Bookmark and Share
|
  • Text +
  • Text -

Computer Science

Towards Emotionally Intelligent Machines: A Comprehensive Mood Classification System

Vasserman, Lucy ('10);  Sood, Sara Owsley

In the future, computers will need to connect with users on an emotional level in addition to performing complex computations. For this reason affective computing, building machines with emotional intelligence, is an important field of research within Artificial Intelligence today. An emotionally intelligent computer must be able to both identify emotions in its user and express emotions itself. This project focuses on the former of these two by creating a system that recognizes emotion, specifically in text. Using Naïve Bayes classification, the system takes a piece of text as input and identifies the mood conveyed in the text (happy, sad, angry, etc). An extension to this project will allow internet users to not only search the web by topics or keywords, but also by mood.
Funding provided by: The Norris Foundation; Owlsey

Structures for Tracking Dialogs

Wienberg, Christopher ('10);  Bruce, Kim

Conversations are one of the most effective and the most common mode of human communication. Despite their ubiquity in the human experience, linguistic theory has only scratched the surface in understanding how conversations function. To expand understanding of conversations, we have extended work done in previous years on a computer model of conversations. The model developed is an implementation of the theoretical work of Professors Bruce and Farkas on the nature of conversations. The model converts natural-language statements to semantics, represented in first-order logic. The model keeps track of the participants discourse commitments, the items already agreed to, the still to be resolved topics of conversations, and possible outcomes of the conversation. The model is automatically updated as each assertion or question of a participant is added to the conversation, and presents this information in a human-readable fashion.
Funding provided by: The Norris Foundation

Research at Pomona