JPH07244592A - Expert system - Google Patents

Abstract

PURPOSE:To make it unnecessary to use an exclusive language by providing this expert system with a link preparing device for extending a link having a meaning by the use of a hyper-text and an inference executing device for using a linked document as a knowledge base. CONSTITUTION:The expert system includes the link preparing device 4 for extending a link having a meaning of user specification to a document 1 and the inference executing device 5 for executing execution specified by the linked document. In the case of executing the same processing as the document 1 by the expert system, the meaning of the document is inputted by a format to be processed by the system. A user inputs the meaning of the document 1 by the format to be processed by the system and the device 5 interpretes the meaning and executes the execution specified by the document 1. When necessary, data to be outputted from the device 5 are sent to an output device 6 and outputted. When the user extends a meaningful link, the expert system using the document as a knowledge base can be constructed.

Description

Detailed Description of the Invention

[0001]

This invention relates to hypertext and expert systems in computer systems.

[0002]

2. Description of the Related Art An expert system is constructed so that the knowledge and know-how of a certain business expert can be incorporated into a computer as a knowledge base so that the user can solve problems in a sequential response format. It is a computer system. The knowledge possessed by a specialist is usually held as a knowledge base by a computer engineer who obtains it from an expert by a method such as an interview and translates it into a dedicated language that can be processed by the computer system. At this time, the engineer needs to draw out the full knowledge of the expert and accurately translate it into a dedicated language. Then, the inference system refers to the condition given by the user and the knowledge base to solve the problem.

Further, in the conventional expert system using hypertext disclosed in Japanese Patent Laid-Open No. 2-263236, the hierarchical relation can be expressed by hypertext. FIG. 25 is a diagram showing the operation of the expert system disclosed in Japanese Patent Laid-Open No. 2-263236. This expert system comprises an upper state including at least one lower state in an expert system having a display section for displaying a window having text information and an input designating section for designating an upper side of the displayed text. It is characterized in that a hypertext control means for expressing a hierarchical relation of data to be expressed by a hypertext using a card is provided and a specific state is determined through the displayed hypertext card.

Cards that make up hypertext include:
The upper status and the lower status connected to it are displayed. When the upper state is determined to be a specific state by the user's designation, the system sets the lower state by inference, creates a card indicating the set lower state, and displays it on the screen. Creating a card that shows the subordinate state
The user interactively answers questions of various cards, and is performed each time one state is determined. Therefore,
By finally confirming the answers to the questions on the displayed card, it is possible to finally set a certain state.

[0005]

In the conventional expert system as described above, a computer engineer draws knowledge from an expert in order to construct a knowledge base, and translates it into a dedicated language that can be processed by an inference system. There was a need. This requires (1) a great amount of labor and time by the engineer, (2) if the knowledge acquired from the expert is ambiguous, it cannot be translated into a dedicated language, or the knowledge cannot be correctly interpreted. (3) Experts cannot accurately convey their own knowledge to engineers, (4) Expert knowledge is confidential and cannot be disclosed to outsiders,
(5) There is a problem that a computer engineer must be called in order to change or add knowledge.

Further, in the conventional expert system disclosed in Japanese Patent Laid-Open No. 2-263236, the answers can be sequentially determined by interactive operation using hypertext, but the data has a hierarchical relationship. There is a problem that it can be applied only when the user has it.

The present invention has been made to solve the above problems, and an object thereof is to construct an expert system having a knowledge base that does not require the use of a dedicated language by using a document in the hypertext format. I am trying. Another object of the present invention is to construct an expert system having a knowledge base that does not require the use of a dedicated language using hypertext for not only documents but also tables or flowcharts. Further, the present invention aims at constructing an expert system having a knowledge base that does not need to use a dedicated language by using hypertext for a program or a database.

[0008]

A first aspect of the present invention is characterized by a link creating device that creates a semantic link using hypertext and an inference execution device that uses a document with a meaningful link as a knowledge base. .

The second invention regards one unit of the document structure such as the chapter or section of the document used as the above-mentioned knowledge base as one knowledge, and when the title of the sentence structure is selected, the knowledge is related to the knowledge. It performs inference.

In a third invention, a condition link is attached to the item in the table in the document used as the above-mentioned knowledge base, and only those satisfying the condition are taken out from the data in the table by the link. It processes only data.

A fourth aspect of the invention relates to the flow (flow chart) in the document used as the above-mentioned knowledge base, links data reference or program execution to each item of the flow, and refers to the data or program execution referred to by the link. Depending on the return value of, the process proceeds to the next item indicated by the flow, and finally the selected process is performed.

According to a fifth aspect of the present invention, a condition is linked to the items in the table, and only the data satisfying the condition is extracted from the data in the table by the link, and the process is performed only on the data.

