BDA Europe 2016

BDA Europe 2016

Support Your Users with a Neo4j Recommender System presentation @ Big Data Analytics Europe 2016

1ea294b37fce89b0bb07beb7526574c7?s=128

Maurits van der Goes

March 22, 2016
Tweet

Transcript

  1. 1.

    Support Your Users with a Neo4j Recommender System Maurits @vanderGoes

    Tuesday 22 March 2016 - Big Data Analytics Europe 2016
  2. 2.
  3. 4.

    “Virtual workers are often more productive and more likely to

    remain with the organization.” (Thompson & Caputo, 2009)
  4. 5.
  5. 6.
  6. 8.

    Architecture API (Scala) Platform (Meteor) Recom. engine (GraphAware) Recom. DB

    (Neo4j) Importer (Java) Platform DB (MongoDB) Hybrid Hybrid DB
  7. 9.

    Logical Data Model in Neo4j User Network Team City Country

    Strength ACTIVE_IN LIVES_IN LOCATED_IN LOCATED_IN MEMBER_OF PART_OF LOCATED_IN HOLDS
  8. 10.
  9. 11.

    GraphAware setup Recommendation engine Algorithms that discover relevant teams Post

    processor Reward & penalize the team score on features (similarities, size, etc) Config Maximum number of recommendations Filter Rules of the platform Blacklist Current relations Logger Terminal loggers with recommendation details & statistics
  10. 16.

    Item-based collaborative filtering algorithm 4,5 2,3 3,7 2,6 3,9 4,1

    Some relations are not shown for the visibility
  11. 17.

    Item-based collaborative filtering algorithm 4,5 2,3 3,7 2,6 3,9 4,1

    Some relations are not shown for the visibility 0,87321 Pearson correlation-based similarity
  12. 18.

    Item-based collaborative filtering algorithm 4,5 2,3 3,7 2,6 3,9 4,1

    Some relations are not shown for the visibility 0,64962 Pearson correlation-based similarity
  13. 19.

    Item-based collaborative filtering algorithm 4,5 2,3 3,7 2,6 3,9 4,1

    Predicted participation score: 3.6 Some relations are not shown for the visibility 0,87321 0,64962
  14. 20.

    A RDBMS vs Neo4j example Depth RDBMS execution time(s) Neo4j

    execution time(s) Records returned 2 0.016 0.010 ~2.500 3 30.267 0.168 ~110.000 4 1543.505 1.359 ~600.000 5 Unfinished 2.132 ~800.000 (Vukotic et al., 2014)
  15. 24.

    Visit Part-up.com Speaker Deck: vdgo.es/BDAE16 Icon credits (the Noun Project):

    Aha-Soft, Akshay Kore, Alfredo Hernandez, Aneeque Ahmed Creative Stall, Gregor Črešnar, Luis Prado, icon 54, Iconathon, Kevin Augustine LO, Klara Zalokar, Magicon, Matt Hawdon, Muneer A.Safiah, Nono Martínez Alonso, Simple Icons, Wilson Joseph Questions: @vanderGoes & m@vdgo.es