Cadmus Offers Algorithmic Twitter Feed Curation

Phil Leggetter
Feb. 15 2011, 12:00AM EST

Filtering, curation, aggregation and relevance have long been touted as the next important steps for social media. In actual fact we've had curation services in the form of Digg, Reddit, StumbleUpon, Delicious and many more for a long time. However, the questions "How do I find the blog post, tweet, facebook update, photo or check-in that is most relevant to me?" and "How do I know I've not missed a really important piece of news?" are still asked daily. So, there still isn't a definite answer - or a definitive service. Cadmus started off as a "real-time filtering service" for a product called ViewPoint, but has grown into a self-standing application and the Cadmus API can help users and developers answer these questions.

The Cadmus application attempts to answer a problem as outlined in a blog entry introducing Cadmus.

The problem we are trying to solve here is that we don’t have all day to stay on top of these social media services. Most people check into these services ever so frequently. And every time they do, they have to scroll through pages of posts to find out what’s going on. From the statistics that we have collected so far, we have found that nearly 20% of all the posts in a user’s stream are similar. This means that every time you login, you have to scroll through pages of posts with low signal to noise ratios.

The app lets you add services such as Twitter, any RSS feed and FriendFeed, but the API currently focuses purely on Twitter as a source of information including methods to curate posts, trends, comments, links, searches, lists and find related tweets. These API methods all return information from a Twitter user stream identified by a Cadmus API key, or via Twitter authentication, if you want to build an application that can use the API for any user.

There are a number of services developed to help you find the most relevant social media posts. They can work in different ways; some define relevance and quality through user voting systems such as Digg, Hacker News and Reddit. Others apply algorithms such as StumbleUpon (combined with user voting) and Pearl Trees (and user organizing stories) and frequently services use a number of methods to try and give their users the most personalized and relevant information. Some undoubtedly work better for some users than others, and there are a number of factors that influence why what works for one user is a turn off for another. But by exposing an API to your service you have the possibility of other products and services being developed on top of yours, and this can increase the chances of your service being successful, even if it's indirectly and the application you develop on top of your API isn't successful in its own right. Cadmus appears to be following this model and time will tell what works for them.

The Cadmus app is an example of a service aiming to ensure that you are less likely to miss tweets that really matter to you. By offering an API they increase their chances of being successful, but ultimately it will depend on how reliable, and clever, the underlying algorithm is. The best way to improve the algorithm is by getting users and feedback, so the API is definitely a good way of doing this.

Phil Leggetter Developer Evangelist at Pusher, Real-Time Web Software & Technology Evangelist, team leader, product developer, micropreneur, managing director of a real-time web and social media software company, blogger and twitter user (@leggetter).

Comments

Comments(10)

Francois

Hi Phil,

Interesting blog post that highlights the algorythm and human ways to solve your two problems "How do I find the web contents relevants to me?" and "How do I know I have missed a piece of news?". And thanks to have quoted Pearltrees (I am one of the co-founders).

I would like to point out one thing I discussed with a sociologist two weeks ago. It is the three needs we have in terms of search on the Internet. The first need is when your are looking for a single item : a hairdresser in SF and classic search engines like Google or Bing are really good at that. There is a need in terms of "news" from your communities, your friends, your collegues and work environment and the news of what you are interested and here a personal mix of Facebook, Twitter and RSS reader is probably the best solution. And the third need is a kind of browsing into a specific topic to discover everything about it, here comes curation services. At Pearltrees our bet is that only human people can do that in an interesting way