According to a sixth aspect of the invention, a link for data reference or program execution is linked to each item of the flow, and the data referred to by the link or the return value of the program execution advances to the next item of the flow and finally. It performs the selected process.

A seventh aspect of the present invention generates a thesaurus consisting of a relation name and an attribute name in a relational database deeply related to the document used as the knowledge base, and has the same meaning as the relation name and the attribute name in the thesaurus. By linking the keywords in the document knowledge base, the database can be searched from the document.

An eighth aspect of the present invention sets search conditions for a document when searching the relational database from the document used as the knowledge base.

A ninth aspect of the present invention enables execution of an application program from a document by linking a keyword in the document used as the above-mentioned knowledge base and an application program name indicated by the keyword.

A tenth aspect of the present invention is to combine the seventh and ninth aspects of the present invention to link the linked application program and the database, thereby inputting the data of the database to the application program executed from the document. , It is possible to store the execution result of the application program in the database.

In the eleventh invention, a keyword in the document used as the above-mentioned knowledge base and a word in the thesaurus of the relational database, or a character string when the application program name is the same word or a synonym word, By linking with, the work of linking the user is supported.

[0019]

In the expert system according to the first aspect of the invention, a hypertext-format document is provided with a link having the meaning specified by the user, and the link having the meaning on the document is referred to.
Executes the process indicated by the meaning link.

In the expert system according to the second aspect of the invention, one unit of document structure such as a chapter or a section in a hypertext format document in which a meaning link is set by the user is regarded as one knowledge, and the unit of knowledge is considered. The processing indicated by the meaning link is executed for each.

In the expert system according to the third aspect of the invention, for a table in a document, a condition link for the data in the item is attached to the item of the table, and the condition is selected from the data in the table to which the condition link is attached. Only those that satisfy the above are taken out, and only the taken out data is the target of the processing for the table.

In the expert system according to the fourth aspect of the invention, with respect to the flow (flow chart) in the document, a link for data reference or program execution is attached to each item of the flow, and the return value of the data or program execution is referenced by referring to the link. Then, the next item to be processed is referred to, and the same process is performed on the item. By repeating this, the last item is selected and the process indicated by the link of the meaning put on the item is executed.

In the expert system according to the fifth aspect of the invention, even if a table exists independently, a condition link for the data in the item is attached to the item of the table, and a condition link is added in the table. Only those that satisfy the condition are extracted from the data, and only the extracted data is subject to the processing for the table.

In the expert system according to the sixth aspect of the invention, even if a flow exists independently, a link for data reference or program execution is linked to each item of the flow, and the link is referenced to return data or program execution. The value is obtained, the next item to be processed is referred to by the value, and the same process is performed on the item. By repeating this, the last item is selected and the process indicated by the link of the meaning put on the item is executed.

In the expert system according to the seventh aspect of the invention, in order to combine the document and the relational database, a thesaurus consisting of the relational name and the attribute name is generated from the relational database, and the document having the same meaning as the relational name and the attribute name is generated. Automatically links with the keyword inside. Then, the relation name and attribute name of the link destination are obtained from the link of the keyword in the document, and the database is searched with this information.

In the expert system according to the eighth aspect of the present invention, by linking the condition specified by the user to the keyword in the document linked to the relation name and attribute name of the database, the database on the document is linked. Set search conditions.

In the expert system according to the ninth aspect of the present invention, the keyword in the document and the application program name indicated by the keyword are automatically linked to execute the application program from the keyword in the document.

The expert system according to the tenth aspect of the invention is executed from the document by establishing a user-specified database search link in the character string portion indicating the input or output of the application program name linked to the keyword in the document. The data of the database is input to the application program or the execution result of the application program is stored in the database.

In the expert system according to the eleventh invention, when the keyword in the document, the word in the thesaurus of the relational database and the application program name are the same word or synonyms, the character strings are automatically linked. To help users find words that have the same meaning when linking.

[0030]

【Example】

Example 1. Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an expert system according to this embodiment. The expert system has a link creating device 4 for creating a link having a meaning specified by a user with respect to the document 1 and an inference execution device 5 for executing the execution shown in the document with the semantic link.

Next, the operation will be described. The document in FIG. 1 “If the patient's body temperature is 37 ° C. or higher, there is a suspicion of a cold.” Is a sentence that means a judgment to diagnose the illness of the patient. If you want the expert system to do the same, first enter the meaning of this document in a format the system can handle. The user uses the link creation device 4 to input the meaning of the document in a format that can be processed by the system, and the inference execution device 5 interprets it to perform the execution indicated therein. Then, if necessary, the data to be output from the inference execution device 5 is sent to the output device 6 and output.

