal Year 2005: ¥4,700,000 (Direct Cost: ¥4,700,000)
Fiscal Year 2004: ¥5,100,000 (Direct Cost: ¥5,100,000)
|
|---|
| Keywords | Artificial market / Text mining / Energy service market / Decision support system / Multi-agent system / Social simulation / Risk assessment / Complex systems |
|---|
| Research Abstract |
This research project obtained the following results in three research elements; the trend analytical method by text mining; artificial market simulation software; system evaluation by actual data. These results were published in international journals. For the purpose of development of this field, we have started a new international workshop.
1. Trend analytical method by text mining: We built a text-mining program that produced input data of an artificial market simulation using the categorization of a related term and classification of frequency and co-occurrence of words and categories. As a test of our proposed method, economic trends were estimated from market reports that Japan Center for International Finance published. These results were compared with financial professional's judgments using the same text data. As a result, our method could correctly estimate 92.2% of training data and 71.9% of out-of-sample data on average.
2. Artificial market simulation: We built an artificial market simulation program that received economic trend data. And we developed a method that integrated text-mining and artificial market simulation program. We tested the simulation program using actual text data and market data. As a result, about 67% of price fluctuation could be simulated on the average of 100 trials.
3. System evaluation by actual data: The decision support system that integrated the above-mentioned text-mining programming and the artificial market simulation was built. Using actual text data and market data, we tested the behavior strategy for market stabilization that the system proposed. As a result, distribution of the market price could be reduced over 70% by the proposed strategy.
|
