 |
|
|
|
╯Conversation Function |
|
|
With full use of the Semantic Recognition Function and the Scenario Function of CAIWA, it is possible to carry on a flexible and natural conversation. Such a fixed response as ﹍I can﹊t understand you,﹎ in regard to ﹍utterance not included in knowledge base﹎ can best be avoided, and responses are dynamically created to carry on a natural conversation.
model;
User:﹍I went to the beach yesterday, you know.﹎
Robot:﹍Hmmm, with whom? (◤response created dynamically)
﹍A User Model﹎ is included in the CAIWA system mounted on the robot.﹛The ﹍User Model﹎ is something like a database that holds information about individual users. It maintains such information as each user﹊s name, birthday and more so that appropriate responses to users are made possible.
The ﹍User Model﹎ also contains a parameter called ﹍Familiarity﹎, whose numerical values, either by decreasing or increasing depending on the familiarity between the robot and the user, affects the contents of conversation. Thus, it is possible to talk in a special way to the users the robot feels closer to (or more familiar with) and act proportionately less familiarly toward other users.
The robot is also able to make a judgment as to which user to talk to when another user interrupts a conversation depending on the level of ﹍Familiarity﹎ and the conversation content (control is made with a parameter called ﹍conversation posture﹎).
|
|
|
 |
 |
|
CAIWA Robot
Voice Recognition
