Ted Willke, Senior Principal Engineer & GM, Datacenter Group, Intel
summary
Demo compares 1,300 molecules visually but notes the n^2 problem for millions of compounds, argues for treating graphs as a data-structure primitive by reusing existing tooling, and shows two concrete directions: integrate graph processing into GraphX on Spark with Python bindings, a REST “data science API,” and a pandas‑like Intel toolkit that builds and runs loopy belief propagation at scale (demo claims a 15 billion node graph and SVD++/PageRank support, runs on a Hillsboro Spark cluster); or embed graph traversal and analytics inside relational engines (SAP HANA and MIT ISTC work shows PageRank/shortest‑path faster than Giraph).
AI-written summary · about
chapters (8)
- 0:00Graphs as big data primitives
- 3:12Graphs in analytics
- 5:20Integrating graph analytics
- 7:09Data science pipelines with graphs
- 11:00Unifying primitives with Spark
- 16:58Graph analytics in SQL
- 23:13Intel bridging graph and big data
- 27:04Demo: Intel’s graph toolkit on Spark
related talks
25mNarayanan Sundaram, Research Scientist, Intel Labs
30mGraph Data Science
18mEhtsham Elahi, Senior Research Engineer, Netflix @ MLconf SEA
50mBig Data on Smallish Machines
49mModern Graph Databases
43mGraphs: Datastructures to Query
38mA Tale of Two Graph Engines on Spark
27mData @scale Seattle - Yucheng Low