JPH1055372A - On-demand interface device and computer-readable recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To retrieve a function matching an inputted request sentence in terms of semantics and easily execute the retrieved function. SOLUTION: From a function explanation data base 4 which holds a function explanatory sentence explaining a function and an action script showing a procedure of operation for actualizing the function in each item, a retrieval means 5 retrieves an item corresponding to a request sentence inputted from an input means 1. A script execution means 6 executes the action script corresponding to the retrieved item to execute the retrieved function. The retrieval means 5 performs key word retrieval or retrieval according to the distance between a sentence vector representing the semantics of the request sentence and the sentence vector of the function explanatory sentence in each item in the function explanation data base.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a help function for assisting a user's operation in a computer application, OA equipment, home electric appliance, and the like. The present invention relates to an on-demand interface device that can be achieved accurately and a computer-readable recording medium on which a program for realizing the function is recorded.

[0002]

2. Description of the Related Art In recent years, computer software such as a word processor or a home appliance such as a video deck displays an operation description of a function by performing a specific operation such as pressing a help key when the operation method is not understood. It has a function called an assist function or a help function. The help function selects and displays help information according to the usage status of the device at the time the help key is pressed. This allows the user to remember all operation methods,
Necessary functions can be easily achieved when needed.

However, in a system in which the number of functions is increasing due to the increase in the number of functions, it is difficult for a user to find a necessary function description by simply displaying all the function descriptions. For this reason, it is necessary to appropriately grasp the function requested by the user and display the corresponding operation description. Japanese Patent Application Laid-Open No. 7-121563 discloses that
After the user presses the help key, enter the keyword,
A help information search device for searching and displaying appropriate help information by keyword matching is described.

[0004]

In the help information retrieval apparatus described above, the user can achieve the desired function by executing an operation based on the displayed operation explanation. However, as the system becomes more multifunctional, it becomes difficult to design a system that can realize all functions with a simple operation procedure, and some functions require complicated operations that take multiple steps to realize them. It has become. In such a case, even if an operation explanation for realizing the target function is displayed, it is troublesome for the user to perform the operation according to the explanation.

[0005] Further, the above-mentioned help information search apparatus performs a search by keyword matching. For example, if the keyword (print, print, print) is given to the print function of a word processor, "Print"
Alternatively, when a request sentence of “print” or “print” is given, a print function is searched by matching keywords.
In addition to the exact matching of codes, the printing function can be searched even in a request sentence such as "purinto" or "printer" by absorbing inflection and ignoring the distinction between hiragana and katakana.

However, in the search by keyword matching, for example, when a request sentence "I want to print on paper" is given, (paper, launch) is obtained as a keyword from the request text, but a keyword (print , Print, print), the print function cannot be searched. Further, it is practically impossible to register in advance all keywords that are considered to be input by the user.
As described above, the method using the keyword matching has a limit in searching for a function that semantically matches the request sentence input by the user.

[0007] The present invention has been made in view of these problems, and an on-demand interface device capable of easily executing a function searched by a help information search, and a computer recording a program for realizing the function. It is an object to provide a readable recording medium. Further, the present invention provides an on-demand interface device that can search for a function semantically matching a request sentence input by a user and can easily execute the searched function, and a program for realizing the function. It is an object of the present invention to provide a recorded computer-readable recording medium.

[0008]

According to the present invention, the above object is achieved by providing means for automatically executing a searched function. That is, the on-demand interface device of the present invention holds input means for inputting a request sentence, a function explanation sentence explaining a function in each item, and an action script indicating an operation procedure for realizing the function. It is characterized by comprising a function explanation database, a search means for searching the function explanation database for an item corresponding to a request sentence, and a script execution means for executing an action script.

[0009] The search means may perform a search based on the degree of matching of the keyword between the request sentence and the function explanation sentence. The search means includes a word dictionary for associating a word with a word vector representing the meaning of the word, and vector generation means for generating a sentence vector representing the meaning of a sentence including a plurality of words using the word dictionary. The search can be performed based on the distance between the sentence vector of the sentence and the sentence vector of the function explanation sentence in each item of the function explanation database. In this case, the user can search for and execute a function that semantically matches the request sentence without knowing the keyword.

In order to function as the on-demand interface device of the present invention, a recording medium on which a program for causing a computer to execute each processing procedure may be used. That is, a program for executing a procedure (search procedure) for searching for a corresponding item from a function description database for a request sentence in order to function as the above-mentioned search means, and an action script for causing the program to function as the above-mentioned script execution means By using a recording medium on which at least a program for executing the procedure for executing the function is recorded, it is possible to automatically execute the searched function.

