Technology > Multi-Stream Retrieval

Multi-stream Retrieval

The Web and Your Web



Information on the Internet is increasing at a breakneck pace. New content, applications and mashups are continuously created. There is a certain subset that we interact with, on a regular basis - web based email, social networks, our favorite music app, our favorite local news source etc. This subset of the web is dear to us. While the whole web i.e The Web is growing, Your Web - the web you trust, is small and unique to you.

Making Computers Understand Your Web



Systems that help you search and discover from the web must taken into consideration the parts of the web - Your Web - that is more important to you than the rest.

The popular approach to model the web is to consider it as a collection of documents that link to each other. Search engines today use links as a prominent signal in understanding the web. With the advent of social networks, another signal, namely sources - people, organizations and other entities - have come into picture.

This additional dimension of sources/endorsers is very useful from an information retrieval standpoint

The affinity towards certain sources of information has been well understood. Social networks and publishing platforms such as Twitter, Facebook, Tumblr, Wordpress have lowered the barrier to endorsing content on the internet, and understanding how these activities influence consumption form an integral part of undestanding Your Web

Contextual Discovery through Your Web



Such a system, that indexes your web, is very useful when you have a complementary system that understands the context of a user and has the ability to conveniently deliver discovery to the user.

Pugmarks is a context engine - a system that understands the context of a user, and uses sophisticated algorithms to deliver great content to the user.

The Multi-stream Retrieval System powering Pugmarks



Pugmarks is powered by a multi-stream retrieval system that uses hundreds of thousands of people and websites as sources and millions of documents to determine the best recommendations available for your context. The recommendations hence recieved are highly precise and personalized, leading to superior experience