|Sunday, June 12, 2005|
|New matching responses feature in HAL3000|
|Savyon, Israel -- June 12, 2005 - Artificial Intelligence Research (Ai) |
(http://www.a-i.com) announced today the release of a new feature for advanced trainers of the latest version of the HAL Learning Algorithm, HAL3000.
The new feature, called 'show matching responses', allows the trainer of a HAL3000 personality (particularly large personalities consisting of over 10,000 turns) to issue a correction with the "wrong?" prefix, followed by a pattern consisting of words and '*'. (examples: 'what*time' or 'time' or 'what*time*go')
This pattern is used by the HAL3000 engine to filter the response bank of the personality. The list of matching responses is then shown in the session log window. Any of the displayed responses can then be used by the trainer for the current correction, by copying and pasting the response or dragging and dropping it into the input field.
The new matching responses feature was introduced as a result of continuous observation of training habits of advanced HAL trainers with large personalities. These trainers often experience difficulties in recalling previously trained responses to given input sequences. The new feature provides them with an immediate, easy to use tool.
Artificial Intelligence Research (Ai) is a groundbreaking research project, focusing on creating genuine Artificial Intelligence – the technology that enables machines to converse with humans in natural language. Ai's research facility is located in Savyon, Israel.