Researchers at the Palo Alto Research Center seem to be creating the ultimate Twitter client. Not because of the number of columns you can have, the types of URL shorteners it supports, or any of the other features that users and developers normally worry about, though. It’s because this particular client aims to make sense of the deluge of tweets many of us see on a daily basis. Even when following a few hundred people, there is a monumental amount of information to dig through and usually leads to information overload.
This new Twitter client, known as the Eddi Project, uses sophisticated algorithms to help you easily makes sense out of a large number of tweets. Rather than the normal routine of scanning through each tweet in your timeline manually and trying to parse everything in your head, Eddi will do it for you, picking out the topics of interest and giving you a clearer picture of what’s going on. The name comes from the idea of eddies in a stream, which indicate an area of contrary motion within a stream. Ed Chi, area manager and principal scientist for the Augmented Social Cognition Research Group at PARC, believes that people choose to dip in and out of their Twitter stream, but don’t necessarily want to consume it all at once.
The researchers have developed two primary ways of interacting with the Twitter stream. The first is a recommendation system. It finds tweets that a user might be interested in based on that users interactions with others and the content of the user’s tweets. This is somewhat similar to technology we already see in the wild for recommending new people to follow.
The second tool is a topic browser. This is the more interesting piece as it is at the heart of summarizing your entire Twitter stream, enabling you to quickly figure out what’s going on at a glance without reading every post. Figuring this out is not as easy as one might think. At the core of the problem is Twitter’s 140-character limit. Most natural language processing relies on having a good amount of text to deal with, but 140 characters is far from adequate. Researchers then realized that search engines have been dealing with this issue from the beginning and started looking at the problem from a different perspective.
The new system should be available for the public some time this summer. Until then, you can play with their recommendations engine Zerozero88, a robot which will help dig up news stories customized for you.