[0011] It is possible to use a recording medium storing a program for causing the search means to perform a search based on the degree of matching of a keyword between a request sentence and a function explanation. The search means may function as a word dictionary for associating a word with a word vector representing the meaning of the word, a vector generating means for generating a sentence vector representing a meaning of a sentence including a plurality of words using the word dictionary, It is possible to use a recording medium that records a program for causing a search to be performed based on the distance between the sentence vector of the request sentence and the sentence vector of the function explanation sentence in each item of the function explanation database.

[0012]

Embodiments of the present invention will be described below. Here, an on-demand interface device of a word processor will be described as an example. FIG. 1 is a schematic configuration diagram showing an example of an on-demand interface device according to the present invention. The on-demand interface device includes an input unit 1 such as a keyboard for inputting a request sentence, an output unit 2 such as a display, a control unit 3, and a function explanation database 4. The control unit 3 implements the functions of a search unit 5 and a script execution unit 6 described later by hardware or software.

FIG. 2 shows an example of the function explanation database 4. The function explanation database 4 contains "centering",
Word processor function names such as "print" and "save," along with descriptions of those functions, instructions on how to implement the functions,
And an action script written in a format that can automatically execute the operation procedure.

FIG. 3 is a diagram for explaining a schematic configuration of the script execution means 6 and its operation. The script execution means 6 includes a command correspondence table 21, a buffer 22 for temporarily storing a command sequence (action script),
And a command execution unit 23. The command correspondence table 21 stores, for example, command names of scripts arranged in alphabetical order and addresses in a program list 25 of all word processing software of a routine for executing the commands.

As shown in the figure, for example, the script command “Ge
The program routine corresponding to "tString ()" is 205
You can see that it starts at address 50. Here, 10023
The routine starting from the address is a menu opening routine.
Specify the menu item to open as an argument. The routine starting from the address 18802 is a routine for selecting an item from the opened menu, and specifies the item to be selected as an argument. 20
The routine starting at address 550 is a routine for obtaining character string information, issues a message prompting the user to specify a character string, and waits until the user specifies the character string with a keyboard or a pointing device such as a mouse.

The buffer 22 temporarily stores a command sequence to be executed. The command execution unit 23 reads the script commands stored in the buffer 22 in order, finds the address corresponding to the command by referring to the command correspondence table, and transfers control to the address. The above operation is repeated until the end of the command in the buffer.

FIG. 4 is a flowchart for explaining the operation procedure by the on-demand interface device of the present invention. When the user who cannot understand the operation of the word processor presses the help key (S11), the editing function of the word processor is displayed together with the explanation of the function by the conventional help function according to the current state of the word processing operation (S1).
2). The function displayed here may be one or plural. If a desired function is found therein (S1
3) It is determined whether or not to execute the function (S1).
9). When executing, if a plurality of functions are displayed, a desired function is selected from the displayed functions (S
20). When a function is selected, the search means 5 stores the action script of the selected function in the buffer 22 (S21). After that, the script execution means 6
The function is realized by sequentially executing the scripts stored in the buffer 22 (S22). If it is determined in step 19 that the execution is not requested only by confirming the function, the process ends.

In step 13, if there is no information on the desired function in the information displayed by the conventional help function, the control means 3 prompts input of a request sentence from the input means 1 (S14). If the request sentence has not been input, the flow ends (S15). When the request sentence is input, the search unit 5 searches the function description database 4 for a function that matches the request sentence (S16), and displays the searched function name and its function description on the output unit 2 (S17). .
The function displayed here may be one or plural. If there is a desired function among the displayed functions (S18), it is determined whether or not to execute the function (S19).

When executing, if a plurality of functions are displayed, a desired function is selected from them (S2).
0). When a function is selected, the search means 5 stores the action script of the selected function in the buffer 22 (S21). Thereafter, the script executing means 6 implements the function by sequentially executing the scripts stored in the buffer 22 (S22). If it is determined in step 19 that the function is not confirmed but only executed, the process is terminated. If there is no desired function among the functions displayed in step 18, the process returns to step 14 again, and prompts for another request sentence.

Next, an example in which the search means 5 performs a search by a keyword search method will be described more specifically. Here, the "centering" function in word processing editing will be described as an example. As described in the function description database shown in FIG. 2, the "centering" function is a function of moving a designated character string to the center of a line, and the operation procedure is a keyword or a pointing device such as a mouse. To specify the character string, and then select the item "centering" from the menu item "edit".