FIG. 2 shows the format in which the document "If the patient's body temperature is 37 degrees or higher, there is a suspicion of a cold." This is a meaning link for causing the inference execution device 5 to perform a process of outputting a character string "I have a suspicion of a cold" when the variable "body temperature" indicating the body temperature of the patient is 37 degrees or more. It is a stretched line.
First, the condition link (>) 11 extends from “37” to “body temperature”, which indicates that the condition link is true when the body temperature is 37 or more. Also, from "body temperature", the reference link 12 is the variable "body temperature".
Indicates that the number assigned to this variable when the “body temperature” in the document is referred to by the condition link 11 is changed to the “body temperature” in the document for reference. In the case of FIG. 2, the numerical value of 37.5 is assigned to the variable “body temperature”, and the condition link 11 that “37.5” is “37” or more is true. Further, the output link 13 is stretched from "body temperature" to "suspected of cold". This is "suspected of cold" when the condition link 11 stretched to "body temperature" becomes true. Is output. Here, the output means to send the character string "There is a suspicion of a cold" to the output device 6 of FIG.

FIG. 3 is a block diagram showing a link creating device. The link creation device is provided with a menu display unit 401 as a user interface for creating a link. Further, it is provided with a variable setting means 402 for setting the variable 2 based on the menu displayed by the menu display means. Also, the menu display means 401
Condition link means 403 for setting a condition link and reference link means 40 for setting a reference link based on the operation of
4, output link means 405 for setting an output link, and output designating means 406 for outputting a keyword or data of a document.

FIG. 4 shows a user interface of the link creating device 4. The user actually links to the document using this user interface. The "condition link" displayed in the menu 31 of FIG. 4 is used when a condition on the document is desired to be set. In the example of the document here, the condition that the body temperature is 37 or more is set. In this case, a condition link is created from "37" in the document to "body temperature". In the actual procedure, first, the range of the character string "37" in the document is designated. Next, the “condition link” of the menu 31 is selected. Then, the condition link menu of FIG. 5 is displayed.
From this condition link menu, select ">", then specify the range of "body temperature" of the link destination, and select "completion" from the menu 31.

The "reference link" in the menu 31 of FIG.
It is used to refer to another keyword or data from the keyword in the document. In the example of the document here, a reference link is established from the keyword “body temperature” in the document to the variable “body temperature”. In the actual procedure, first, the range of the character string "body temperature" in the document is specified. Next, select "reference link" from menu 31 and specify the variable "body temperature" of the link destination,
Select “Complete” in the menu 31.

The "output link" in the menu 31 of FIG.
It is used when it is desired to send the keywords and data in the document to the output device 6 of FIG. In the example of the document here, an output link is made from the keyword "body temperature" in the document to "there is a suspicion of a cold." In the actual procedure "First, the" body temperature "in the document
Specify the range of the character string. Next, "output link of menu 31" is selected, the link destination "suspected of cold" is specified, and "completion" of menu 31 is selected. This output link is executed by the inference execution device 5 of FIG. 1 only when the condition is satisfied when the condition of the link source keyword is set.

The "designate output" of the menu 31 of FIG. 4 is used when it is desired to unconditionally send the keyword or data in the document to the output device 6 of FIG.

The link information generated using the link creating device shown in FIGS. 3 and 4 is displayed in the format of FIG. 2 by the user interface. Inference execution device 5 of FIG.
Interprets and executes a document in the hypertext format in which a semantic link is placed in the document as shown in FIG. The algorithm of the inference execution performed here is shown in FIG. The algorithm shown in FIG. 6 starts when a value is set for variable 2. In the example shown in FIG. 1, the inference execution device starts inference when a value of 37.5 is input for the variable “body temperature”. The value of the variable is input by the operator from the keyboard. Alternatively, a value may be set for a variable by a program. In the inference execution device, first, the link provided in the document is searched (51). If no link is found here, the process ends (52). When a link is found, it is searched whether or not another link is provided for the character string of the link source (53). If another link is set here, that link is targeted for processing (54), and the processing of 53 is performed again. If no other link is set, the type of the link is referred to, if it is a conditional link, conditional link processing (56), if it is a reference link, reference link processing (57), if it is an output link, output link processing (5
8) is performed to search for a link not yet processed (59). If no link is found here, the process ends (60). When the link is found, the link is processed and the process of 53 is performed again.

