Hunch Launches API: Go Make Decisions

While Google leads the way in web search and Yahoo leads the way in answers, many upstart services are attempting to help you make decisions by giving you smarter answers to your questions. Microsoft has billed its new search offering Bing as both a "decision engine" and an "answer engine." Now there's Hunch, a new start-up founded by Flickr's Caterina Fake, which also aims to be a decision engine, but using a very different model. By first asking the user questions ranging from food preferences to pet peeves, Hunch tries to provide answers to questions that best match the user's interests based on crowdsourced data collected from other users.

Like other new search and answers services with APIs like the Bing API and Wolfram Alpha API, there is now a new Hunch API. The API is currently available in two forms, client-side and server-side. The Hunch developer's page discusses the differences in detail:

There are two different Hunch APIs. The first is the client-side JavaScript API, which you use if you want your users to connect with Hunch directly from your own Web site. The second is the server-side JSON API, which you use if either your server needs to pull data from Hunch, or if you are developing an application that is not web-based (such as a desktop or mobile device application).

Hunch Decision Engine

Hunch's client-side JavaScript API provides two functions which allow developers to embed ready-made Hunch interface widgets inside of a DIV on their webpages. The API also gives developers access to many lower level callback functions to provide greater control over things like the amount of questions a user needs to answer before the system is sufficiently trained to make suggestions. The server-side version of the Hunch API is a RESTful interface accessible using HTTP GET requests, which only returns data encoded in JSON format.

One very intriguing feature available to users of Hunch's API is direct access to the correlation data that it uses to provide decision results. Hunch claims that this data is "anonymized and aggregated to protect" user privacy. The Hunch API page gives an example of what a developer could use the interface to discover: could ask our taste database for the correlation between people who own guns and the people who like SUVs, or what the most salient personal trait is for predicting whether someone likes the movie Napoleon Dynamite. Over time we hope to provide a complete database of people's aggregated taste preferences.

Currently, use of the Hunch API comes with several conditions. As the Hunch API page states, "our API is free for non-commercial use. We ask that you attribute Hunch by providing a link to our Web site wherever you use our API." In order to use features of the Hunch API, developers must sign up for an access key. While there is no limit to the total amount of users making requests with the server-side API through a particular public developer key, each user can only make 1000 requests per day per IP address.

Hunch might not be the perfect tool for answering every question, but its interface is certainly fun and easy to use. And by creating a powerful API that gives developers the ability to build on this service and data could turn out to be just the right decision by Hunch.

Be sure to read the next Social article: Twitter Announces Retweet API


Comments (0)