temporal graph algebra
play

Temporal Graph Algebra VERA ZAYCHIK MOFFITT JOINT WORK WITH JULIA - PowerPoint PPT Presentation

Temporal Graph Algebra VERA ZAYCHIK MOFFITT JOINT WORK WITH JULIA STOYANOVICH SEPTEMBER 1, 2017 Graph Evolution https://www.kenedict.com/apples-internal-innovation-network-unraveled-part-1-evolving-networks/ 2 Interesting and Important


  1. Temporal Graph Algebra VERA ZAYCHIK MOFFITT JOINT WORK WITH JULIA STOYANOVICH SEPTEMBER 1, 2017

  2. Graph Evolution https://www.kenedict.com/apples-internal-innovation-network-unraveled-part-1-evolving-networks/ 2

  3. Interesting and Important Questions § What is the likelihood of an individual to join a community? § Which roads exhibit abrupt congestion and at what time? § Which websites have the highest increase in popularity/rank over the past year? § What is the rate of densification of the graph? § Have any changes in network connectivity been observed? § At what time scale can interesting trends be observed? 3

  4. Goal Principled and systematic support for querying and analytics of evolving graphs 4

  5. Existing Models – Time as Data Approach – Add time property ◦Need a new node for each change of property or period of validity Are Alice and Bill connected? ! Time needs special treatment 5

  6. Snapshot Reducibility 6

  7. Existing Models – Snapshot Sequence Which pairs of people are connected by a journey? ! No explicit references to time 7

  8. Contributions § Conceptual representation of an evolving graph § Captures evolution of both topology and properties § Temporal Graph Algebra (TGA) § Concisely express wide range of common analysis tasks 8

  9. Temporal Graph Model Definition 3.1.1 (TGraph). A TGraph 𝒣 is a 7- • ρ : E → (V x V ) total function, tuple (V, E, Π , ρ , ξ , λ v , λ e ), where: • ξ 𝑈 : (V ∪ E) x T → B total function, • V – set of nodes, • λ 𝑈 : (V ∪ E) x P x T → Val partial function • E – set of edges, • P – set of available properties, [2015/1,2015/7) [2015/2,2015/5) [2015/2,2015/5) v 1 e 1 v 2 type=co-author type=person type=person cnt=3 name=Alice name=Bob school=Drexel [2015/5,2015/10) [2015/7,2015/10) [2015/1,2015/10) [2015/5,2015/6) v 2 e 1 v 3 e 1 type=co-author type=person type=person type=co-author cnt=4 name=Bob name=Cathy cnt=3 school=CMU school=Drexel 9

  10. TGA Operators § Provide temporal versions of common graph operations: § subgraph § aggregation § vertex- and edge-map § union, intersection, difference 10

  11. Node Creation § attribute-based node creation § add new nodes representing a matching input pattern § window-based node creation § Change temporal resolution of 𝒣 11

  12. Attribute-based node creation § Add new nodes to represent matching pattern x 1 f v ( x 2 ) f e ( x 1 ,studentAt, x 2 ) school= x 2 type=school type=studentAt students = count( x 1 ) adds nodes Drexel and CMU and edges to them 12

  13. Attribute-based node creation [2015/1,2015/7) [2015/2,2015/5) [2015/2,2015/5) v 1 e 1 v 2 type=co-author type=person type=person cnt=3 name=Alice name=Bob school=Drexel [2015/5,2015/10) [2015/7,2015/10) [2015/1,2015/10) [2015/5,2015/6) v 2 e 1 v 3 e 1 type=co-author type=person type=person type=co-author [2015/1,2015/7) cnt=4 name=Bob name=Cathy cnt=3 e 3 school=CMU school=Drexel type=studentAt [2015/7,2015/10) e 4 [2015/5,2015/10) [2015/1,2015/7) type=studentAt e 5 e 4 type=studentAt type=studentAt [2015/1,2015/7) [2015/5,2015/10) [2015/7,2015/10) Drexel CMU Drexel type=school type=school type=school students=2 students=1 students=1 13

  14. Window-based node creation 3 months [2015/1,2015/7) v 1 type=person name=Alice school=Drexel [2015/4,2015/10) [2015/7,2015/10) [2015/1,2015/10) [2015/4,2015/7) v 2 e 1 v 3 e 1 type=co-author type=person type=person type=co-author cnt=4 name=Bob name=Cathy cnt=3 school=CMU school=Drexel 14

  15. Example: NYC Cabs 15

  16. Node Influence over Time Are there high influence nodes and is that behavior persistent over time? 16

  17. Node Influence, with TGA 1. Select a subset of the data representing the 5 years of interest, using trim: 2. Compute in-degree (prominence) of each node during each time point using aggregation and pattern p1 x 1 x 2 x 3 deg=count( x 2 ) 17

  18. Node Influence, with TGA 3. Aggregate degree information per node across the timespan of G2 using the window-based node creation operator: 4. Transform the attributes of each node using the vertex-map operator: 18

  19. Summary § TGraph model represents evolution of graph topology and properties § TGA provides a concise set of operations over TGraphs § Precise semantics § More expressive than current state of the art § Desirable temporal properties 19

  20. Thank You! Questions? 20

Recommend


More recommend