FIG. 7 shows the processing condition processing 56 of FIG. In the conditional link processing, first, the character string of the link source of the target conditional link is referenced, and the link is searched for another reference link having the character string as the link source (62). If there is no reference link, the link source character string is temporarily assigned to the variable A (63). If there is a reference link, the reference link processing is performed for that reference link (6
4). Then, the reference destination character string obtained by the reference link processing is temporarily assigned to the variable A (65). Next, refer to the character string at the link destination of the target conditional link,
The link is searched for another reference link having the character string as a link source (66). If there is no reference link, the character string of the link destination is temporarily assigned to the variable B (67). If there is a reference link, target that reference link,
Reference link processing is performed (68). Then, the character string of the reference destination obtained by the reference link processing is temporarily changed to the variable B.
(69). Next, the type of the condition link (>, <, =, ≤, selected from the condition link menu of FIG. 4)
Either ≧ or ≠) is referred to and the variable X is temporarily substituted (6a). Then, it is judged whether the relationship of AXB is established between the variable A substituted in 63 or 65 and the variable B substituted in 67 or 69 (6b). If this is not the case, the process ends (6c). If the condition is satisfied, the link is searched for an output link having the link destination character string of the target conditional link as a link source (6d). If the output link does not exist, the process is terminated (6f). If it exists, output link processing is performed (6e), and the processing ends (6f).

FIG. 8 shows the reference processing 57 of FIG. In the reference link processing, first, it is searched whether there is another reference link whose link source is the link destination of the target reference link (72). If there is no reference link, the reference destination of the character string of the link source is set to the character string of the link destination (73), and the process ends (77). If there is a reference link, the reference link is processed (74), the reference link process is performed (75), and the reference source character string of the reference link is first processed. (76), and the process ends (77).

FIG. 9 shows the output process 58 of FIG. In the output link processing, first, it is searched whether or not there is another reference link whose link source is the link destination of the target output link (82). If there is no reference link, the character string of the link destination is sent to the output device 6 of FIG. 1 (83), and the process ends (86). If there is a reference link, the reference link processing is performed for that reference link (8).
4). Then, the character string of the reference destination of the reference link is shown in FIG.
To the output device 6 (85), and the process is terminated (8
6).

Next, an example in which the above-described algorithm shown in FIG. 6 is specifically applied will be described using the example of the document shown in FIG. When the variable is set to 37.5, the document link is searched. In this case, since the condition link 11, the reference link 12, and the output link 13 exist, the link search is successful. For example, when the retrieved link is the reference link 12, the link source is the "body temperature" of the document.
Then, it is checked whether the "body temperature" in the document is linked to another link. In this example, since it is the link destination of the condition link 11, YES is established and other links are searched. Similarly, in the case of the output link 13, since "body temperature" of the document is the link destination of the condition link 11, YES is established, and another link is searched. Next, when the condition link 11 is searched, the link source “37” of the condition link is not a link destination of another link, and thus N
O is established. Since the type of this link is a conditional link, conditional link processing is performed.

When the conditional link processing is executed as shown in FIG. 7, it is checked whether the link source "37" of the conditional link 11 is the link source of another reference link. In this case, since it is not the link source of another reference link, NO
Is established. Then, "37" is substituted for the variable A. Next, it is checked whether the link destination of the condition link 11 is the link source of another reference link. In this case, YES is established because the “body temperature” of the document is the link source of the reference link set in the variable “body temperature”. And
Reference link processing is performed.

In the reference link treatment, as shown in FIG. 8, it is checked whether or not there is another reference link ahead of the variable "body temperature". In this case, NO is established because there is no reference link. Next, the character string "body temperature" of the link source of the reference link 12 is set to the character string "37.5" which is the reference destination of the reference link 12. When the reference destination is a variable, the value itself entered in the variable is used as the reference destination. When the reference link processing is completed in this way, the processing shown in FIG. 7 is restarted.

Since the condition link 11 is being processed in FIG. 7, the reference destination of the link destination of the condition link 11 is further substituted into the variable B. That is, for the variable B, “37.
5 ”is substituted. Next, since the type of the conditional link 11 is <, <is assigned to the variable X. Next, it is determined whether AXB is established, but in this example, it is determined whether 37 <37.5 is established. In this example, since the condition is satisfied and the output link 13 is set up at the link destination of the condition link 11, that is, "body temperature", the output link process is performed.

In the output link process, as shown in FIG. 9, it is inspected whether the reference destination of the output link 13 is further extended. In this example, since the reference link does not exist, the link destination character string "there is a suspicion of a cold." Is sent to the output device and is output. As described above, by using the algorithm shown in FIGS. 6 to 9, the inference execution device can execute the inference with the input of the value of the variable as an opportunity and output the result.

