JP2003330923A - Automatic conversation device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide simple sentences for continuing a conversation while preparing for corresponding properly to a human's talk, avoiding the conversation from being interrupted. <P>SOLUTION: The automatic conversation device has two different types of databases, a database relating to common words and a database relating to proper nouns. The database relating to common words includes simple conversational sentences, and the database relating to proper nouns includes mainly explanatory items such as definition of a word. The device outputs simple sentences while forming a conversation by retrieving the data of the proper noun. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to an interactive robot having a topic database.

[0002]

2. Description of the Related Art Conventionally, there has been an interactive robot which analyzes a human voice and replies to the content of the voice or provides a new topic related to the content. In the past, only simple conversations were possible, so it was possible to respond immediately by listening to human voices, but in order to make complicated conversations, it is necessary to equip the robot with a large database that is a topic of conversation. There is. It takes a long time to search this database, so there is a blank time when you do not speak anything.
The flaw is that the conversation does not proceed smoothly.

[0003]

SUMMARY OF THE INVENTION The present invention solves the above-mentioned drawbacks, and the connection is maintained even while the robot searches for topical data so that the conversation between the human and the robot becomes smooth without interruption. I tried to talk.

[0004]

In order to achieve the above object, the automatic interactive apparatus of the present invention comprises a receiving means for receiving the first data, an analyzing means for analyzing the received contents, and a data for storing the second data. Storage means, data retrieval means for retrieving the second data, output means for outputting the second data and third data, respectively, and parallel for outputting the third data during retrieval of the second data It is characterized by having a processing means.

Further, when the retrieval of the second data is completed and the preparation for outputting the second data is completed, the third data is retrieved.
It is characterized by having a canceling means for canceling the output of the data.

Further, when the second data is searched or the third data is output, the second and third data are output.
Data is data related to the first data.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings.

(Outline) First, an outline of the operation of the proposed apparatus will be described. FIG. 8 shows how the present apparatus responds to the contents spoken by humans. Suppose a human being said, "I read Shuhei Fujisawa's novel today" (801).
This device analyzes a human story and extracts some words.
A corresponding sentence is prepared for each word (802).

However, there are some words that can be immediately prepared for correspondence and some that are not. "Today""Novel""Read"
Although the general noun and verb can be prepared immediately, the personal name "Shuhei Fujisawa" is a proper noun, so we must search the encyclopedia database and collect data on this personal name (803).

However, since the main point of the story depends on this person's name, it is necessary to prepare the data. In the meantime, the device needs to connect. For that, use sentences about general nouns that can prepare the story quickly. For example, the story is connected by a sentence such as “What kind of work did you read?” (804). The above is the outline of the operation of this device.

(Block Diagram) FIG. 1 is a block diagram of an automatic dialogue apparatus according to the present invention.

First, 101REC is a receiving device for receiving a human voice or a sentence to this device which may be either a voice or a character code signal. In the case of voice, it has a function of converting into a character code signal.

The CPU 102 is a central processing unit, which performs control and calculation for analyzing based on the received signal, searching a database described later, and outputting.

Reference numeral 103SEN is a transmission device for outputting data as voice or characters from this device. In the case of voice, the character code is converted into a voice signal, and in the case of displaying in characters, a font and a display device are provided and displayed in a character pattern. This is a key switch for instructing the area person, and details will be described later.

The 104ROM is a read-only memory that stores a program for operating this apparatus.

Reference numeral 105HD is a database device for storing data. Here, we have a database that corresponds to common nouns and a dictionary of proper nouns.

The 106 RAM is responsible for a work area used for temporary storage and calculation of transmitted / received data.

Reference numeral 107BL is a bus line for connecting the above devices and exchanging information.

(Details of Database) Details of the above-mentioned database will be described. As shown in FIG. 4, the database is composed of a unique data part (401) and a general data part (402). The proper data is composed of a word associated with a proper noun and its description. The general data is composed of words other than proper nouns and transmission data related thereto.

FIG. 5 shows an example of unique data. For example, assume that there is an item (501) by a novelist Shuhei Fujisawa. A plurality of words related to this proper noun are gathered as data (502). For example, first being a "novelist" (503). This is a definition. Next, items such as historical novels, Japanese literature, Yamagata prefecture (hometown) follow.

