4. Our Mission
Connect the world’s professionals to make them
more productive and successful.
Our Vision
Create economic opportunity for every
professional in the world.
Members First!
16. Problem Definition
Address job seeker* need: find dream job
– Huge cost of consumption
Lag between view and application is in hours/days
– Extremely high level of expectation
– No forgiveness for less than perfect recommendations
Accuracy is key!
(*) 20% active, 60% receptive -- 10/12 Job Seeker Survey, 20K in 7 countrie
17. How LinkedIn matches people to jobs
Job Corpus Stats
Matching Transition probabilities
Connectivity
Binary yrs of experience to reach title
title industry …
Exact matches: education needed for this title
geo description …
company functional area geo, industry,
…
User Base Soft Similarity
(candidate expertise, job description)
transition
Filtered 0.56
probabilities,
Similarity
Candidate similarity, (candidate specialties, job description)
… 0.2
Transition probability
Text (candidate industry, job industry)
General Current Position 0.43
expertise title
specialties summary Title Similarity
education tenure length 0.8
headline industry
Similarity (headline, title)
geo functional area
experience … 0.7
Ensemble
.
.
Scorings
.
~250B Member Job Pairs a day! 17
18.
19. Feature Engineering – Entity Resolution
Companies
‘IBM’ has 8000+ variations
- ibm – ireland
- ibm research
- T J Watson Labs
- International Bus. Machines K-Ambiguous
- Deep Blue
Huge impact on the
business and UE
Ad targeting
TalentMatch
Referrals
Asonam’11, KDD’11
19
20. Feature Engineering – Would you move
Open to relocation ?
Region similarity based on profiles or network
Region transition probability
predict individuals propensity to migrate and
most likely migration target
Impact on job recommendations
20% lift in views/viewers/applications/applicants
21. What should you transition to .. and when ?
Probability of switch
Months since graduation
21
24. Power of aggregation..
Before
employees worked at
Yahoo! (247)
Google (139)
Microsoft (105)
Oracle (93)
IBM (68)
Before
employees worked at
Microsoft (1379)
IBM (939)
Yahoo! (608)
Oracle (558)
24
25.
26. Segmented Models
Demographic Segmentation
Students (or recent grads)
US vs International members
Industry Specific models
e.g. Finance vs Technology
Behavioral Segmentation
Job Seekers (Active)
Daily Users vs Monthly Users
27. Job Seeking
Types
Active
Passive receptive
Not a job seeker
Modeling
Ordered logistic reg.
Impact
~10x application rate between Active and Passive receptive
27
32. A/B Testing
Is A better than B.. Let’s test
Beware of
- novelty effect
- cannibalization
- potential biases (time, targeted population)
- random sampling destroying the network
effect
Enjoy testing furiously! Don’t forget to A/A test first
job views per 5% bucket range - 6/5/11 job views 6/19/11
9,000
7,000
8,000
7,000
6,000
6,000 5,000
5,000 4,000
job views per 5% bucket range -
4,000 6/5/11
3,000 job views 6/19/11
3,000
2,000
2,000
1,000
1,000
0 0
0 5 10 15 20 25 0 5 10 15 20 25
(“Seven Pitfalls to Avoid when Running Controlled Experiments on the Web”, KDD’09
“Framework and Algorithms for Network Bucket Testing” WWW’12 submission)
32 32
33. Some final remarks
Most people aren't actively looking for jobs.
– Many people are but most aren’t
– Complicates evaluation and training
Important not to offend
– JYMBII: I am more senior than that!
– What is the price of a bad recommendation? (PYMK vs. JYMBII)
You can’t always get what you want
– Every employer wants the hottest candidate.
– The perfect candidate already works for you.
33
34.
35. Credits
Engineering : Abhishek Gupta, Adam Smyczek, Adil
Aijaz, Alan Li, Baoshi Yan, Bee-Chung Chen, Deepak
Agarwal, Ethan Zhang, Haishan Liu, Igor Perisic, Jonathan
Traupman, Liang Zhang, Lokesh Bajaj, Mario
Rodriguez, Mitul Tiwari, Mohammad Amin, Monica
Rogati, Parul Jain, Paul Ogilvie, Sam Shah, Sanjay
Dubey, Tarun Kumar, Trevor Walker, Utku Irmak
Product : Andrew Hill, Christian posse, Gyanda
Sachdeva, Mike Grishaver, Parker Barrile, Sachit Kamat
Alphabetically sorted
Mission: For us, fundamentally changing the way the world works begins with our mission statement: To connect the world’s professionals to make them more productive and successful. This means not only helping people to find their dream jobs, but also enabling them to be great at the jobs they’re already in. Vision: But, we’re just getting started. By our measure,there are more than 640 million professionals in the world. And roughly 3.3 billion people in the global workforce. Ultimately, our vision is to create economic opportunity for every professional, which we believe is an especially crucial objective in light of current macroeconomic trends.Our most important core value is that members come first.
Speak to the growth.
Speak to the growth.
Talent Match: job posting flow: When recruiters post jobs we in real time suggest top candidates fit for the job
Similar Profiles: For recruiters, useful in sourcingIf a recruiter finds a profile that he is interested in, this helps him in finding other similar profiles that may be of interest
As part of Talent pipeline, this new features helps recruiters in importing resumesWhen recruiters import resume we suggest in real-time LinkedIn profile that might correspond to those resumes so they can link the prospect with a LinkedIn account.
As part of Talent pipeline, this new features helps recruiters in importing resumesWhen recruiters import resume we suggest in real-time LinkedIn profile that might correspond to those resumes so they can link the prospect with a LinkedIn account.
As part of Talent pipeline, this new features helps recruiters in importing resumesWhen recruiters import resume we suggest in real-time LinkedIn profile that might correspond to those resumes so they can link the prospect with a LinkedIn account.
Job recommendations are a feedback on how well they write their profile on Li. Main problem definition because there are secondary problem definitions.
As part of Talent pipeline, this new features helps recruiters in importing resumesWhen recruiters import resume we suggest in real-time LinkedIn profile that might correspond to those resumes so they can link the prospect with a LinkedIn account.
8000 name variants of IBMWe use the definition of entity resolution terminology k−ambiguous and k−variant from [10]. Same company name can denote multiple company entities but each occurrence of a company name references a single entity only. A name referring to k different entities is called k − ambigous. Additionally, An entity which can be referred to by k different names is called k − variant.Ranker approach does not work. A given name may not be resolvable in the sense that the company entity has not being created yet…Classification problemGiven a pair of (member position, company entity), a binary classifier would determine whether there is enough evidence to resolve the member position to the company entity. This would address the problem of the ranking approach in that an unresolvable member position would most likely remain unresolved because the classifier has insufficient evidence for any company entity. It is certainly possible that there could be multiple company entities with sufficient evidence for a member position.
Unreasonable effectiveness of Big Data.. This chart shows the probability of holding a title across all titles, plotted vs number of months after graduation. Notice the spikes.. They are ~12 month almost perfectly aligned.. Remember the itch that you had when you finished 2 years at your company
Huge impact on the business: Target more aggressively active JS while not spamming the others. Personalize the set of recommendations- Ads
Exposing a % of users to a new treatmentMeasuring the effect on metrics of interestRunning statistical tests to determine whether the differences are statistically significant, thus establishing causality