Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
JOBANDTALENT AT
RECSYS CHALLENGE 2016
RECSYS CHALLENGE 2016
LEARNING TO RANK IN RECSYS 2016
▸ Learning to Rank Framework
▸ Listwise, Pairwise, Pointwise
▸ Relev...
RECSYS CHALLENGE 2016
DATA EXPLORATION
▸ Understanding data.
▸ Probability of Item being interacted "Popularity".
▸ Probab...
RECSYS CHALLENGE 2016
RETRIEVERS
▸ Job title/Job tag match
▸ Relevant items per interaction
▸ Interactions made by user
▸ ...
RECSYS CHALLENGE 2016
COLLECTORS
▸ Freshness.
▸ Item popularity.
▸ User behavior.
RECSYS CHALLENGE 2016
LEARNING PROCESS
▸ Speed up the training process.
▸ Reducing the number of features applied.
▸ Reduc...
RECSYS CHALLENGE 2016
RESULTS & CONCLUSIONS
Run Internal Official Improved
Baseline 22849 30721
Baseline BM25 (b=0.2 / k1=1...
SEE YOU THE NEXT YEAR!
Jobandtalent at recsys challenge 2016
Upcoming SlideShare
Loading in …5
×

Jobandtalent at recsys challenge 2016

1,307 views

Published on

Recsys 2016 workshop

Published in: Engineering
  • Be the first to comment

Jobandtalent at recsys challenge 2016

  1. 1. JOBANDTALENT AT RECSYS CHALLENGE 2016
  2. 2. RECSYS CHALLENGE 2016 LEARNING TO RANK IN RECSYS 2016 ▸ Learning to Rank Framework ▸ Listwise, Pairwise, Pointwise ▸ Relevance Judgements ▸ Training & Evaluation dataset ▸ LambdaMART ▸ Gradient tree boosting & NDCG loss function
  3. 3. RECSYS CHALLENGE 2016 DATA EXPLORATION ▸ Understanding data. ▸ Probability of Item being interacted "Popularity". ▸ Probability of Item being interacted by his terms titles ▸ Probability of recurring interactions. ▸ Probability of a user interacting Items and geographical distance
  4. 4. RECSYS CHALLENGE 2016 RETRIEVERS ▸ Job title/Job tag match ▸ Relevant items per interaction ▸ Interactions made by user ▸ Impressions shown to user ▸ Collaborative filtering, Item to Item/User to User ▸ Clustered users.
  5. 5. RECSYS CHALLENGE 2016 COLLECTORS ▸ Freshness. ▸ Item popularity. ▸ User behavior.
  6. 6. RECSYS CHALLENGE 2016 LEARNING PROCESS ▸ Speed up the training process. ▸ Reducing the number of features applied. ▸ Reducing the number of elements used to train. ▸ Internal evaluation. ▸ NDCG@30 metric.
  7. 7. RECSYS CHALLENGE 2016 RESULTS & CONCLUSIONS Run Internal Official Improved Baseline 22849 30721 Baseline BM25 (b=0.2 / k1=1.2) 23591 36857 20 % Interactions made by the user 162846 229949 523 % Training process 176312 241422 5 % Impressions with decay 400217 434433 79,94 % Item popularity 435325 475940 10 % Clustering 442049 488192 3 % Collaborative filtering 457391 507022 4 % Best effort @ 2000 480396 533232 5 % Best effort @ 9000 482083 535899 1 %
  8. 8. SEE YOU THE NEXT YEAR!

×