Example 2. FIG. 10 is a diagram showing this embodiment. 3, 4, 5 and 6 are the same as those in FIG. The document 91 is linked with the meaning shown in FIG.
Here, the document of 91 has a document structure of 1 chapter interview, 1 chapter 1 body temperature, 1 chapter 2 pulse. When the user selects the topic “1.1 body temperature” in this document 91, the document described in “1.1 body temperature” and the link provided on the document are described in the inference execution apparatus 5 in FIG. 1. Execute as a group of knowledge. In the case of the example in FIG. 11, there are two, "if the patient's body temperature is 37 degrees or more, there is a suspicion of a cold." And "if the patient's body temperature is 40 degrees or more, an antipyretic drug is administered." The sentence is regarded as one piece of knowledge, and the character strings “suspected of cold” and “administer antipyretic drug” are sent to the output device 6 in FIG. 1. Similarly, when the user selects the topic "1. Questioning" in the document 91, the document described in "1. Questioning" and the link provided on the document are described in the inference execution device 5 of FIG. Execute as a group of knowledge. In the case of the example of FIG. 11, 2 of “1.1 body temperature” and “1.2 pulse”
One section is regarded as one piece of knowledge, and the character strings "I have a suspicion of a cold", "I administer an antipyretic drug", and "I measure blood pressure" are sent to the output device 6 in FIG.

Example 3. FIG. 12 is a diagram showing this embodiment. FIG. 12 is a block diagram of the link creating device 4 in this embodiment. In this embodiment, selection condition link means 407 is newly provided. The selection condition linking means links the items forming the table.

The table of FIG. 13 shows that the symptom and degree are referred from the variable 114, the line having the same character string or numerical value as the variable is selected, and the coping method for the line is output. By linking the selection condition link 111 to the item that is the condition of the table, the inference execution device 5 of FIG. 1 refers to the character string of the link destination and the data of the item of the table that is the link source, and Only the lines with the same are processed. The “table output designation” 113 is designated as a table item. Among the data of the designated items, only the data which is the processing target by the above selection condition link is the output device 6 of FIG.
Sent to. In the example of FIG. 13, only the rows with the fever of the item “symptom” are processed by the selection condition link 111, and only the rows of the item “degree” of 2 are processed by the selection condition link 112. . As a result, the item "symptoms"
Is generated, and only the second line with "degree" of 2 is processed.
Then, by specifying the table output of the item "coping method", the data "coping agent" and "administration of antibiotic A" of the second line are sent to the output device 6 of FIG.

As described above, the selection condition link links the items in the table, but the character strings "symptoms" and "degrees" indicating the items are not targets of processing, but, for example, " The values "fever" and "headache" belonging to the item "symptom" are the actual processing targets. That is, as shown in FIG. 13, the selection condition link 112 is set up for the variable “symptom” and the item “symptom” in the table, but when the inference execution device actually executes the inference, Compare the "fever" and "headache" that are actually registered under the item "symptom" in the table with the value "fever" entered in the variable, instead of the item "symptom" being processed This limits the processing target. Therefore, the selection condition link has a character different from that of the condition link described above.

Example 4. FIG. 14 is a diagram showing this embodiment. 14 is different from the embodiment described above in that
The point is that the Y link means / N link means 408 is provided. The Y link means and the N link means link to the flowchart. When a judgment is made in the flowchart and the process branches depending on the judgment result, the Y link and the N link are set. The link creating device creates the Y link and the N link when the condition is judged in the flowchart and the process branches.

The flowchart of FIG. 15 shows boxes 121, 1
The conditions shown in 24 and 125 are judged. First, box 121
If the condition is satisfied, the box (124) indicated by the Y link (122) is processed, and if the condition is not satisfied, N is processed.
The box (125) indicated by the link (123) is processed. This is repeated, and finally, one of the end boxes 126, 127, 128, and 129 having no link to the other is selected, and the processing shown therein is performed. The variable 12a is
The character string "38 degrees" for "body temperature", "yes" for "depression",
Indicates that "none" has been assigned to "eczema". 12
b is a conditional link (>), 12c and 12d are conditional links (=), 12k, 12e and 12f are reference links, 12
g, 12h, 12i, and 12j are output designations.

The user creates a semantic link on the flow chart by the link creating device 4, and the inference executing device 5 performs the following process by the semantic link. First,
Search for the condition link 12b of the box 121 and execute the condition link 12
The link destination "37 degrees" of the reference link 12k from the link source of b is compared with the value of "body temperature" of the variable 12a, 38 degrees, and it is recognized that this condition is satisfied, and the link 1 of Y
22, the condition of the next box 124 is determined. As with the box 121, refer to the condition link 12c of the box 124,
The value “present” of the variable 12a “eczema” is compared with the reference destination “present” of the reference link 12e generated from the link source “eczema” of the condition link 12c, and it is recognized that this condition is satisfied, and from the box 124 By referring to the Y link that appears, the next box 126 is processed. The box 126 is not provided with a conditional link, and only the output of the character string 12 g “measles” is designated. Therefore, the inference execution device 5 outputs the character string “measles” to the output device 6. The character string “measles” is output from the sending and output device 6.

Example 5. FIG. 16 is a diagram showing this embodiment. In this embodiment, database link means for linking in a database having data to be referred to from a document is provided. The database link means 409 creates a thesaurus as described later from the database and establishes a reference link from the document to the thesaurus.

