While social networking sites already help users maintain connections with friends and interest groups, a recent trend in web services has been the growth of semantic technologies that connect people across different websites. New York startup AdaptiveBlue offer a service called “Glue,” which uses semantic web techniques to understand the content of popular web pages that describe lifestyle objects such as books, music, and movies. It then lets you do many useful, in-context things based on that data, such as learn more about the movie or actors on IMDB, buy the soundtrack on Amazon, read the historical background on Wikipedia, and a interact with a variety of other sites and services.
But perhaps Glue’s most compelling feature is the ability to link the activities of users across different websites, based on nothing more than the content of the pages they visit. For example, if a friend posts a recommendation while viewing the Slumdog Millionaire page on IMDB, Glue can display that friend’s comment when you visit the Slumdog Millionaire page on Netflix.com.
Until recently, the primary way users could interact with Glue was through browser extensions like a Firefox plugin, Adaptive Blue’s widgets, or the Glue iPhone app. However, developers can now harness the potential of Glue’s semantic data using the recently released Glue API. The Glue API provides a REST-based interface which returns data in XML or JSON format. The API allows developers to access most of the functions found in the Glue Firefox plugin, such as the ability to view friend activity or to post Twitter-like comments (known as “twocents”) about a particular book or movie. Currently, the only user authentication method available is basic HTTP authentication. For more, see our Glue API Profile.
In a recent blog post, Adaptive Blue’s Mark Tabry described the methods of the Glue API that he thinks are the most interesting:
object/get provides a direct translation between a URL and the object it represents. Given the URL of any of the hundreds of sites Glue understands, the API will return as much rich metadata as possible. So for a movie, for example, you would be returned the title, director, release year, actors, and description, all in nicely formatted XML or JSON. object/links is almost the inverse function. It uses the aggregate browsing experience of our users to find as many URLs representing an object as it can.
One challenge to any service that tries to use semantic web techniques is that most web pages that lack machine readable descriptions of their content. While Glue is able to understand the content of popular websites like IMDB or Wikipedia, many other websites simply lack the descriptive metadata that software such as Glue requires. Adaptive Blue’s approach to this problem has been to propose a simple metadata standard that can be easily implemented by online publishers. This data format, which they call “AB META,” allows websites to add descriptive metadata to pages using standard HTML META headers.
Now that Glue’s released their API they’ve invited developers to win a fully paid weekend trip to NYC by creating the best mashup in their Glue API Launch Challenge. This announcement includes 10 interesting ideas for Glue apps and 2 follow-up blog posts highlight the possibilities of using the API for visualizing the web of people and things as well as semantics and search.