Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
2000-10-30
2004-04-13
Abebe, Daniel (Department: 2655)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C704S270000, C704S257000
Reexamination Certificate
active
06721706
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to devices that simulate personal interaction with a user through various outputs modalities such as light pulsations, synthetic speech, computer generated animations, sound, etc. to create the impression of a human presence with attending mood, ability to converse, personality, etc.
2. Background
With increasing sophistication in technology, the variety of possible features and options associated with many appliances can be daunting. This phenomenon is exemplified by satellite and cable TV where the number of program choices is unwieldy in some cases. Many examples exist, including cell phones, personal computer applications, e-trading systems, etc. In such environments it is useful for the machines to take some of the routine work out of making choices from among an overwhelming number of options. However, often, the solutions are not much less painful than the problems they are supposed to address. For example, user interfaces that filter a large number of choices using custom templates for each user must be trained as to the user's preferences. For example, a user can enter his/her preferences by actively classifying his/her likes and dislikes (“customization”). This can also be done passively such as by having a computer process “observe” the selections made by the user over time (“personalization”). Such systems are discussed in a variety of patent applications assigned to Gemstar and Philips Electronics. U.S. Pat. No. 5,515,173 for System And Method For Automatically Recording Television Programs In Television Systems With Tuners External To Video Recorders; U.S. Pat. No. 5,673,089 for Apparatus And Method For Channel Scanning By Theme; US Patent U.S. Pat. No. 5,949,471 Apparatus And Method For Improved Parental Control Of Television Use. Another example is U.S. Pat. No. 5,223,924.
The user-interfaces that permit the specification of preferences, either explicitly or passively, are often sophisticated enough to be fun and intuitive. More and more such systems have evolved toward seemingly “smart” systems that try to seem like human helpers rather than control panels. For example, help dialogs in complex software applications such as Microsoft® Office® accept natural language sentences and give text responses quasi-synchronously with animated characters. Another example of an interface that accepts natural language questions is AskJeeves®, a search engine for the Internet.
User interfaces are evolving rapidly away from function oriented systems where a sequence of steps are performed, the last step of which results in the desired result, to object oriented interfaces which allow the user to manipulate the subject matter of concern providing immediate feedback on the results. The ultimate embodiment of this idea is noncommand interfaces such as virtual reality environments that permit a user to manipulate controls and experience the effects rather than performing a prescribed series of steps to achieve a result. Agents are another type of noncommand interface. The eventual goal of some research is to provide highly intelligent agents that know all about the user's needs and lifestyle and can retrieve exactly the information required of the user at any given time. Practical embodiments of such noncommand interfaces are, however, not quite ready for introduction into practical products. One of the problems is that the software technology required is highly sophisticated and requires a great deal of programming and/or training. For now, computers still need to be told what to do and the dream of interfaces that are truly easy to use, those giving the impression of being “smart,” are for most applications, still rather far off.
Computers that look smart without actually being smart, however, have been around for a while. A very different area of computer software development began in the mid-1960s with an MIT computer scientist Joseph Weizenbaum who created a simple program with no utility apart from how smart it appeared to be. It simulated the role of a Rogerian psychotherapist by attempting to draw the user out and encouraging the user to continue talking. This program was called ELIZA. Many users found ELIZA very convincing. Even Weizenbaum's secretary demanded to be left alone with the program. However, ELIZA's conversation is effective only as long as one is willing to accept the very passive conversational style of a Rogerian psychotherapist, in which either the “therapist” usually responds by repeating part of what the “patient” has said or by asking the “patient” to explain or elaborate on what he has just said. What makes ELIZA seem convincing is that it does not do this in a rote manner. If the user mentions his mother, for example, ELIZA might ask the user to tell “her” more about his family.
Below is a sample conversation reproduced from ELIZA:
User: Men are all alike.
ELIZA: In what way?
User: They're always bugging me about something or other.
ELIZA: Can you think of a specific example.
User: Well, my boyfriend made me come here.
ELIZA: Your boyfriend made you come here.
User: He says I'm depressed much of the time.
ELIZA: I'm sorry to hear you are depressed.
User: It's true. I am unhappy.
ELIZA: Do you think coming here will help you not to be unhappy?
User: I need some help, that much seems certain.
ELIZA: What would it mean to you if you got some help?
User: Perhaps I could get along with my mother.
ELIZA: Tell me more about your family.
User: My mother takes care of me.
ELIZA: Who else in your family takes care of you.
User: My father.
ELIZA: Your father.
User: You are like my father in some ways.
ELIZA: What resemblance do you see?
So far, virtually all these so-called conversation simulators have used a basic method that generates shallow responses based on template-matches (rules) without any real understanding. For example, the template: “I wish I were <x>”(where <x>represents any series of words) matches on the user statement: “I wish I were taller.” The template is associated with the machine response: “Why do you wish you were <x>?” The machine responds with the statement: “Why do you wish you were taller?” What distinguishes one conversation simulator from another is not so much its sophistication and complexity as the size and variability of its database of responses. More sophisticated variants have a larger database of templates and responses including whimsical responses that can make them more interesting than the passive, flat responses of ELIZA.
Some conversation simulators provide information on specific topics, rather than general conversation simulation. For example, conversational simulators have been used for providing information regarding a particular topic. Basically, their libraries of responses anticipate questions about some subject and provide “canned” responses. Some conversation simulators have been programmed to appear as if they had a life story to relate. They would talk about their story when they could not come up with a good template match to keep the conversation going.
A typical conversation simulator may be described as having two parts: a user-interface shell and a database. The user-interface is a computer program that remains essentially constant irrespective of which personality or information database is used. The database is what gives the conversation simulator its personality, knowledge, etc. It contains the specific answers and information about questions for a topic. The database has pre-defined answers linked together by question templates. The realisticness of the conversation simulator depends on how well the creator of the database has anticipated the questions people are likely to ask and the patterns that are common to classes of questions with the same answer. The user-interface accepts questions from a person, searches through the templates and returns the (or a random of the) most appropriate answer (or answers) corresponding
Eshelman Larry J.
Gutta Srinivas
Milanski John
Pelletier Daniel
Strubbe Hugo J.
Abebe Daniel
Koninklijke Philips Electronics , N.V.
Thorne Gregory L.
LandOfFree
Environment-responsive user interface/entertainment device... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Environment-responsive user interface/entertainment device..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Environment-responsive user interface/entertainment device... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3211555