This research was conducted by Megumi Matsutani, director of ZOZO Research Institute; Shunsuke Awaihara, co-founder and technical director of Hanjuku Virtual and advisor to ZOZO Research Institute; Yuta Saito, co-founder and scientific director; and Yusuke Narita, representative director and assistant professor at Yale University.
Open Source Development & Implementation Platform for Social Decision-Making Algorithms
The open source social decision-making algorithm (Open Bandit Pipeline) is a one-of-a-kind infrastructure implementation that can reliably predict the performance of new measures and algorithms when they are introduced into a real service environment, and can also verify their accuracy. This implementation platform is also being used in the marketing initiatives of ZOZOTOWN, a fashion shopping site operated by the ZOZO Group, and has realized an increase in click and purchase rates. In addition, it has been released free of charge as open source software since August 2020, contributing to the optimization and automation of decision-making in a wide range of areas.
Along with the software, the data collected during the implementation on ZOZOTOWN is available to the public as the Open Bandit Dataset. This data is a large scale data set containing more than 26 million fashion recommendation logs in total, and is a unique data set that can be used in conjunction with the software to verify the performance of various recommendation algorithms.
Reason for Award
In the midst of the recent oligopoly of data and technology in artificial intelligence research by a few large companies in the U.S. and China, this project was recognized for its potential to meet the societal need to publicize and democratize artificial intelligence technology by making its data and software development infrastructure available to practitioners and researchers around the world.
Future Outlook
The award-winning "Open Source Development & Implementation Platform for Social Decision-Making Algorithms" is highly versatile and continues to be improved for more accurate predictions. Through this technology, we plan to continue to release OSS and data with the aim of further developing artificial intelligence technology and helping to address various issues faced when introducing artificial intelligence technology in the real world.
This research is part of a joint research effort between Hanjuku Virtual Corporation and ZOZO Research Institute, and research and development of various other technologies is underway. Semi-mature Virtual Corporation will continue to promote the public and democratization of data science and technology.
About Half-Baked Virtual, Inc.
Semi-Future Virtual Corporation combines data, algorithms, mathematics, and thought to design businesses, policies, and a vision for the future of society. In particular, the company has some of the best people and technology in Japan in "market design," "counterfactual virtual machine learning," and "causal inference," and it conducts joint projects and research with numerous companies as well as its own basic research and software development. These activities are carried out by scientists and engineers with business experience who belong to or have graduated from Yale University, MIT, Stanford University, the University of Tokyo, Tokyo Institute of Technology, Tokyo University of the Arts, and other universities, transcending industry-academia and national boundaries.
| name of company | Half-mature Virtual Co. |
| Establishment | July 28, 2020 |
| Location | 5-5-17 Kugayama, Suginami-ku, Tokyo |
| representative | Representative Director Yusuke Narita |
| Business Overview | Construction, implementation, and propagation of techniques for designing society with data, algorithms, and mathematics. |
About the Japan Open Innovation Awards
As international competition for the creation of innovation intensifies, a major challenge is to quickly implement the results of research and development and other activities in society, and to link them to the resolution of social needs and the creation of new value. As a way to achieve this, open innovation, which promotes new initiatives by combining knowledge, technology, and management resources across organizational boundaries, is attracting attention.
In light of this situation, in order to further promote open innovation in Japan, it was decided to honor initiatives with high leadership and originality that are expected to serve as role models for the future with the "Japan Open Innovation Award.
In this award, the Minister's Prize for each area of responsibility, the Secretary's Prize, and the President's Prize for economic and academic organizations will be awarded for open innovation initiatives that are exemplary, have a large social impact, and are sustainable, and the most outstanding of each award will be recognized as the Prime Minister's Prize. The best of each award will be presented as the Prime Minister's Award.
*Click here for the Cabinet Office website.
https://www8.cao.go.jp/cstp/openinnovation/prize/
Related documents related to this study
For more information on this study, please contact
https://press-tech.zozo.com/entry/20200818_release_zozoresearch
The open source development and implementation platform for social decision-making algorithms can be downloaded from the following link.
Open source social decision-making algorithms
(Open Bandit Data):https://research.zozo.com/data.html
Implementation Infrastructure (Open Bandit Pipeline):https://github.com/st-tech/zr-obp






