Keynote talk at 4th International Workshop on Social Recommender Systems (SRS 2013)
In conjunction with 22nd International World Wide Web Conference (WWW 2013). More details: http://cslinux0.comp.hkbu.edu.hk/~fwang/srs2013/
3. Outline
• About LinkedIn!
• Social Recommender Systems at LinkedIn!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
3
11. Outline
• About LinkedIn!
• Social Recommender Systems at LinkedIn!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
11
14. Outline
• Social Recommender Systems at LinkedIn!
• LinkedIn Today: Recommend News!
• Jobs Recommendation!
• Related Searches Recommendation!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
14
15. Linkedin Today: News Recommendation
• Objective: serve valuable professional news, leading to
higher engagement as measured by metrics such as CTR
15
16. News Recommendation: Explore/Exploit
16
item j from a set of candidates
User i
with
user features
(e.g., industry,
behavioral features,
Demographic
features,……)
(i, j) : response yijvisits
Algorithm selects
(click or not)
Which item should we select?
! The item with highest predicted CTR
! An item for which we need data to
predict its CTR
Exploit
Explore
Agarwal et. al 2012
19. News Recommendations: Revised Algorithm
• Explore/Exploit scheme!
• Explore: choose an item at random with a small probability (e.g., 5%)!
• Exploit: choose highest scoring CTR item (e.g., 95%)!
• Temporal smoothing: more weight to recent data!
• Impression discounting: discount items with repeat views!
• Segmented model: segment users in CTR estimation!
• Opportunity: Multi-arm bandit problem
19
20. Outline
20
• Social Recommender Systems at LinkedIn!
• LinkedIn Today: Recommend News!
• Jobs Recommendations!
• Related Searches Recommendation!
• Social Graph Analysis!
• Social Update Stream and Virality!
• Scaling Challenges
21. Jobs Recommendation
• Goal: recommend dream jobs to job
seekers!
• Challenges!
• Lag between view and application, offer,
acceptance!
• High level of expectations
21
22. Jobs Recommendation
22
17
Corpus StatsJob
User Base
Filtered
title
geo
company
industry
description
functional area
…
Candidate
General
expertise
specialties
education
headline
geo
experience
Current Position
title
summary
tenure length
industry
functional area
…
Similarity
(candidate expertise, job description)
0.56
Similarity
(candidate specialties, job description)
0.2
Transition probability
(candidate industry, job industry)
0.43
Title Similarity
0.8
Similarity (headline, title)
0.7
.
.
.
derived
Matching
Binary
Exact matches:
geo, industry,
…
Soft
transition
probabilities,
similarity,
…
Text
Transition probabilities
Connectivity
yrs of experience to reach title
education needed for this title
…
Ensemble
Scorings
Bhasin et. al 2012
23. Magic Is In Feature Engineering
• Open to relocation?!
• Region similarity based on profile or network!
• Region transition probability!
• Predict members’ propensity to migrate and
potential regions
23
24. What Should You Transition To And When
24
• Probability of holding a title wrt time: spikes 12 months aligned
Wang et. al, WWW’13
25. Job Seeking: Socially Contagious
25
[Zhang, 2012]
• Prob. of quitting increases as the #of recently left connected colleague
26. Outline
26
• Social Recommender Systems at LinkedIn!
• LinkedIn Today: Recommend News!
• Jobs Recommendation!
• Related Searches Recommendation!
• Social Graph Analysis!
• Social Update Stream and Virality!
• Scaling Challenges
27. Related Searches Recommendation
• Millions of Searches everyday!
• Help users to explore and refine their queries
27
Reda et. al, CIKM’12
31. Outline
• About LinkedIn!
• Social Recommender Systems at LinkedIn!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
31
39. Outline
• About LinkedIn!
• Social Recommender Systems at LinkedIn!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
39
44. Skill Recommendation
• Predict a skill even if not
present in the profile!
• Based on likelihood of
member having a skill!
• Features: company, industry,
skills, ...
44
Profile
Tokenize
SkillsTagger
Phrases
Skills
Skills Classifier
Profile features
Recommended Skills
45. Suggested Skill Endorsements
• Binary Classification!
• Features!
• Company overlap, School overlap,
Industrial and functional area similarity,
Title similarity, Site interactions, Co-
interactions, ...
Candidate
generation
Classifier
Features
- Company
- Title
- Industry
...
Suggested
Endorsements
45
48. Outline
• About LinkedIn!
• Social Recommender Systems at LinkedIn!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
48
49. Scaling Challenges: Related Searches Example
• Kafka: publish-subscribe messaging system!
• Hadoop: MapReduce data processing system !
• Azkaban: Hadoop workflow management tool!
• Voldemort: Key-value store
Metaphor
Hadoop
Search
Backend
Kafka
Voldemort
Related
Searches
Backend
Front
End
HDFS
49
50. Outline
• About LinkedIn!
• Social Recommender Systems at LinkedIn!
• Social Graph Analysis!
• Virality in Social Recommender Systems!
• Scaling Challenges
50
52. Acknowledgement
• Thanks to Data Team at LinkedIn: http://data.linkedin.com!
• We are hiring!!
• Contact: mtiwari[at]linkedin.com!
• Follow: @mitultiwari on Twitter
52
You!
Applied Reseacher/
Research Engineer