Twitter will soon give developers the ability to filter messages by popularity and language.
The company said in a blog posting Wednesday that it will add two new fields to the public metadata it associates with messages on its service, which are popularly known as Tweets. The first is a "filter_level" that allows developers to screen out only popular items, roughly equivalent to the way users of Twitter's home page can choose to view only "Top Tweets" in search results.
In its public FAQ, Twitter defines "Top Tweets" as those that "many other Twitter users have engaged with and thought were useful." It says they are filtered out by an algorithm that looks for items that many people are interacting with and sharing.
Twitter will also add the ability to filter by language, the same feature that is available to those using its advanced search page. The new functionality will give developers the obvious ability to provide more relevant results to users, but it will also provide a new way to analyze global reactions to events on Twitter.
The new fields will give developers a better way to sort through the mass of Twitter messages returned for popular searches and subjects. Twitter is working to fine tune users' ability to sort through its data, also adding features such as verification of people using Twitter, such as public figures and journalists.
The two new fields will be added to Twitter's streaming APIs (application programming interfaces), which provide a steady flow of Twitter messages in real-time to developers. The ability to filter by language will also be enabled for applications that query for messages on a per-request basis. The filtering by popularity feature will go live on Feb. 20, with the language feature to be added shortly after that.
For developers that wish to use the new filtering capability, the company said Twitter messages will initially be marked "medium," "low" or "none," values, with a "high" value to be added in the future. The "medium" and "high" levels are on par with the "Top Tweets" filter in its search results.