Project

Conversation Finder for Socially Aware Communication

A socially intelligent communication agent that assists the user in her telecommunication tasks needs "communication intelligence." One way to acquire such intelligence is to rely on the human social intelligence of co-located people to decide on whether, when, and how to alert a user of an incoming communication. As a prerequisite, the agent needs to know who is part of an ongoing conversation; a network of small, body-worn, wireless sensors helps the agent to determine this. A novel, fully distributed decision-making process is used to detect conversations; these nodes have binary speech detectors and low-range radio transceivers, and communicate asynchronously with each other on a single channel. They independently come to a conclusion about who is part of the user's current conversation by looking at alignment and non-alignment of the speaking parties, and can be queried wirelessly for this list of people.

A socially intelligent communication agent that assists the user in her telecommunication tasks needs "communication intelligence." One way to acquire such intelligence is to rely on the human social intelligence of co-located people to decide on whether, when, and how to alert a user of an incoming communication. As a prerequisite, the agent needs to know who is part of an ongoing conversation; a network of small, body-worn, wireless sensors helps the agent to determine this. A novel, fully distributed decision-making process is used to detect conversations; these nodes have binary speech detectors and low-range radio transceivers, and communicate asynchronously with each other on a single channel. They independently come to a conclusion about who is part of the user's current conversation by looking at alignment and non-alignment of the speaking parties, and can be queried wirelessly for this list of people.