Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Financial Security Scoring Service Responsibility and future

Financial Security Scoring Service Responsibility and future

LINE Developers
PRO

May 29, 2019
Tweet

More Decks by LINE Developers

Other Decks in Technology

Transcript

  1. Intertust x LINE Security Summit 2019, Spring Financial Security Scoring

    Service Responsibility and future May 29th, 2019 K.Sekimizu LINE/LVC/LINE Credit/LINE証券準備設⽴会社/LINE Financial Security-team
  2. Summary of abstraction in this speech Digital Identity E-KYC FIDO

    OpenID Connect Web Service Provider Data Business Provider As Security - Fraud Detection System As Business - Scoring Strategy User Privacy
  3. Todayʼs Agenda About Scoring Service Implementation & issues Our responsibility

    & proposal Conclusion
  4. Chapter1 About Scoring Service

  5. Why? Why do we(LINE) provide Scoring Service? As-Is To-Be Un-fair

    • Discrimination by • Nationality • Family Status • Gender Identity • Not Fun, Serious • Educational/Business Background • Income Fair • Current Status by • Service behavior • Based on individual • Fun • via experience of LINE Service Positive
  6. Whatʼs ʻScoringʼ? Define: Give some ʻIncentiveʼ to user based on

    behavior of using service User Service Sign-up Check Communicate Buy Sell User Behavior Scoring Service Service Service ref Incentive Architecture:
  7. History How calculate ʻscoreʼ? Proposition & Process Process exmaple: Proposition

    What kind of person do they drive safety ? List of Users User Behavior Make a model based on user behavior model User Behavior Deploy Score Service ref Matching Objective Scoring Service
  8. Chapter2 Implementation & issues

  9. Rule matching vs Machine Learning Proposition What kind of person

    do they drive safety ? Method Easy to Understand Explainability Novelty/Treasure Fair & Fun Rule Matching Machine Learning • Old • History of the accident • etc • User behavior x teacher model • Good • Bad • ? • Bad • Good • ? > < <>? Lead sentence
  10. Rule matching Machine Learning Score Service ref • Easy to

    understand / explainablity • Novelty/Treasure • Fairness & Usability Proposition Explaninability ? ? Lead sentence Balancing between methods
  11. Lead sentence Service As Security Service Shopping Service Marriage agency

    Score Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Multi service & multi proposition
  12. Lead sentence Service As Security Service Shopping Service Marriage agency

    Score Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning mismatching Multi service & multi proposition
  13. Lead sentence Service As Security Service Shopping Service Marriage agency

    Score Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Proposition Rule matching Machine Learning Selectable Explainable Multi service & multi proposition
  14. Chapter3 Our responsibility & proposal

  15. xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx

    『AIと憲法』⼭本⿓彦 ⽇本経済新聞出版社 ,2018から引⽤ Why MUST avoid pessimistic story/scenario Story
  16. Level2 Considerations Ensure modelʼs Explainability & can not balancing also

    service must understand the above as a promise Providers should self weight Monopoly do-not-use should not overtake individual Implementation: Give a method Opt-in/Opt-out for each Logic (not attribute) to users.
  17. Implementation User Scoring Service Scoring Service Scoring Service Scoring Service

    Scoring Service Scoring Service Lead sentence Allow Allow Allow Allow Deny Deny
  18. Implementation User Scoring Service Service Need to get agreement by

    user with appropriately as same as at yet. User can select Allow/Deny to provision a score to each service OAuth2.0/OIDC OAuth2.0/OIDC User logs. User score. Lead sentence
  19. Level1 Considerations Ensure Digital Identiy Prepare & compliance to privacy

    policy Implementation: E-KYC, FIDO, OpenID Connect , OAuth, Inhouse platform & operations
  20. Implementation Lead sentence KYC FIDO etc. OIDC/OAuth Keep assurance of

    ID Proofing Keep assurance of Authentication Keep assurance of Federation Keep High Assurance of user behavior It can be used/ scoring, more
  21. Finally Conclusion

  22. Time frame Level1 Considerations Level2 Considerations now going… Just started.

    We are now, here. Web Service Provider Data Business Provider
  23. Thanks !