It is assumed that the word processing user tries to center the document title and does not understand the operation method, and presses the help key (S11). Here, it is assumed that the function explanation is displayed by the conventional help function (S12), but there is no information on "centering" in the help function (S13).
Next, it is assumed that the user has input “move to center” in response to the input request in step 14. Search means 5
Function explanation database 4 by request sentence "move to center"
Is searched (S16).

In the search, in this example, Kanji and Katakana character strings are extracted as keywords from the whole sentence of the request sentence and from the function name and the function explanation of the function explanation database, and used for the search. Therefore, (center, move) is extracted from the request sentence as a keyword, and (centering, character, line, center, move) is extracted from the function item of “centering” in the function description database. Similarly, (print, document, printer, print) is extracted as a keyword from the function item “print”, and (save, document, file, save) is extracted as a keyword from the function item “save”. Search is based on the AND of the keywords in the request sentence
Performed by search. In this case, the function item “centering” is searched because both keywords (center, move) extracted from the request sentence match, and the function items “print” and “save” are not searched because the keywords do not match.

Hereinafter, keyword matching is performed with all the function items in the function description database, and the following operation will be described on the assumption that the function item searched as a result is only "centering". In step 17, the searched function item “centering” is displayed on the output means 2.
Since this function is exactly required by the user, the process proceeds from step 18 to step 19, and it is assumed that execution is selected in step 19. Then, the search means 5 copies the action script of the searched function item "centering" to the buffer (S21). Therefore, the buffer 22
In the table, the following command string is stored. (GetString (), OpenMenu (HENSYU), Choose (CENTERING)) Next, the script executing means 6 sequentially executes the commands stored in the buffer 22 (S22). That is, Ge
Transfer control to address 20550 with tString (), then OpenMe
With nu (HENSYU), control is transferred to address 10023 with HENSYU as an argument, and finally, control is transferred to address 18802 with Choose (CENTERING) with CENTERING as an argument. Where the argument
HENSYU and CENTERING mean “edit” and “centering” respectively.

The operation of the script execution means 6 will be described once again together with the operation of each routine described above.
First, the user is prompted to specify a character string at address 20550, and waits until the user specifies the character string. Next, the menu item "edit" corresponding to the argument HENSYU is opened at address 10023. Finally, the item "centering" corresponding to the argument CENTERING is selected at the address 18802. This series of actions
This is the same as the operation procedure of the “centering” function executed as it is. That is, the “centering” function is automatically executed by the script execution means 6.

In the above, the process in which the function item "centering" is retrieved from the request sentence "move to the center" and the action script representing the processing is automatically executed has been described. Thus, when you enter the request statement "move to center",
The function item "centering" is searched by the search means,
Subsequently, if requested by the user, the "centering" function can be automatically executed by the script execution means. Here, the description has been limited to the "centering" function, but the same applies to other functions. Particularly, for functions requiring complicated operations, the script execution means substitutes the operation procedure for the burden on the user. Is greatly reduced.

In the above description, it is described that one function item is searched. However, when a plurality of function items are searched at the same time, the user is allowed to select an appropriate one from among them and to select one item. After that, the action script is copied and executed as described above. If the search is not performed by the AND search, a process such as performing an OR search may be performed. A keyword may be given to each function item in advance, and may be stored in another buffer.

Next, an example in which vector retrieval is used as retrieval means will be described. FIG. 5 is a schematic configuration diagram of the vector search means, and includes a word dictionary 31 that holds word vectors,
The search unit 32 includes a search unit 32 that performs a search based on a vector, a vector generation unit 33 that generates a vector, and a storage unit 34 including a buffer that temporarily stores information necessary for the search.

FIG. 6 shows a part of the word dictionary 31. The word dictionary 31 holds a word vector representing the meaning of each word. The word vector expresses the meaning of the word in a distributed manner in association with the N feature words. As an example, a list of feature words is ｛text,
When the document, position, space, print, move, shape, memory, and display are defined as｝, the relevancy between a certain word and these N feature words can be multi-valued or most simply 2 depending on the degree of the relevance. The N-dimensional vector generated by expressing the value is the word vector of the word.

For the sake of simplicity, a description will be given of an example in which the relevance to a feature word is expressed by a binary value indicating the presence or absence of a relationship. Has a relationship with “1”, and has a low relationship with the characteristic words “position”, “space”, “print”, “move”, and “memory”, and thus becomes “0”. , 0, 0, 0, 0, 1, 0, 1).
Regarding the creation of the word vector, see the IEICE Tech.
9 (1993-1), "Associative search from a large-scale document database" (The Institute of Electronics, Information and Communication Engineers) proposes a method for constructing a large-scale word vector that reflects human knowledge by using an encyclopedia. .

