JPH02120976A - System for forming confirmation sentence of japanese sentence analyzing result - Google Patents

Abstract

PURPOSE:To simply confirm the contents of retrieval execution simultaneously with the analyzed result of an input Japanese sentence by providing the title system with a means for forming and displaying a confirmation sentence expressing the meaning contents of the input Japanese sentence understood by the system. CONSTITUTION:When a Japanese sentence expressing an operation command for information retrieval is inputted to a Japanese sentence analyzing part 1, the analyzing part 1 refers a word dictionary 2, recognizes a Japanese character string, forms intermediate expression format 1a and supplies the formed expression format 1a to an input sentence meaning contents confirming sentence forming processing part 3 and the processing part 3 forms a confirmation sentence 3a indicating the meaning contents of the Japanese sentence while referring tables 4, 5. The processing part 3 supplies the confirmation sentence 3a to a user confirmation part 6, and in the case of updating the meaning contents, updates the confirmation sentence 3a and then confirmation sentence 3a is used as a user's input sentence to execute the analysis of the input sentence again from the Japanese analyzing part 1. Even if the system analyzes a Japanese sentence in error, an instruction such as retrieval can be prevented from being executed in error.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a Japanese sentence analysis method in an information processing system that includes a Japanese language analysis section and an information search processing section and performs processing such as information search in Japanese. In particular, the present invention relates to a Japanese sentence analysis result confirmation sentence generation method suitable for clarifying the meaning of information processing contents.

[Conventional technology]

Conventionally, as a processing system for this type of information retrieval, etc., for example, there has been a
7), pp. 1325-1326 [As described in RISJ]
There are known devices that analyze Japanese sentences and perform searches based on the analysis results. However, in this document, the analysis results of Japanese sentences are not presented to the user, and if the Japanese sentences are analyzed incorrectly, an incorrect search will be performed and an incorrect search result will be obtained.

[Problem to be solved by the invention]

In the above-mentioned conventional technology, consideration is given to showing the user how the input Japanese text is analyzed, how the system understands it, and how the system is searching, and prompting the user to confirm it. Therefore, if the system incorrectly analyzes Japanese text, it may search for incorrect information and obtain incorrect search results.

Therefore, it is an object of the present invention to solve the problems of the prior art described above, and to analyze and present the analysis results of Japanese sentences expressing operation instructions for processing input information search etc. in an easy-to-understand manner to the user. Based on the results, a confirmation sentence that expresses the clear meaning of the input Japanese sentence is generated and shown to the user, prompting him to confirm it. This avoids erroneous information retrieval processing, performs accurate information processing, and helps the user understand what he or she wants. The object of the present invention is to provide a Japanese sentence analysis result confirmation sentence generation method that obtains processing results as follows.

[Means to solve the problem]

In order to achieve the above object, the present invention analyzes an input Japanese sentence expressing an operation instruction such as information retrieval, converts it into an operation instruction command language, and executes an instruction such as information retrieval using the converted operation instruction command language. A Japanese data processing system is characterized by the semantic content of the input Japanese text understood by the processing system (how the processing system understands the input Japanese text and executes commands such as searches). 1i! generates a confirmation sentence representing the meaning of the input Japanese sentence (meaning of the input Japanese sentence) and displays it! Equipped with certified statement generation and display means.

Specifically, the confirmation sentence generation means is configured as follows. That is, from the input Japanese sentence, the data header Att
The word equivalent to ribute (attribute item name), this A
By extracting words corresponding to Value (attribute item data) corresponding to ttribute, words expressing logical instructions, and words expressing logical operations, and combining these extracted words according to the analysis results of the input Japanese sentence. , is configured to generate a TJ'fI main sentence that shows the clear meaning and content of the Japanese sentence expressed by the analysis result.

