Visual Analytics for R&D Intelligence @Funding the Commons & DeSci Tokyo 2024

This is my presentation slide in Funding the Commons 2024.
https://www.fundingthecommons.io/tokyo-2024

Talk summary is the follows:
R&D intelligence aims to gain insights for decision-making in R&D strategy. To achieve effective R&D intelligence, it is important to utilize science and technology (SciTech) information, such as academic papers and patents. SciTech information is big data that contains research topics, researchers, organizations, collaborations, and knowledge linkages. Therefore, data technologies such as data mining and visualization play an essential role. In fact, VALUENEX Japan Inc. develops and provides data analytics tools. Additionally, I have been implementing various data analytics projects for companies and public institutions. These projects include R&D trend analysis, researcher discovery, and exploring new business opportunities. In this talk, I will show visual analytics methods for R&D intelligence and examples of insights through several case studies. One of the case studies is "Panoramic View Analysis of Academic Papers on Decentralized Science." I hope this talk will contribute to future discussions about the utilization of new data arising from changes in the research ecosystem.

Transcript

1. Visual Analytics for R&D Intelligence Takayoshi HAYASHI VALUENEX Japan Inc. Advanced Informatics Research July 24, 2024 Funding the Commons Tokyo & DeSci Tokyo 2024
2. 1 We create “Intelligence” from Big Data around the world 世界に氾濫する情報から「知」を創造する R&D Intelligence • We provide insights for decision-making in R&D strategy to our clients. • Mainly, we analyze science, technology and startup information. Clients • Company: R&D, Intellectual Property, Business Development, Management Divisions etc. • Public Institution: JST, AMED, NEDO, AIST, NICT, NISTEP, JPO, Japan Ministry of Defense etc. Services • Data Analytics Tools • Research & Consulting
3. 2 Science and Technology information contains various contents such as research topics, researchers, organizations, countries, collaborations, and knowledge linkages. Academic Paper https://patents.google.com/patent/JP2015065740A/en https://onlinelibrary.wiley.com/doi/epdf/10.1002/eej.22666 Patent
4. 3 Science and Technology information has become “Big Data.” Total (1990-2023) Scopus：75,000,000 / Lens：220,000,000
5. 4 Visual Analytics for R&D Intelligence : Panoramic View Analytics R&Dインテリジェンスのためのビジュアルアナリティクス：俯瞰解析
6. 5 A list display of search results may result in opportunity loss of knowledge discovery. Search results list of ”Decentralized Science” in Lens.org We can not read all. There are 1300 papers. So, we tend to focus on the top items. We can not understand easily the relationship between neighboring papers. https://link.lens.org/uOndk45Qb4 Search conditions (a) ”Decentralized Science“ in All Fields：49 papers (b) Cited by (a)：1,072 papers (c) Citing (a)：286 papers (d) Total：1,371 papers published by June 2024 We are looking at only a portion of the information without understanding the overall picture. Opportunity loss of knowledge discovery ① ②
7. 6 Visual Information Seeking Mantra Overview first, zoom and filter, then details-on-demand Gain an overview of the entire collection. Zoom in items of interest. Filter out uninteresting items. Select an item or group and get details when needed. https://www.cs.umd.edu/~ben/papers/Shneiderman1996eyes.pdf Basic principle for visual information seeking. It is proposed by Ben Shneiderman in 1996. * Ben Shneiderman is Distinguished Professor in the University of Maryland. His research field is human–computer interaction. We have been proposing visual analytics for document information. Panoramic View Analytics (俯瞰解析)
8. 7 Panoramic View Analytics: Overview first, zoom and filter, then details-on-demand In academic research on DeSci, there are several areas such as ”Computing / Machine Learning”, ”Blockchain / Metaverse”, ”Science”, ”Applications.” Science Issue Applications (Healthcare, Agriculture, Environment etc.) Computing / Machine Learning Blockchain / Metaverse 1 plot ＝1 paper Distance among plots is similarity among documents. Similar documents are plotted nearly. Dissimilar documents are plotted far away. X and Y axes are not predefined. Plot position is calculated optimally to represent similarity among documents. • We can understand overview of academic research. • If we compare own knowledge, we can discover new perspectives.
9. 8 Panoramic View Analytics: Overview first, zoom and filter, then details-on-demand (a) Papers including DeSci (b) Papers cited by (a) (c) Papers citing (a) We can zoom in interest perspective. The differences between papers including DeSci and cited/citing papers are as follows: (a) Peer review, Blockchain, DAO, Metaverse (b) Advanced ICT (Federated learning, ChatGPT, Metaverse, Blockchain), Specific domain (Medical, Agriculture, Vehicle) (c) Metaverse, DAO, Parallel system, Blockchain, Peer review Blockchain Metaverse DAO Peer review ChatGPT Metaverse Federated learning Blockchain Cancer Blockchain Metaverse DAO Parallel system Peer review
10. 9 Panoramic View Analytics: Overview first, zoom and filter, then details-on-demand Fei-Yue Wang Reza M. Parizi We can understand the fields of each researcher. • Fei-Yue Wang (Chinese Academy of Science): Parallel system, DAO • Reza M. Parizi (Kennesaw State Univ.): PI of Decentralized Science Lab. Federated learning, Blockchain. He uses the term “Decentralized Science” in the context of distributed computing. • Hiro Taiyo Hamada (Araya / DeSci Tokyo): Peer review. He proposes new incentive mechanism of peer review. Hiro Taiyo Hamada Parallel system DAO Federated learning Blockchain Peer review