FIG. 6 is a general word database. First, there is a word (601) as a headline. Although a group of data (602) is stored so as to correspond to this word, unlike the unique data, this is not the definition of the word but the conversational sentence related to the word is stored. For example, when it comes to the word "novel," it sounds interesting.
(603) There is a conversation sentence such as "Is it a long novel?"
These conversations are used as time links.

(Operation Processing) An example of operation processing of this apparatus is shown in FIG.
~ It demonstrates with reference to the flowchart of FIG. The program for this operation is stored in the ROM.

When the power of the device is turned on, the process begins with receiving a human story. The story may be input as text instead of voice. The data once received is converted into a character code and stored in the work area of the RAM (step S
201).

The input received sentence is decomposed into words, and the content itself of the sentence is analyzed. As shown in 802 of FIG. 8, items corresponding to each word are organized (step S20).
2).

As a result of the analysis, it is determined whether the received story can be immediately returned with a simple content such as a simple greeting or a complicated content that requires a long time to reply (step S20).
3).

If the content is simple, a reply sentence is created and transmitted to the transmission device. As a result, the present apparatus speaks or displays a sentence to a human (step S2).
04).

After that, the process returns to the receiving process again.

On the other hand, if the story is complicated and an immediate reply cannot be made, an interrupt process is generated and a reply sentence is assembled (step S205).

Here, the interrupt processing will be described with reference to the flowchart of FIG. After the interruption process, the database is searched (step S301) and a reply sentence is created (step S302).

After that, a preparation completion signal is issued (step S303), and the interrupt processing is terminated.

On the other hand, during interrupt processing,
In this process, the connected conversation is output. In this connected conversation, the conversation sentence derived from the general word is output as it is (step S206).

Next, it is judged whether or not the interrupt processing is completed (step S207).

If the interrupt processing is not completed, the connected conversation is continued. Conversational sentences are based on words obtained from received signals. When the interrupt is completed, the transmission process based on the database is performed. (Step S208) It is possible to transmit a connected conversation and have a place while forming a full-fledged conversation sentence in the above process.

(Other Embodiments) In the above embodiment,
As a connecting conversation, I tried to use a conversation sentence based on the received word, but it can not always be used to use the same conversation sentence, so after outputting the conversation sentence about the received word all the other words Send a conversation sentence. For example, as shown in Fig. 7, the received sentence is "I read a novel written by Shuhei Fujisawa today."
In that case, while searching the database for Shuhei Fujisawa,
702, 7 of the conversational sentences of the common words "today" and "novel"
The output is 03, 704, and 704, but when this ends, the conversational sentence to be output is interrupted. Then, paying attention to the word "interesting" in 704, the conversation sentence of this word is further extracted and output as in 706. In other words, the word is pulled out and the conversation sentence is drawn out endlessly.

[0035]

As described above, according to the present proposal, the robot talks while the robot searches for the topic data so that the conversation between the human and the robot can be smoothly and smoothly. I chose

[Brief description of drawings]

FIG. 1 is a block diagram of the present apparatus.

FIG. 2 is a flowchart of processing.

FIG. 3 is a flowchart of processing.

FIG. 4 is a conceptual diagram of a database.

FIG. 5 is a block diagram of a proper noun database.

FIG. 6 is a block diagram of a general word database.

FIG. 7 shows an example of a connected conversation.

FIG. 8 is a process diagram from a human story to a robot output.

Claims (3)

[Claims] 1. Receiving means for receiving first data, analyzing means for analyzing received contents, data storing means for storing second data, data retrieving means for retrieving the second data, and second retrieving means. An automatic dialogue apparatus comprising: output means for outputting data and third data respectively; and parallel processing means for outputting the third data during retrieval of the second data. 2. The retrieval of the second data is completed and the second data is retrieved.
2. The automatic dialogue apparatus according to claim 1, further comprising a stopping unit that stops outputting the third data when the preparation for outputting the data is completed. 3. The second and third data are data related to the first data when the second data is searched or the third data is output. The automatic dialogue apparatus according to claim 2.