Also, when generating a confirmation sentence, A t of the database
Words equivalent to “tribute” and “tribute”
A logical directive word relationship table in which logical directives representing logical directive relationships between words corresponding to Value corresponding to e, logical directive character strings corresponding to the logical directives are defined in advance, and logic representing logical operations of search conditions. A logical operator relationship table in which operators and logical operation character strings corresponding to the logical operators are defined in advance is configured to be referenced.

[Effect]

The effect based on the above configuration will be explained.

The confirmation sentence generated and displayed by the confirmation sentence generation and display means indicates the meaning of the input Japanese sentence, which indicates how the processing system is trying to understand the Japanese sentence input by the user and execute instructions such as information retrieval. It represents the content. The user looks at the displayed confirmation sentence and checks whether the confirmation sentence and the input Japanese sentence have the same meaning. If they are the same, the user can execute the command; if they are different, the user can modify, delete, add, etc. Let's do it.

This prevents a command such as a search from being executed erroneously even if the system incorrectly analyzes Japanese text.

In a specific invention, when generating a 611 certified sentence, the words representing the attribute item name, the corresponding attribute item data, the logical instruction, and the logical operation are extracted from the input Japanese sentence, and the input Japanese sentence is analyzed. According to the results, these extracted words are combined to generate a confirmation sentence.

In addition, when generating a confirmation sentence, a logical instruction word relationship table that defines a word expressing the logical instruction and/or a logical symbol corresponding to this word, a word expressing the logical operation, and/or a logical symbol corresponding to this word, A logical operator relationship table that defines the logical operators to be used is referenced. Here, the following contents are registered in these tables,
Been there again. That is, a conjunction representing a logical product or a logical sum is registered as a logical operation string, and a symbol corresponding to the logical operation string is registered as a logical operator in the logical operator relationship table, and the attributes of the database are registered.
A word representing a logical directive relationship between a word corresponding to e and a word corresponding to Value corresponding to Attribute is used as a logical directive string, and a symbol corresponding to the logical directive string is used as a logical directive. Register in the relationship table.

For simplicity, a term corresponding to Attribute in the database is defined as term A, and a term corresponding to Value corresponding to Attribute is defined as term D.

Through the Japanese language analysis process, the A term and the D term are related by a logical indicator representing a logical indicative relationship, and the A term, the D term, and the logical indicator are collectively defined as a condition.

Add a particle representing the nominative to the character string that corresponds to term A, and immediately after that, refer to the logical demonstrative relationship table and express the logical demonstrative relationship between terms A and D in the character string that corresponds to term D. Connect the logical directive with the corresponding logical directive character string attached. The above method generates a condition confirmation statement.

A plurality of conditions can exist in the input sentence, and each condition is related by a character string representing a logical operation. Refers to the logical operator relationship table using the character string representing the logical operation, extracts the logical operator character string corresponding to the character string representing the logical operation, and attaches it immediately after the condition confirmation statement.
By performing this process for all conditions, it is possible to generate a confirmation sentence that clearly expresses the meaning of the analysis result of the input sentence, allowing the user to easily confirm the content of the search execution. .

〔Example〕

Hereinafter, one embodiment of the present invention will be described using FIGS. 1 to 23, and an expanded embodiment of the present invention will be described using FIG. 24.

FIG. 1 shows the functional blocks constituting the present invention. 1 is a Japanese language analysis unit, 2 is a word dictionary, 3 is an input sentence semantic content rIi recognition sentence generation processing unit, and 4 is a logical indicator relationship table. , 5 is a logical operator relationship table, 6 is a user confirmation section, and 7 is an operation instruction command language conversion processing section.

