This document summarizes a new social media application called Conn'r. Conn'r aims to connect users for simple conversations without profiles or friend connections. It will randomly match two users who can chat instantly without registration. The goal is to satisfy the basic human need for connection in a quick and hassle-free way. Advanced machine learning algorithms will also analyze conversation logs to flag unwanted content like spam or abuse. The application is nearing completion and plans are outlined for alpha and beta testing along with future ambitions around analytics and research.
2. The “social web”
• Register, confirm email, and fill up thirty five
fields.
• Decorate profile with party pictures.
• “Friend” everyone you know, and some you
don’t.
• It’s all about building connections.
• It’s worked well for most of us.
4. The Concept
• Our application enables two randomly
selected users to chat with each other.
• It’s not about finding new friends or dates, it’s
about having a conversation.
• No registration or filling up profiles, just click a
button and start talking.
• Instant gratification.
5. Target Market
• Almost everyone.
• It satisfies the most basic human need to
connect – the same reason why the social web
is so big.
• Individuals who want to simply talk to
someone new without the hassles of
registration, “friends” and profiles, etc.
6. A bigger goal
• We will allow users to flag unwanted elements
(spam, offensive, abusive), and log those
conversations.
• Powerful AI algorithms will analyze these logs,
and pre-emptively flag such users.
• This functionality will be opened as a public
API, like a new generation of “spam filters”.
7. Nuts and Bolts
• A powerful real time communication
framework ensuring near zero latency.
• Technology
– WebServer – Django + FastCGI
– Frontend – Adobe Flex™
– Backend – C++
– Custom Message Queue server.
• Extremely scalable – both horizontally and
vertically.
9. Timeline
• Backend almost finished, work on frontend
continues.
• Alpha launch mid April.
• Beta over summer.
• Long term ambitions:
– Log processing & Machine learning
– Analytics and research aspects
10. Demo
// TODO: Insert picture of kitten
saying “oh noes, I broks your code”
11. Us
Akshit Sharma Anirudh Sanjeev
• 4th year, IIT Kharagpur • 4th year, IIT Kharagpur
• Interests: Machine learning, • Interests: Frontend design,
prediction models, web scalable backend, rapid
development. prototyping, web security.
• Internships: Microsoft • Internships: Google (India),
Research, Hong Kong, Microsoft Research, Iviz
EPFL(Switzerland), Google Securities.
SOC, IBM IRL • anirudh@anirudhsanjeev.org
• asharma.iitkgp88@gmail.com