FIG. 17 is a diagram showing an example of the database and thesaurus. Reference numeral 134 is a database having data to be referred to from the document 131. In this example,
The database 134 has a table “examination”, and the table “examination” has attributes “patient name”, “symptom”, “examination day”, “administered drug”, and “examination result”. The table list 132, the attribute list 133, and the attribute link 137 are thesauri that are automatically generated by extracting the table name and attribute name of the database. Hereinafter, this is referred to as database information. By linking the meaning between the automatically generated database information and the character string in the document, the document and the database are combined. The table list 132 has a table name in the database as a character string, and the attribute list 133 holds the attribute name in the table in the table list 132 as a character string in a table format for each table. Attribute link 137 is table list 1
From each table of 32, the attribute list 13 which the table has
It is a link to 3. Here, the user uses the link creation device 4 of FIG. 1 to designate the output of “administered drug” of the document “In the second and subsequent examinations, the previous administered drug is confirmed.”
A reference link 135 is drawn from the character string “administered drug” in the document to “administered drug” in the attribute list 133. This means to obtain the data of the attribute "administered drug" of the table "diagnosis" in the database.

With the link of this meaning, the inference execution unit 5 of FIG. 1 performs the following processing. First, the meaning link provided in the document 131 is checked. The reference link 135 provided under “Administered drug” in the document 131 refers to “administered drug” in the attribute catalog list 133. Further, since the output designation 139 is made for “administered drug” in the document 131, the search expression “π-administered drug (examination)” of the data “administered drug” in the database is generated. π indicates the projection of the database search. The database 134 is searched by this database search formula, and the resulting data is sent to the output device 6 in FIG. In this example, all data of the attribute “administered drug” in the database 134 is sent. The speculative execution unit of this embodiment refers to reference link 135 “administered drug”.
In the above-described embodiment, the character string “administered drug” itself is output. However, in this example, since the reference link 135 refers to the database information, the inference execution device uses the character string “administered drug”. Instead of outputting "drug", the data actually stored under the item "administered drug" of the relational database is accessed and output to the output device. In the example shown in FIG. 17, antibiotic A and antibiotic B are sent to the output device and displayed.

In this way, the link creating device 4 creates a thesaurus and establishes a reference link between the document and the thesaurus so that when the thesaurus is referenced, the inference executing device causes the corresponding data in the database to correspond. To access the data in the database as a result.

Example 6. FIG. 18 is a diagram showing this embodiment. In FIG. 18, a search condition linking unit 410 for linking the search condition for searching the database is provided. FIG. 19 is a diagram for explaining the operation of this search condition linking means. 131, 132, 133, 13
4, 135, 136 and 137 are the same as those in FIG. The variable 141 indicates that the character string “Taro Yamada” is assigned to the variable name “patient name”. The user is shown in FIG.
Similar to 7, using the link creation device 4, the output designation is “administered drug” of the document “In the second and subsequent examinations, the previous administered drug is confirmed.” And the character string “administered drug” in the document is used. A reference link 135 is attached to “administered drug” in the attribute list 133.
This means to obtain the data of the attribute "administered drug" of the table "diagnosis" in the database. Further, the user establishes a condition link (=) 142 from the variable name “patient name” of the variable 141 to the “patient name” of the attribute list 133 in order to obtain only the data of the patient currently targeted. Thereby, the condition that the attribute “patient name” of the table “examination” is “Taro Yamada” is set.

With the link of this meaning, the inference execution unit 5 of FIG. 1 performs the following processing. First, the meaning link provided in the document 131 is checked. From the reference link 135 provided under “Administered drug” in 131, the “administered drug” in the attribute list 133 is referenced. Further, since the output designation 139 is made for “administered drug” in the document 131, the search expression “π-administered drug (examination)” of the data “administered drug” in the database is generated. π indicates the projection of the database search. Next, it is determined whether or not another link is put in the “administered drug” table of the attribute list 133, and the condition link (=) from the variable name “patient name” of the variable 141 to the attribute “patient name” of the attribute list 133 (= ) Recognizing that 142 is stretched,
The conditional expression “σ patient name = Taro Yamada (medical examination)” is generated.
Then, this conditional expression is added to the previous database search expression to generate the database search expression “π-administered drug (σ patient name = Taro Yamada (examination))”. The database 134 is searched by this database search formula, and the resulting data is sent to the output device 6 in FIG. In this example, database 1
34 "antibiotics A" are sent.