When the user inputs Japanese text JP expressing operation instructions for information retrieval into the Japanese language analysis section 1, the analysis section 1
Referring to the word dictionary 2, the Japanese character string is recognized, an intermediate representation 1a is generated, and this is held. Preserved intermediate representation 1
A is supplied to the input sentence semantic content confirmation sentence generation processing unit 3, which generates a confirmation sentence 3a showing the meaning and content of the Japanese sentence while referring to Table 4.5, and holds this. If you want to supply the retained confirmation sentence 3a to the user confirmation unit 6 and update its meaning, after updating the confirmation sentence 3a, use the Mf, T1 confirmation sentence 3a as a user input sentence and convert it into Japanese. Execute again from analysis section l. If the semantic content is not updated, the intermediate expression format 1a is retained by the operation instruction command language conversion processing unit 7, and mechanically converted from the intermediate expression format 1a to the operation instruction command language 7a from formal language to formal language. From FIG. 2 onwards, details of the functional blocks in FIG. 1 will be explained.

FIG. 2 shows the hardware configuration of an embodiment of the present invention, in which 8 is an input/output device, 9 is a processor, and 11 is a storage device. The device 11 consists of the following storage section. 12 is a working area, 10 is a Japanese language analysis processing storage area, 20 is a word dictionary storage area, 30 is an input sentence meaning content confirmation sentence generation processing unit storage area, 40 is a logical indicator relationship table storage area, and 50 is a logical operator Reference numeral 60 represents a relationship table storage area, 60 represents a user confirmation unit storage area, and 70 represents an operation instruction command language conversion processing unit storage area.

Figure 3 is an overview PAD diagram of the Japanese language analysis department (Probl).
em Analysis Diagram) and (d
(ef stands for definition), which shows the configuration of Japanese language analysis processing, and consists of the following processing units. First, term recognition processing 101, then modification expression recognition processing 10
2. Next, it consists of intermediate representation generation processing 103.

FIG. 4 is a PAD diagram of the term recognition process 101, in which the phrase separation Japanese character strings input by the user are recognized, and the phrase tables shown in FIGS. 12 and 17 (provided in the working area 12) are ), and this process will be explained below with reference to the PAD diagram.

The system waits for input of a Japanese character string for segmentation from the user (10101), and assigns the Japanese character string input by the user to the character string variable INF (10102). Next, while the variable INF contains a clause, the following processing is performed (10103).

Refer to Word Dictionary 2 to determine whether a word stem is included in the phrase (10104). If it is, register the part of the phrase that corresponds to the stem as the word stem in the phrase table, and use the rest of the character string. The part is registered as an attached word in the clause table (10105).

Next, the regular words and keyword attributes of the word dictionary 2 are each registered in the phrase table (10106).

If 10104 is negative, it is determined whether or not a number string exists in the clause (10107), and if it exists, the number string is registered as the stem of the clause table (1010B), and the remaining character strings are added to the clause table. Register as a word (10
109). After 10104, the clause is deleted from the variable INP (10110). 10101 to 10110, the clause-separated Japanese sentence shown in Figure 11 is recognized,
It is stored in the phrase table shown in FIG. Similarly,
10101 to 10110, the phrase separated Japanese sentence shown in FIG. 16 is recognized and stored in the phrase table shown in FIG. 17. The clause table at this time stores clause information, and using the information in this clause table, FIGS. 5 and subsequent figures show the process of identifying the modifying relationship of the input sentence and generating an intermediate expression.

FIG. 5 is a PAD diagram of the modified expression recognition process 102,
It shows the processing up to the point of identifying the modifying relationship of the input sentence using the information in the bunsetsu table, and this processing will be explained below in accordance with the PAD diagram.

While data exists in the phrase table, the following processing is performed (10201). Refer to the clause table and check if the clause is a logical indicator clause and the regular word of the clause is the second clause.
To determine whether the phrase is registered in the logical directive relationship table shown in Figure 1 (10202), if so, refer to the phrase table and determine whether the phrase immediately before the phrase is attribute item data (10203), In the case of , the logical indicator corresponding to the logical indicator character string in the logical indicator relationship table 4 is extracted, concatenated immediately before the clause immediately before the clause, and the clause and the clause immediately before are combined into a data item (D term). (10204). After the determination in step 10202, refer to the clause table to determine whether the clause is a logical product conjunction or a disjunction (10205).If yes, the clause immediately before and after the conjunction clause is It is difficult to determine whether the attribute item data are the same or the attribute item names are the same.
206), in the case of response, it is difficult to determine whether the clauses immediately before and after the conjunction clause have the same attribute item name or not 10207)
, in the case of response, the relevant conjunction clause and the clauses immediately before and after are collectively determined as an Attribute term (A term) (1
0208). If 10207 is negative, the conjunction clause and the clauses immediately before and after are collectively determined as a data item (D term) (10209).

