Data processing: speech signal processing – linguistics – language – Linguistics
Reexamination Certificate
1999-10-07
2003-11-11
Edouard, Patrick N. (Department: 2654)
Data processing: speech signal processing, linguistics, language
Linguistics
C704S009000, C704S275000
Reexamination Certificate
active
06647363
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a method for automatically responding to an inquiry from a user; the method comprising:
executing a machine-controlled human-machine dialogue to determine a plurality of pre-determined query items specifying information to be verbally presented to the user;
retrieving a plurality of information items from a storage in dependence on the query items;
generating at least one natural language phrase to present the obtained information items according to a presentation scenario; and
verbally presenting the generated phrase(s) to the user.
The invention further relates to a system for automatically responding to an inquiry from a user; the system comprising:
means for executing a machine-controlled human-machine dialogue to determine a plurality of pre-determined query items specifying information to be verbally presented to the user;
means for retrieving a plurality of information items from a storage in dependence on the query items;
means for generating at least one natural language phrase to present the obtained information items according to a presentation scenario; and
means for verbally presenting the generated phrase(s) to the user.
2. Description of the Related Art
Automatic inquiry systems, for instance for obtaining traveling information, increasingly use automatic human-machine dialogues. Typically, a user establishes a connection with the inquiry system using a telephone. In a free-speech dialogue between the machine and the user, the machine tries to establish a number of query items required to obtain the desired information from a storage, such as a database. During the dialogue, the system typically issues an initial question. For a travelling information system, such an initial question might be “From which station to which station would you like to travel ?”. During the following dialogue, the system uses speech recognition techniques to extract query items from the user's utterances. For example, an automatic public transport information system needs to obtain the departure place, the destination place and a desired time/date of traveling to perform a query. The system may issue verifying/confirming statements to verify that it correctly recognized a query item. To establish query items for which no text has been recognized yet, the system may issue explicit questions, like “When would you like to leave ?”. The question may also be combined with a verifying statement. For example, in a situation where Amsterdam has been recognized as the departure place, but a departure time or specific departure station in Amsterdam is still unknown, the system could ask: “When would you like to leave from Amsterdam Central Station?”. Once all its essential query items have been recognized, the answers to the query are obtained from a storage. This results in a collection of information items, which are to be presented to the user in spoken form. Normally, textual presentations of the information items are inserted in a sequence of preformatted phrase or sentence templates, referred to as a presentation scenario. Usually, the templates are prosodically enriched. If no suitable speech presentation is available, for instance in the form of sampled phrases/words, the prosodically enriched text may be converted to speech using a speech synthesis technique. The speech output is fed back to the user. Depending on the nature of information items to be presented, a suitable presentation scenario may be chosen from a predetermined collection of scenarios. For instance, a different scenario may be used for any of the following situations:
no suitable connection was found,
one connection was found,
more than one connection was found,
the connection involves no changing over
the connection involves one change
the connection involves multiple changes.
Traditionally, inquiry systems were operated by human operators. The operator performed a dialogue with a user and entered query items into the system. The operator then performed a database query, and the results were displayed on a screen. The operator then read out the requested information to the user. The traditional inquiry systems tended to be self-centered or oriented towards the operator and lacking orientation towards the end-user. Automatic inquiry systems were build around these systems by adding a dialogue function and a presentation function. This sometimes exposed the lack of user-orientation of the system in the form of rigid or menu driven dialogue schemes.
From “Dialog in the RAILTEL Telephone-Based System”, Proceedings ICSLP 1996, Vol.1, pp.55-553, it is known that it is desired to take the user's intention into consideration during the human-machine dialogue. This may improve the recognition rate as well as confirmation process during the dialogue. It may also improve the selection process during the dialogue for systems wherein the user may choose between several sets of information items (e.g. information relating to a train or to a bus; main time table information, information relating to delays, etc) or may choose between several services, like presenting information or acting as an automated telephone switchboard.
SUMMARY OF THE INVENTION
It is an object of the invention to provide a method and system of the kind set forth, which is more user oriented, enhancing the acceptance of an automatically operating inquiry system.
To achieve the object, the method is characterized in that the method comprises, based on utterance(s) of the user, determining an intention of the user from a predetermined set of intentions; the intention reflecting a preferred way of presenting the information items; and selecting the presentation scenario from a predetermined set of presentation scenarios in dependence on the determined intention.
The inventor has realized that in many human-machine dialogues information is present, either explicitly or implicitly, that enables the system to determine or to infer certain intentions of the user that influence the way the information is to be presented most effectively. For instance, in many situations it is possible to derive from the dialogue, which of the information items retrieved during the query, are important to the user. In train travel information inquiries in general, the departure time will usually be of primary interest, allowing the user to determine when he should arrive at the departure station. However, if a user indicates that he wishes to travel from A to B and arrive around 9 A.M., it may be assumed that he is also very much interested in the exact arrival time at the station of destination. At least more so, when compared to users who ask for a connection from A to B that leaves around 9 A.M. Mentioning the arrival time early in the presentation will allow the user to check whether the arrival time of the connection indeed corresponds with his request, and to discard the incoming information if it does not. By choosing a presentation scenario which is tailored to the intention of the user, instead of selecting a scenario solely based on the information obtained as a result of the query, the user will feel more appreciated by the system. Consequently, the acceptance of the system will increase.
In systems which allow the user access to different sets of information or offer several services, the dialogue will normally involve determining in which set of information or in which service the user is interested. Such determining of the intention and acting upon it is not the subject of the invention. The invention relates to determining the intentions of the user that influence the way in which certain information items are presented effectively, and not to determining which set of information items or service the user wants to have access to.
In many cases it is not desired to pose additional questions to the user during the dialogue phase in order to determine intentions of the user that affect the way the information is presented most effectively. This will prolong the dialog
Bromberg & Sunstein LLP
Edouard Patrick N.
ScanSoft, Inc.
LandOfFree
Method and system for automatically verbally responding to... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for automatically verbally responding to..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for automatically verbally responding to... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3143208