11. 10 Panoramic View Analytics: Overview first, zoom and filter, then details-on-demand Fei-Yue Wang If we are interested in Prof. Fei-Yue Wang, we can know his activities in detail. • Parallel system: Cyber-Physical-Social System, Enterprise Management by using DAO, Metaverse and CPSS etc. • DAO: Proposal of DeSci mechanism, Governance of DAO etc. ▪ Keywords parallel, physical, system, cyber, intelligence, social, cpss, knowledge, intelligent, dao ▪ Papers list • Metaverses and DeMetaverses: From Digital Twins in CPS to Parallel Intelligence in CPSS • A Novel DAO-Based Parallel Enterprise Management Framework in Web3 Era • Parallel Driving with Big Models and Foundation Intelligence in Cyber–Physical– Social Spaces Parallel system ▪ Keywords intelligence, decentralize, organization, parallel, daos, autonomous, dao, governance, future, organizational ▪ Papers list • A New Architecture and Mechanism for Decentralized Science MetaMarkets • A Novel Approach for Predictable Governance of Decentralized Autonomous Organizations Based on Parallel Intelligence • Parallel Governance for Decentralized Autonomous Organizations enabled by Blockchain and Smart Contracts DAO Parallel system DAO
12. 11 Reverse engineering of emerging concept • We can clarify the related topics of DeSci by applying panoramic view analytics to papers including DeSci and citation papers. • Anyone can easily and quickly grasp the overview and trend. Decentralized Science Blockchain Computing Science Applications • Blockchain • DAO • Parallel system • NFT • Metaverse • Federated Learning • Edge Computing • ChatGPT • Peer review • Scientometrics • Medical / Bio • Vehicle • Agriculture • Environment Research category Related tech/topic Elemental tech/topic Emerging concept We can obtain more detailed information by using additional reference papers.
13. 12 Case Studies of Companies 企業の活用事例
14. 13 Case Study: New business creation by Meiji Holdings ※Meiji Holdings Co.,Ltd., “Integrated Report 2023”, p61 https://www.meiji.com/global/investors/results-presentations/integrated-reports/pdf/2023/integrated-reports_2023_en_all.pdf They wanted to develop new cocoa bean products. Therefore, they applied panoramic view analytics to patent information on cocoa bean and chocolate. As a result, they could identify R&D field they should approach.
15. 14 Case Study: Identifying synergy among companies by Asahi Kasei ※ Asahi Kasei Corporation, “Intellectual Property Report 2023”, p7 https://www.asahi-kasei.com/r_and_d/intellectual_asset_report/pdf/ip_report2023e.pdf In 2023, Asahi Kasei and Mitsui Chemicals established a joint venture in the nonwoven fabrics industry (不織布). Before this decision-making, Asahi Kasei applied panoramic view analytics to patents of both companies. As a result, they could identify technology synergy area.
16. 15 Case Study: Research & Business Development by ENEOS / Preferred Networks ※ VALUENEX web site https://www.valuenex.com/jp/news-list/2023/global-internship-program-supports-eneos VALUENEX has Silicon Valley office. We are promoting the research & business development with client companies and university students such as Stanford University and UC Berkley.
17. 16 Change of Research Ecosystem and Data Technology 研究エコシステムの変化とデータテクノロジー
18. 17 Change of Research Ecosystem and Science and Technology information • Research ecosystem will expand by the movement like DeSci etc. As a result, people who join in research process will diversify. For example, citizen, foundation etc. • Scholarly communication is evolving. For instance, research data and preprints may be published prior to the formal academic paper. Additionally, research visions and progress could be shared with the public during the early stages of the R&D process. Research Grant Research Activity Commercialization Use in Society Paper, Patent Grant Product, Service Government Academia Company Citizen Crowdfunding, DAO, Donation Research Vision, Progress Voice of Customer Research Data, Preprint Open Review +Foundation Action Information Startup New! New!
19. 18 Science and Technology Information for Citizen • I am interested in creating an environment where everyone can access, understand, and utilize diverse research information. • Therefore, I would like to contribute to develop data technology for the above purpose. Science and Technology Information for … Scientist / Engineer Business / Policy Making Citizen
20. Contact Takayoshi Hayashi VALUENEX Japan Inc. Advanced Informatics Research email: hayashi@valuenex.com X: @hayataka88 Thank you so much!
21. 20 Appendix
22. 21 Panoramic View Analytics: Algorithms クラスターA クラスターB クラスターC Cluster 1 Cluster 3 Cluster 2 Vectorizing documents by combination of words Location in high-dimensional space 2D Visualization Calculating TF-IDF (Term Frequency - Inverse Document Frequency). This is important score representing frequency and bias. The below chart is represented by 3- dimensional space. But, in fact, there are tens of thousands of dimensions. The relationship among documents is compressed from high dimensions to two dimensions. ※ There are many vectorization methods like topic model, doc2vec etc. ※ There are many dimension reduction methods like multi dimensional scaling, principal component analysis, T-SNE, UMAP etc. After vectorizing each document, the relationship among documents is compressed from high dimensions to two dimensions. Word 1 Word 2 Word 3 ・・・ Doc 1 0.4 0 0.2 Doc 2 0.5 0 0.1 Doc 3 0 0.1 0.5 ・・・
23. 22 Science and Technology information is important source including human knowledge. https://scholar.google.com/schhp ← Newton also used this phrase.
24. 23 Overview first, zoom and filter, then details-on-demand Before 2017以前 2018-2020 2021-2024.6 Time series change • Before 2017: Peer review, Scientometrics, Specific domains (Vehicle, Medical, Agriculture etc.) • 2018-2020: Federated Learning, Blockchain • 2021-2024.6: Federated Learning, Blockchain, ChatGPT, Metaverse, NFT, DAO