Next, the sentence vector will be described. Take the sentence "Print a document on a printer" as an example. A sentence vector extracts words contained in a sentence,
The word vectors are added and then normalized. As a method of extracting a word, there is a method of simply extracting a character string using a word dictionary and a method of using morphological analysis.

In the present case (document, printer, printing)
Two words are extracted. These word vectors are (1,1,0,0,0,0,0,0,0), (0,
1,0,0,1,0,0,0,1), (0,1,0,
0, 1, 0, 0, 0, 1). Adding them together gives (1,3,0,0,2,0,0,0,2),
The normalized version of this is "print the document on a printer"
Is a sentence vector of the sentence Here, the normalization means that the vector is made to have a fixed length. In this example, when the length of the vector is normalized to 10, a sentence vector is obtained as V in the following equation.

V = (10 / L) (1,3,0,0,2,0,0,0,2) = (2.4, 7.1, 0.0, 0.0, 4.7, 0.0, 0.0, 0.0, 4.7) L = (1 ² + 3 ² + 0 ² + 0 ² + 2 ² + 0 ² + 0 ² + 0 ² + 2 ^{2) 1/2} = 18 ^1/2 will be described search method. When two sentences are given, when the directions of the sentence vectors are close to each other, it is determined that the similarity between the two sentences is high. Since the vector is normalized, the closeness of the direction is represented by the size of the inner product. That is, when the inner product of two sentence vectors is large, the similarity between the two sentences is high. Therefore, the inner product of the sentence vector of the request sentence and the sentence vector of each item is determined, and the larger one is determined as the search result.

FIG. 7 is a flowchart showing the flow of processing in the search means using vector search. First, the vector generation unit generates the sentence vector QV of the request sentence from the request sentence as described above (S31). Next, the index i is initialized to 1 (S32), and the function description database 4
Taking out the function description of the i th function item from, generates the sentence vector BV _i (S33). Calculate the inner product p _i of QV and BV _i, stored in the buffer of the memory unit 34 (S
34). If i points to the last function item, step 3
7 (S35), otherwise increase i by 1 (S36) and return to step 33. After calculating the inner product of all the function items with the sentence vector (S35), the largest one among the inner products p _j stored in the buffer is selected (S37), and the function item corresponding to the index number is searched. The result is obtained (S38).

As an example, a search in the case where a request sentence “I want to print on paper” is given will be described. First, two words (paper, embossed) are extracted from the request sentence. However,
In the word dictionary, the inflected words were registered as stems. Next, a sentence vector QV of the request sentence is generated by adding these word vectors and normalizing them (S31). 2
The word vectors of the two words (paper, embossed) are (1,1,0,0,1,0,0,0,1), (0,1,
0, 1, 1, 0, 0, 0, 0), the sentence vector QV of the request sentence is obtained as in the following expression.

QV = (10 / L) (1,2,0,1,2,0,0,0,1) = (3.0, 6.0, 0.0, 3.0, 6.0, 0.0, 0.0, 0.0, 3.0) L = (1 ² +2 ² +0 ² +1 ² +2 ² +0 ² +0 ² +0 ² +1 ² ) ^1/2 = 11 ^{1/2 At} steps 33 and 34, the sentence vector QV obtained above and each item of the function explanation database are described. The inner product of the function explanation and the sentence vector is calculated. Here, as an example, the inner product of the function item "centering" and the function item "print"
Let's calculate the inner product of First, for the function item “centering”, four words (character, line, center, move) are extracted from the function description “move character to center of line”, and the word vector of the four words is Each (1,
1,0,0,0,0,1,0,1), (0,1,1,
1,0,0,0,0,0,), (0,0,1,1,0,
0,0,0,0,), (0,0,1,1,0,1,0,
0,0), and the sentence vector BV is given by the following equation.

