Dilip Ganesan
In this assignment we are going to analyze the recommendation system of LinkedIn. With north of 135 million members, its mission is to connect the world’s professionals to make them more productive and successful.With terabytes of data flowing through the systems, generated from member’s profile, their connections and their activity on LinkedIn, they have to have a recommender system so the users/advertisers/recruiters see what they really want.
The target audience are employee(professionals) and employers(recuiters) From User perspective, some of the information that i see daily are
What people are talking about now
Who to follow - This is based on my list of great personalities who i follow, it lists people from same domain or from same technology platforms.
Job Recommendation - This is based on my current profile, my education background, my location, my technology skill set.
PYMK(People you may know) - This is bases on my previous schools of education, previous employers details and etc.
Talent Match - This is for job recruiters based on job post they do a profile match and suggest highly qualified candidates.
New Recommendations - Top articles shared by industry leaders.
Companies to follow - The use a combination of content matching and collaborative filtering.
From companies point of view, Advertising is their cash cow. The more the adverstisement they are able to direct to correct users the more the get the clicks and transactions out of advertisement. For recruiters who have paid subscription the system has to recommend the best and brightest candidates else they might be looking for different websites. So keeping in view of the their interest and customers interest their recommender system is one of the best compared to other job sites.
Some of the technologies used by linkedin for recommender systems are: Hadoop Lucene R mahout
I personally feel, time has come a change in how we look at our feeds. Traditionally whether it is twitter or facebook the feeds are meant to be scrolled. I think some one has to break this trend of scrolling and comeup with novel way of showing recommendations and entire site as a whole. I am looking for that one site which is going to break that chain.