Next Billions - Spoken Web Social Navigation

The Internet is accessible largely through text-based technologies (text messaging, email, web browsing) in English and is thus not useful to the billions of people without sufficient literacy, without regular access to a computer, or speaking many thousands of other native languages. To address this problem, our IBM India Research Lab envisions the Spoken Web, a network of interconnected voice driven applications accessible through any phone with audio content generated by users. However, voice driven applications remain difficult to use, error prone, and slow to navigate due to their sequential nature, particularly when audio content is user-generated (and linked) and thus growing, unorganized, and noisy.

We want to understand and improve the navigation patterns of users of voice driven applications. Given the likelihood of shared information needs among users in a given region, we expect significant similarities in navigation patterns. Given this, one promising approach is social navigation: use well-trodden paths to recommend future directions. In particular, we are exploring whether collaborative filtering can be used to recommend useful navigation paths on a voice driven application deployed to rural farmers in India.

Associated Teams

Social Computing Group