LinkedIn Chief Scientist Christian posse report on recommendations

According to the KDD Conference, he was very impressed with the lectures by LinkedIn's chief scientist posse, not only from his French accent, but also from LinkedIn's meticulous data analysis.

Title and abstract are as follows:

Key lessons learned building recommender systems for large-scale social networksChristian posse
LinkedIn Inc.

Abstract

By helping members to connect, discover and share relevant content or find a new career

Opportunity, recommender systems have become a critical component of user growth and
Engagement for social networks. The multidimenstmnature of engagement and diversity
Members on large-scale social networks have generated new infrastructure and Modeling
Challenges and opportunities in the development, deployment and operation of recommender
Systems. This presentation will address some of these issues, focusing on the modeling side
Which new research is much needed while describing a recommendation platform that enables real-
Time recommendation updates at scale as well as batch computations, and cross-leverage
Different product recommendations. Topics covered on the modeling side will include optimizing
For multiple competing objectives, solving contradicting business goals, modeling user intent and
Interest to maximize placement and timeliness of the recommendations, utility metrics beyond CTR
That leverage both real-time tracking of explicit and implicit user feedback, gathering training data
For new product recommendations, virality preserving online testing and virtual profiling.
Categories & subject descriptors: h.2.8 [Database Management]: Data Mining.
Author keywords: recommender systems, real-time updates, multi-objective
Optimization, user intent modeling, online testing.

Bio

Dr. Christian posse is principal scientist at LinkedIn Inc. Where he leads the development
Recommendation solutions as well as the next generation online experimentation platform. Prior
LinkedIn, dr. posse was a founding member and technology lead of Cisco Systems Inc. Network
Collaboration Business unit where he designed the search and advanced social analytics of pulse,
Cisco's network-based search and collaboration platform for the enterprise. Prior to Cisco, dr.
Posse worked in a wide range of environments, from holding faculty positions in US universities,
To leading the R & D at software companies and a US National Laboratory in the social networks,
Biological networks and behavioral analytics fields. His interests are diverse and include predictive
Analytics, search and recommendation engines, social networks analytics, computational social and
Behavioral Sciences, computational linguistics, and information fusion. He has written over 40
Scientific peer-reviewed publications and holds several patents in those fields. dr. Posse has a PhD

In Statistics from the Swiss Federal Institute of Technology, Switzerland.

LinkedIn data:

> 50% of connections are from recommendations (pymk)
> 50% of job applications are from recommendations (jymb)
> 50% of Group joins are from recommendations (gyml)

(A wide range of LinkedIn recommendation types)

What is a Recommender System? -- The Explanation of posse is a recommender selects a product that if acquired by the "buyer" maximizes value of both "buyer" and "seller" at a given point in time, from the perspective of job recommendation, it is to balance the interests of candidates and recruiters directly and maximize the benefits.

User Experience matters most --- this is an important point raised by Posse.
Not only users' interests, willingness to apply for a job, but also users' clicking processes (User flow) when they access LinkedIn will affect the recommendation results. In addition, the recommended description method (set right expectations) is very important to explain the results (explain recommendations. Design the entire system from the user's perspective

Specific data:

Job recommendation Use Cases

User Experience	Optimization	Impact on Job Application Rate
User intent/location	Homepage personalization	2.5x
User flow/user intent	Before vs. After having applied to a job	7X
User Flow	LinkedIn homepage vs. Jobs Homepage	10x
Location	Center rail vs. Right rail	5x
Message	Followers vs. Leaders	2x

In addition, interact with the user is also very important, this test of product design capabilities

Leverage social referral, which is written by posse in the recsys paper. I did not make a record at the site, but I remember it was very helpful to improve the recommendation performance. The paper name is:

M. amin, B. yan, S. sriram,. bhasin and Christian posse, social referral: using network connections to deliver recommendations, Proceedings of the sixth ACM Conference on recommender systems (recsys '2017)