I think having a social newsfeed in your site has become a bare minimum requirement for most of the sites that aim to be an ecosystem in a certain field.
I thought of having that in ShortFilmWindow. Since I was totally out of touch, I had to really start digging lots of information around it.
What I have realised is that getting the data structure right is the biggest key. An interesting graph structure tailored to your user behaviour will impact system performance greatly.
I have collected some resources that might help folks understand the problem, define it better and take right decisions.
Most of these are good starting point and a person can dig deeper if need be.
They talk about graph structure and system for searching through it in a efficient way. There are two courser courses that I am gonna take and make everyone in my development team take too (they are starting in a week).
I thought of having that in ShortFilmWindow. Since I was totally out of touch, I had to really start digging lots of information around it.
What I have realised is that getting the data structure right is the biggest key. An interesting graph structure tailored to your user behaviour will impact system performance greatly.
I have collected some resources that might help folks understand the problem, define it better and take right decisions.
Most of these are good starting point and a person can dig deeper if need be.
They talk about graph structure and system for searching through it in a efficient way. There are two courser courses that I am gonna take and make everyone in my development team take too (they are starting in a week).
Facebook Graph Structure and Search Systems:
Facebook graph search with neo4j: http://tinman.cs.gsu.edu/~raj/8711/sp13/neo4j/Neo4J_%20Report.pdf
Facebook distributed data store (Tao): https://www.cs.cmu.edu/~pavlo/courses/fall2013/static/papers/11730-atc13-bronson.pdf
Facebook (Unicorn): http://www.vldb.org/pvldb/vol6/p1150-curtiss.pdf
Twitter Scalability:
Coursera Courses:
Anyways next step is personalization. But I have done enough ML courses to not get intimidated by it. The above thing still intimidates me.