Dyads on Twitter

NLDB 2019 Slides

Transcript

1. Analysis and Prediction of Dyads in Twitter IsaInuwa-Dutse MarkLiptrott YannisKorkontzelos Department of Computer Science March ,
2. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Outline Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion /
3. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Social Network & Relationships useful for linking micro and macro levels of social ties [ ] /
4. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Social Network & Relationships useful for linking micro and macro levels of social ties [ ] Sociometry /
5. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Social Network & Relationships useful for linking micro and macro levels of social ties [ ] Sociometry and quantification of social ties /
6. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Social Network & Relationships useful for linking micro and macro levels of social ties [ ] Sociometry and quantification of social ties function of closeness and size ... /
7. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Social Network & Relationships useful for linking micro and macro levels of social ties [ ] Sociometry and quantification of social ties function of closeness and size ... social support clique [ ] (see Fig. , based on [ ]) /
8. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Social Network & Relationships useful for linking micro and macro levels of social ties [ ] Sociometry and quantification of social ties function of closeness and size ... social support clique [ ] (see Fig. , based on [ ]) /
9. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion ...taxonomy Figure: Classification of social groups and the degree of cohesiveness /
10. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion ... social cohesion OSMPs – enable validation of social theories /
11. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion ... social cohesion OSMPs – enable validation of social theories Difficult to detect cohesive groups/content veracity /
12. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion ... social cohesion OSMPs – enable validation of social theories Difficult to detect cohesive groups/content veracity ... many event-typeties /
13. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion ... social cohesion OSMPs – enable validation of social theories Difficult to detect cohesive groups/content veracity ... many event-typeties (see Fig. ) /
14. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Connections on Twitter (a) dyads /
15. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Connections on Twitter (a) dyads (b) RT and hashtag /
16. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Connections on Twitter (a) dyads (b) RT and hashtag (c) @mention /
17. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Connections on Twitter (a) dyads (b) RT and hashtag (c) @mention (d) Relative proportion Figure: Connection types on Twitter Homophily, central to human's social interaction[ ] /
18. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Connections on Twitter (a) dyads (b) RT and hashtag (c) @mention (d) Relative proportion Figure: Connection types on Twitter Homophily, central to human's social interaction[ ] ... studied at various levels – location/popularity [ ], follower–following [ ] /
19. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Connections on Twitter (a) dyads (b) RT and hashtag (c) @mention (d) Relative proportion Figure: Connection types on Twitter Homophily, central to human's social interaction[ ] ... studied at various levels – location/popularity [ ], follower–following [ ] Explore dyads or reciprocal ties ... /
20. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Dataset & Model dyadic tie: a relation R over a set D is dyadic iff aRb = , ∀a, b ∈ D, see Fig. github.com/ijdutse/dyads_in_Twitter /
21. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Dataset & Model dyadic tie: a relation R over a set D is dyadic iff aRb = , ∀a, b ∈ D, see Fig. Figure: Pair of connections github.com/ijdutse/dyads_in_Twitter /
22. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Dataset & Model dyadic tie: a relation R over a set D is dyadic iff aRb = , ∀a, b ∈ D, see Fig. Figure: Pair of connections queried over M accounts and retrieve dyadic ties github.com/ijdutse/dyads_in_Twitter /
23. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Dataset & Model dyadic tie: a relation R over a set D is dyadic iff aRb = , ∀a, b ∈ D, see Fig. Figure: Pair of connections queried over M accounts and retrieve dyadic ties (see ) criteria: user's network G to determine reciprocal ties: G = {u|∃u ∈ G} s.t. u ∩ u = github.com/ijdutse/dyads_in_Twitter /
24. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Dataset & Model dyadic tie: a relation R over a set D is dyadic iff aRb = , ∀a, b ∈ D, see Fig. Figure: Pair of connections queried over M accounts and retrieve dyadic ties (see ) criteria: user's network G to determine reciprocal ties: G = {u|∃u ∈ G} s.t. u ∩ u = Table: Many directed connections i.e. ∃a, b ∈ D, a −→ b = and b −→ a = . Category Seed Size Visited Users Retrieved Remark Unverified dyads , , , , utilised for prediction Verified dyads , , , – not used for prediction -edge and null tie , – , utilised for prediction github.com/ijdutse/dyads_in_Twitter /
25. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Empirical Analysis dyads in % and % of unverified and verified profiles respectively (Fig. ) /
26. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Empirical Analysis dyads in % and % of unverified and verified profiles respectively (Fig. ) Figure: Proportion of dyads ... fewer dyads in a random collection /
27. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Empirical Analysis dyads in % and % of unverified and verified profiles respectively (Fig. ) Figure: Proportion of dyads ... fewer dyads in a random collection proposed a dyad prediction model (Fig. ) /
28. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Empirical Analysis dyads in % and % of unverified and verified profiles respectively (Fig. ) Figure: Proportion of dyads ... fewer dyads in a random collection proposed a dyad prediction model (Fig. ) /
29. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Raw data /
30. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Prediction Framework (a) embedding /
31. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Prediction Framework (a) embedding (b) convnetoperation /
32. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Prediction Framework (a) embedding (b) convnetoperation (c) encoding & FCN Figure: Prediction pipeline /
33. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Result Figure: prediction performance /
34. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter /
35. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups /
36. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups ... content veracity /
37. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups ... content veracity Conducted an empirical analysis to understand dyads on Twitter /
38. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups ... content veracity Conducted an empirical analysis to understand dyads on Twitter Dyadsutility: improve detection tasks/content veracity /
39. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups ... content veracity Conducted an empirical analysis to understand dyads on Twitter Dyadsutility: improve detection tasks/content veracity prediction model /
40. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups ... content veracity Conducted an empirical analysis to understand dyads on Twitter Dyadsutility: improve detection tasks/content veracity prediction model .... and large scale datasets consisting of dyadic ties /
41. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Conclusion Porous connections on Twitter .... challenge detection of cohesive groups ... content veracity Conducted an empirical analysis to understand dyads on Twitter Dyadsutility: improve detection tasks/content veracity prediction model .... and large scale datasets consisting of dyadic ties future: focus on transitiveties and clustering framework based on state-typeties /
42. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Reference See the full paper for further information and details about the references: /
43. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Reference See the full paper for further information and details about the references: Inuwa-Dutse, I., Liptrott, M. and Korkontzelos, Y., , June. Analysis and Prediction of Dyads in Twitter. InInternationalConferenceonApplicationsofNaturalLanguagetoInformation Systems (pp. - ). Springer, Cham. /
44. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Thank you /
45. Social Relationships Social Media Ecosystem Dyads Analysis & Prediction Result & Conclusion Thank you /