Example 7. FIG. 20 is a diagram showing this embodiment. In FIG. 20, a program execution link unit 411 for linking a document to execute a program related to the document is provided. FIG. 21 is a diagram for explaining the operation of the program execution link means 410. Reference numeral 151 is a document, 152 is a program execution link for executing a program from the document, and 153 is program information in which the names of preregistered programs are displayed in a table format. Document "In the first consultation,
If there is a program that checks the medical examination item described in "Checking the medical examination item", Document 15
By establishing a program execution link 152 between “checkup item check” in 1 and program information 153, when the user selects “checkup item check” in the document,
The inference execution device 5 of FIG.
, The program name is obtained from the program information 153, and the program is executed. Program information 1
53 is automatically generated from the program name when the program is registered by the user. The user inputs the program name, the program execution module, and the program start execution script when registering the program. Figure 1
The inference execution device 5 internally holds a correspondence table of a program name and its start execution script, and executes the program by referring to this correspondence table from the program name.

Example 8. FIG. 22 is a diagram showing this embodiment. In this embodiment, the database linking means 409, the search condition linking means 410 and the program execution linking means 411 are organically combined.
151 and 152 are the same as those in the above-described embodiment. 1
32, 133, 134 and 137 are the same as those in the above-mentioned embodiment. 141 and 142 are the same as those in the above-described embodiment. Reference numeral 161 is program information registered by the user, and the program name is “medical examination item check (IN
(Symptom)) ”and“ IN (symptom) ”represents an input parameter of this program. By making a reference link 162 from this input parameter to the “symptom” of the attribute list 133, the data of the attribute “symptom” of the database table “examination” is input to the examination item check program.

Example 9. In the expert system in this embodiment, when the keyword in the document, the word in the thesaurus of the relational database, the application program name is the same word or a synonym, the character string is automatically linked, and the user Makes it easier to find words with the same meaning when linking. FIG. 23 is a diagram showing this embodiment. When linking in the above example,
It was done by the operator using the link creation device to specify the character string with the keyboard or mouse,
In order to reduce the linking work by the operator, this embodiment is provided with a link generation support means 412 that automatically links words having the same meaning, similar words, or synonyms. FIG. 24 is a diagram for explaining the operation of the link generation support means 412. The link generation support means 412
The synonyms or synonyms are automatically linked from the documents, variables and database information shown in FIG. In this case, the reference links are automatically generated because they are synonyms or synonyms. For example, since the “examination” of the document and the “examination” of the table list 132 of the database information have the same word, the reference link 173 is automatically set from the document to the table list. Similarly, the reference link 171 and the reference link 172 are automatically set. In this example, a reference link is provided only for the same word, but a reference link may be automatically provided for a synonym or a similar word. By automatically setting the reference link in this way, the user can reduce the trouble of setting the link.

Further, as described above, the link is not automatically established, and the synonym or synonym is simply displayed specially so that the user can determine whether or not the link should be established. I don't mind.

Example 10. It is conceivable that the functions of the above-described first to ninth embodiments may be selectively combined or may be arbitrarily used. For example, if there is a document with a table or flowchart embedded outside the document, you can create a link to the document and at the same time you can link to the table or flowchart. It is possible to construct a knowledge base using hypertext.

[0065]

As described above, according to the first aspect of the invention, the user can construct an expert system using a document as a knowledge base by linking the document with a meaning link.

Further, according to the second invention, one unit such as a chapter or a section of the document structure can be regarded as one knowledge, and the inference can be executed for a certain set of knowledge.

According to the third aspect of the invention, the user links the meanings to the items in the table in the document so that the data satisfying the condition is processed from the data in the table, and the table in the document is processed. You can carry out the execution indicated by your knowledge.

According to the fourth aspect of the invention, with respect to the flow (flow chart) in the document, the user links each item with data reference or program execution, so that execution indicated by the knowledge of the flow in the document is executed. Can be done.

According to the fifth aspect of the invention, the user links the meanings to the items in the table, and the data satisfying the condition is processed from the data in the table, and the execution indicated by the table knowledge is performed. be able to.

According to the sixth aspect of the invention, the user can perform the execution indicated by the knowledge of the flow by linking the data reference or program execution to each item in the flow (flow chart).

According to the seventh invention, the document and the database can be combined, and the database can be searched from the document.

Further, according to the eighth aspect, it is possible to set a database search condition on a document with respect to a document combined with a database, and it is possible to perform a database search with the condition set from the document.

According to the ninth invention, the document and the program can be combined, and the program can be executed from the document.

According to the tenth aspect of the invention, by combining the database combined with the document and the program combined with the document, the execution of the program referring to the data of the database and the execution result of the program are stored in the database. Execution of the program can be performed from the document.

According to the eleventh aspect, when the user links to the document, it is possible to easily obtain the information of the database or the information of the program having the same or the same meaning as the word in the document.

[Brief description of drawings]

FIG. 1 is a block diagram including a link creation device and an inference execution device according to an embodiment of the present invention.

FIG. 2 is a diagram showing meaning links provided on a document according to an embodiment of the present invention.

