Trends and Issues of Web Service Development Using Social Network Analysis

■イベント 
「第31回獨協インターナショナル・フォーラム」
https://www.dokkyo.ac.jp/international/international_center/forum/2019.html

■登壇概要
タイトル：Trends and Issues of Web Service Development Using Social Network Analysis
発表者： 
DSOC 研究開発部 SocSci Group 前嶋 直樹 / Naoki Maejima

▼Twitter
https://twitter.com/SansanRandD

Transcript

1. Trends and Issues of Web Service Development Using Social Network Analysis Naoki Maejima Sansan, Inc. DSOC R&D group
2. ※ 掲載されている内容等は発表時点の情報です。 ※ 公開に当たり、資料の⼀部を変更・削除している場合があります。
3. Data Strategy and Operation Center About me Naoki Maejima • Researcher, Sansan, Inc. DSOC R&D • PhD candidate, The University of Tokyo • Twitter: @naoki_maejima
4. Data Strategy and Operation Center Today’s agenda • What/Who is Sansan ? • Our Research • Applying SNA to Web Services Development • Trends • Our products • Businessperson type analysis • Know-who search (upcoming) • Issues
5. What / Who is Sansan?
6. Data Strategy and Operation Center Our mission and products Mission • Turning encounters into innovation Products … centered on business cards • Sansan (B2B app suite) A cloud-based contact management tool for companies, centered on scanning business cards • Eight (B2C app) A contact management app and business social network (small similarity to LinkedIn) You can download it right now, for free!
7. Data Strategy and Operation Center Four characteristics of business card exchange data • Universality, ubiquity (in Japan) Business card exchange is a common business practice in Japan, in all industries. • Accuracy A business card is an official medium that conveys the contact info of individuals and companies; it’s highly unlikely to contain false/inaccurate information. • Real-time Business card scan date is recorded, or users can set exchange date. • Accompanies face-to-face interaction Since business cards are paper-based media, conversations or other face-to-face interactions can be assumed to be behind the exchange.
8. Data Strategy and Operation Center DSOC SocSci team • DSOC (Data Strategy & Operation Center) R&D Group • SocSci team – Researchers trained in social sciences • Economics • Sociology • Management science • Computational social sciences • Each business card exchange makes a large social network. → Therefore, a social science approach should be effective. • We do both empirical research and web service development.
9. Our Research
10. Data Strategy and Operation Center Our research themes • Inter-organizational network and corporate branding • Based on BBES (B2B Brand Engagement Score) survey • Heterogeneity of network formation among industries (upcoming) • Conditions for “magical encounters” … and much more! *Anonymize individuals among Eight data and use registered business card information and profiles within the scope approved to in Eight’s Terms of Use.
11. Data Strategy and Operation Center Ongoing project@NetSciX2020 • Examining effect of temporal pattern of dyad-wise shared partner growth on success of encounters on tie strength • Time-series clustering method (dynamic time warping + k-means) *Anonymize individuals among Eight data and use registered business card information and profiles within the scope approved to in Eight’s Terms of Use. • We found 3 patterns: • Gradual ← most likely to strengthen ties • Bursting • Potential
12. Data Strategy and Operation Center Sansan Data Discovery (a joint research platform) Collaborative research with leading outside experts. Is the concept of “weak ties” valid in respect to career changes? w/ Habitech, Inc.; University of Illinois at Urbana-Champaign Is the labor market becoming fragmented? w/ Keio University; The University of Tokyo; Yale University; The Institute of Statistical Mathematics Turning the formation of personal networks into a science w/ The University of Tokyo How are life-changing encounters created? w/ Hosei University Research themes
13. Data Strategy and Operation Center “Dawn of Innovation” (NetSci 2019 Visualization Prize) w/ Qosmo, Inc.; JEMAPUR; Habitech, Inc
14. Applying SNA to Web Services Development
15. Data Strategy and Operation Center The rise of people analytics “People analytics” is now prevalent (Waber, 2013) . • Realtime workplace data collected using wearable sensors to measure and improve productivity Organizational network analysis (ONA) is attracting attention (Leonardi and Contractor 2018) . • Examples of ONA goals • Discovery & integration of closed teams (silos) that lack external contact • Identification of influencers • Identification of individuals and teams with new ideas • Leadership assessment • Evaluation of personnel allocation mismatch • Diversity and inclusion assessment
16. Data Strategy and Operation Center Tools and services applying SNA Tools • OrgMapper • HOW4 • OrgAnalytix • Polinode • Socilyzer • Openteams Consulting service • Complete Coherence • Social network laboratory …and more
17. Data Strategy and Operation Center Businessperson Type Analysis
18. Data Strategy and Operation Center Businessperson Type Analysis • Showing user’s networking type with a radar-chart • Five scores (intra-organizational index and inter-organizational index) • Keyperson: degree centrality • Teamwork: network closure (local clustering coefficient) • Executive: position rank of alters • Innovator: variety of industries of alters • Pioneer: rarity of business cards
19. Data Strategy and Operation Center Backend: bipartite projection Building 2-mode user-card network to obtain 1-mode user–user network (Breiger, 1976) User Business card
20. Data Strategy and Operation Center What type of relationship does it capture? 18 • Comparing network from co-ownership of business-cards (left) and network from event co- occurrence of Google Calendar in Sansan, Inc. (right) • Slicing network by event category (e.g., discussion, moving from place to place, dinner) of network based on the calendar to examine which network resembles the network based on business card co-ownership. • Index: • Graph correlation (with QAP test) • Edge overlap ratio • Using top 5,000 edges of both networks • Based on standardized residuals of matrix
21. Data Strategy and Operation Center Results 19 Network based on business card co- ownership relatively captures the official collaborative relationship (e.g., customer service) rather than an informal one (e.g., lunch or dinner). Category of relationship Graph correlation Edge overlap ratio Customer services 0.27*** 0.31 Discussion 0.26*** 0.31 Moving 0.27*** 0.31 Regular meeting 0.12*** 0.18 HR 0.14*** 0.19 Deskwork 0.11*** 0.17 Lunch or dinner 0.11*** 0.17 Study meeting 0.17*** 0.22 Other 0.23*** 0.28 Total 0.25*** 0.29 Q *** p < 0.0001 with QAP test
22. Data Strategy and Operation Center In the works: Know-who search Searching for members who have specific knowledge in your organization, based on business cards • e.g., If you’ve acquired business cards from Google, AWS, Preferred Networks, etc., we infer you’re involved with artificial intelligence. • Knowledge and information flow via connections (major premise of SNA). User’s neighbors connected with the suggested member are also shown. • This leads to close triads (transitivity) and is predicted to decrease the cost to contact the suggested member. Our Manager ! Know-who search (α version)
23. Data Strategy and Operation Center Some issues Network visualization, a strength of SNA, is hard to interpret for ordinary users. • > “Some visualizations are totally useless (the "network" hairball that people still draw for large networks, just to prove that they can), and nothing works for everything.” (Martin, 2018) Customers need just a prediction or decision, not hints for decision-making. • “SNA seems to tell us many things…but so what?” Customers need formal organization charts, rather than informal sociometrics. • They feel the result is obvious without technology. • We must make public and clarify the value SNA offers!