Skip to content

Time Series Graphs

Yogesh Simmhan edited this page Aug 4, 2016 · 6 revisions

Time Series Graphs

Neo4J DynaGraph (SocioPatterns.org)

  • Presented by Abhilash
  • Frame is a snapshot. Points to next snapshot, and all vertices and edges in the same frame.
  • Vertices and edges modeled as nodes in Neo4J
  • Newo4J relations between edge, its source and sink vertex.
  • Temporal indexing: Tree to help identify the snapshots within a given time range.
  • Types of queries
  • Snapshots in a time range satisfying a graph property
  • Return all vertices that matches a given predicate in a given snapshot
  • Return a subgraph BFS
  • TODO:
  • Additional reading: Data and query model for TS graphs, Timeseries in Titan [AS]

Temporal Graph Analytics (Amol Deshpande)

  • Presented by Abhilash
  • Space: Node, neighbor, graph, size
  • Time: Point, interval
  • Snapshot over time. Key-Val pairs for vertices and edges. Different schemas for different vertices/edges.
  • Node-centric model.
  • Temporal node with many time-specific states for a given vertex.
  • Set of temporal nodes: Temporal nodes within same interval.
  • Temporal Subgraph
  • Dimensions for filtering (Cube)
  • Node (select), Attribute (Filter), Time (Timeslice)
  • Map operation can be applied on temporal node, or a snapshot of the node. one result for template or one result per snapsoht.
  • Evolution, ...
Clone this wiki locally