BV = (10 / L) (1,2,3,3,0,1,1,0,1) = (2.0,3.9,5.9,5.9,0.0,2.0,2.0,0.0,2.0,) inner product p of ^{L = (1 2 +2 2 +3} 2 +3 2 +0 2 +1 2 +1 2 +0 2 +1 2) 1/2 = 26 1/2 QV and BV is calculated by the following equation. p = 3.0 × 2.0 + 6.0 × 3.9 + 0.0 × 5.9 + 3.0.
0 × 5.9 + 6.0 × 0.0 + 0.0 × 2.0 + 0.0 × 2.
0 + 0.0 × 0.0 + 3.0 × 2.0 = 53.1 Next, a sentence vector BV is generated from the function explanation “Print document” on the function explanation “Print”. Has already been sought as an example in (2.4,7,1,
0.0, 0.0, 4.7, 0.0, 0.0, 0.0, 4.7). The inner product p of QV and BV is calculated as in the following equation. p = 3.0 × 2.4 + 6.0 × 7.1 + 0.0 × 0.0 + 3.0.
0 × 0.0 + 6.0 × 4.7 + 0.0 × 0.0 + 0.0 × 0.0.
0 + 0.0 × 0.0 + 3.0 × 4.7 = 92.1 From the above, in this case, the inner product of the request statement and the function item “centering” is 53.1, and the inner product of the function item “print” is 92.1, and it is determined that the request sentence “I want to print on paper” has a higher similarity to “print” than “centering”. If there is no other function item whose inner product with QV is 92.1 or more, "print" is obtained as a search result. After that, similarly to the case of the keyword search described above, the name of the searched function and the description of the function are displayed, and when the user selects the execution, the action script of “print” is copied to the buffer 22. , The script execution means 6 executes the scripts in the buffer in order.

In this way, the search means searches for a sentence that semantically matches the request sentence, and is automatically executed. In this example, a sentence vector is generated from the function description of each function item each time the search is performed. However, the sentence vector of the function description may be generated in advance and stored in another buffer. In this case, the step 33 of generating a sentence vector from each function item becomes unnecessary, and the sentence vector may be simply taken in from the buffer, and the processing speed can be increased. Also,
In this example, only the top searched features are displayed,
A plurality of function items having a large inner product may be displayed to allow the user to select.

Further, in order to function as the on-demand interface device of the present invention, a program for executing the above-described processing is stored in a floppy disk or a CD.
The information may be recorded in a computer-readable recording medium such as a ROM in advance, and may be used by being installed in a computer.

[0039]

According to the present invention, in the case where the operation of the device is not understood, the function of requesting by simply inputting the request statement without performing the necessary operation can be achieved.
Work efficiency can be greatly improved. In addition, if a search using a feature vector representing the meaning of a sentence is used, even if the user does not know the keyword, a function that is semantically applicable to a given request sentence is searched and automatically executed. , The user interface is greatly improved over the conventional help function.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of an on-demand interface device according to the present invention.

FIG. 2 is an explanatory diagram of a function explanation database.

FIG. 3 is a diagram for explaining the configuration and operation of a script execution unit.

FIG. 4 is a flowchart illustrating an operation procedure.

FIG. 5 is a configuration diagram of a vector search unit.

FIG. 6 is a diagram showing a part of a word dictionary.

FIG. 7 is a flowchart showing the flow of processing of a search unit.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input means, 2 ... Output means, 3 ... Control means, 4 ... Function explanation database, 5 ... Search means, 6 ... Script execution means, 21 ... Command correspondence table, 22 ... Buffer, 23 ...
Command execution unit, 25: All program lists of word processing software, 31: Word dictionary, 32: Search unit, 33: Vector generation unit, 34: Storage unit

Claims (6)

[Claims] An input means for inputting a request sentence; a function explanation database for holding a function explanation sentence explaining a function in each function and an action script indicating an operation procedure for realizing the function; Search means for searching for a corresponding item from the function description database for a sentence;
An on-demand interface device comprising: script execution means for executing the action script. 2. A function description database for storing input means for inputting a request statement, a function description statement explaining a function in each function, and an action script indicating an operation procedure for realizing the function. On the other hand, a computer-readable recording medium on which a program for functioning as a search unit for searching a corresponding item from the function description database and a script execution unit for executing the action script is recorded. 3. The on-demand interface device according to claim 1, wherein said search means performs a search based on a degree of matching between a request sentence and a function explanation sentence. Interface device. 4. A computer-readable recording medium according to claim 2, wherein a program for causing said search means to perform a search based on a degree of matching between a request sentence and a function explanation sentence is recorded. Computer readable recording medium. 5. The on-demand interface device according to claim 1, wherein the search means includes a word dictionary for associating a word with a word vector representing the meaning of the word, and a plurality of words using the word dictionary. Vector generating means for generating a sentence vector representing the meaning of the sentence, and performing a search based on the distance between the sentence vector of the request sentence and the sentence vector of the function explanation sentence in each item of the function explanation database. An on-demand interface device, comprising: 6. The computer-readable recording medium according to claim 2, wherein said search means includes a word dictionary for associating a word with a word vector representing the meaning of the word, and a plurality of words using the word dictionary. Function as a vector generating means for generating a sentence vector representing the meaning of the sentence,
A computer-readable recording medium storing a program for causing a search to be performed based on a distance between a sentence vector of the request sentence and a sentence vector of a function explanation sentence in each item of the function explanation database.