FIG. 3 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 4 is a diagram showing a user interface of a link creating device according to an embodiment of the present invention.

FIG. 5 is a diagram showing a condition menu at the time of selecting a condition link of the link creating device in the embodiment of the present invention.

FIG. 6 is a diagram showing a process flow of an inference execution device according to an embodiment of the present invention.

FIG. 7 is a diagram showing a processing flow for a conditional link of an inference execution device according to an embodiment of the present invention.

FIG. 8 is a diagram showing a processing flow for a reference link of an inference execution device according to an embodiment of the present invention.

FIG. 9 is a diagram showing a processing flow for an output link of an inference execution device according to an embodiment of the present invention.

FIG. 10 is a block diagram including a link creating device and an inference executing device according to an embodiment of the present invention.

FIG. 11 is a diagram showing links of meanings provided on a document in an embodiment of the present invention.

FIG. 12 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 13 is a diagram showing meaning links provided in a table in a document according to an embodiment of the present invention.

FIG. 14 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 15 is a diagram showing links of meanings provided in a flow (flow chart) in a document according to an embodiment of the present invention.

FIG. 16 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 17 is a diagram showing meaning links provided between a document and database information according to an embodiment of the present invention.

FIG. 18 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 19 is a diagram showing links of meanings provided between documents and database information and links of database search conditions provided on documents according to an embodiment of the present invention.

FIG. 20 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 21 is a diagram showing a semantic link provided between a document and program information according to an embodiment of the present invention.

FIG. 22 is a diagram showing meaning links provided between a document, database information, and program information according to an embodiment of the present invention.

FIG. 23 is a block diagram of a link creating device according to an embodiment of the present invention.

FIG. 24 is a diagram showing reference links automatically placed in the same word between a document, database information, and program information in one embodiment of the present invention.

FIG. 25 is a diagram showing an operation of a conventional expert system.

[Explanation of symbols]

 1 Document 2 Variable 3 Link Creation Device 4 Inference Execution Device 5 Output Device 11 Conditional Link (>) 12 Reference Link 13 Output Link 31 Link Type Selection Menu 41 Conditional Link Menu 91 Document 92 Variable 101 Conditional Link (>) 102 Reference Link 103 Output link 105 Conditional link (>) 106 Reference link 107 Output link 111 Selection condition link 112 Selection condition link 113 Table output designation 114 Variable 12a variable 131 Document 132 Database information (table name) 133 Database information (attribute name) 134 Database (table) "Examination") 141 Variables 142 Condition links (=) 151 Documents 152 Program execution links 153 Program information 161 Program information 162 Reference links

Claims (11)

[Claims] 1. An expert system having the following elements: (a) A link creating device for linking a meaningful link to a document and using the linked document as a knowledge base;
(B) An inference execution device that executes inference based on the knowledge base. 2. The document has a predetermined document structure, the link creating device creates a knowledge base with one unit of the document structure as one knowledge, and the inference executing device creates one of the document structures. The expert system according to claim 1, wherein the inference is executed by designating one unit. 3. The link creating device is further characterized in that a link having meaning is attached to an item forming a table, and the linked table is used as a knowledge base.
Or the expert system described in 2. 4. The link creating device further creates a link having meaning to items constituting the flowchart, and uses the flowchart with the link as a knowledge base. The described expert system. 5. An expert system having the following elements: (a) A link creating device in which a link having meaning is attached to an item constituting a table, and the linked table is used as a knowledge base, and (b) the knowledge base. An inference execution device that executes inference by. 6. An expert system having the following elements: (a) a link creating device in which a link having meaning is attached to an item constituting a flowchart, and the linked flowchart is used as a knowledge base; and (b) the knowledge base. An inference execution device that executes inference by. 7. An expert system having the following elements: (a) a relational database having relation names and attribute names,
(B) A link creating device for extracting a relation name and an attribute name of the relational database to generate a thesaurus, linking the document to the thesaurus, and forming a knowledge base, (c)
An inference execution device that executes inference by the knowledge base and accesses a database from a document. 8. The expert system according to claim 7, wherein the inference execution device sets a search condition when searching a database from a document. 9. An expert system having the following elements: (a) an application program for executing a predetermined process, (b) a link creating device for establishing a meaningful link from a document to the application program, (c) the link An inference execution device that executes an application program from the above document using the inference execution device. 10. An expert system having the following elements: (a) a relational database having relation names and attribute names,
(B) An application program that executes a predetermined process, (c) A link that has meaning to the application program from a document is created, and a relational name and attribute name of the relational database are extracted to generate a thesaurus. A link creating device for linking to a thesaurus, (d) an inference executing means for executing an application program from the document using the link and accessing a database from the application program using the link. 11. The link creation device comprises link generation support means for supporting linking to a synonym or a synonym when a link is established. The expert system according to 9 or 10.