After 10201, while there is data in the clause table (
10210), performs the following processing. First, the phrase in question is
It is difficult to determine whether the clause has not yet been determined as an A term or a D term (10211), and if so, refer to the clause table and determine whether the clause is an attribute item name or not (10212),
If so, the clause in question shall be determined as Section A. 1021
If 2 is negative, the clause is determined as the D term. (1
0214). 10201 to 10214, each clause of the input Japanese sentence is determined as an A term or a D term. Here, the A term is A ttr of the database.
The term D is a word corresponding to the value of A jjribute in the database. The qualifying expression processing shown in 10201 to 10214 is a qualifying process for qualifying relationships in Japanese sentences seen from the database side, and is an example of natural language analysis. The process of generating an intermediate expression representing the logical relationship between the A term and the D term determined in FIG. 6 using the determined A term and D term is shown.

FIG. 6 is a PAD diagram of the intermediate representation generation process 103,
It shows the processing up to the generation of the intermediate representation, and this processing will be explained below with reference to the PAD diagram. The A term at the end of the A term and the D term determined in the qualifying expression recognition process 102 is determined as a search target and stored (10301). Next, perform the following processing for all A terms and D terms (10
302).

Adjacent terms A and D are determined as the minimum unit of the search condition (10303). After 10302, the following process is performed for all the minimum units (10304). A logic instruction is placed at the beginning of term D. 103
05), in the case of response, connect the D term immediately after the A term and select the item in the search condition (determined as cow (10306).103
If 05 is negative, the logical operator '=-'' is connected immediately after the A term, and the D term is connected immediately after that to be determined as a condition in the search condition. (10307). After 10304, all The following processing is performed for the conditions (10308).

Refer to the phrase table and between the determined conditions,
Distortion power/judgment 103
09), in the case of response, the interval between the conditions is the remainder operator''
or" (10310). If 10309 is negative, insert a logical operator "and" between the conditions (10311). After 10308, for all conditions in the search condition, The following processing is performed (10312).It is determined whether the logical operator "Or" exists in the search condition (10313), and if so, the A term of the condition immediately before and after the logical operator "Or" I want to determine whether it is the same -A term.103
14) In the case of response, insert curly braces “ (” before the condition immediately before the logical operator “or” in the search condition, and insert curly braces “ ) immediately after the condition immediately after the logical operator “Or” ° Insert (10315).

Next, it is necessary to determine whether the logical directives of the conditions immediately before and after the logical operator "or" are the same and are other than the logical directive "=-10316), and if yes, the logical The logical operators in the conditions immediately before and after the operator are enclosed in curly braces “
(Concatenate immediately before ``, and add logical directives in curly braces to ``=
Replace with =" (10317). Determine whether the logical indicator is =", which indicates negation (10318),
If so, change “or” to “and” (1031
9). Modified expression recognition processing of 10201 to 10214,
By the intermediate expression generation processing of 10301-10319, the intermediate expression shown in FIG. 13 is generated from the phrase data shown in FIG. 12. Similarly, the intermediate expression shown in FIG. 18 is generated from the phrase data shown in FIG. 17. here,
An intermediate expression consists of a search condition that expresses the logical relationship between terms A and D for information retrieval, and a search target. Using this intermediate expression, S
Conversion to QL (database search language for relational databases in international standards) is also possible.

FIG. 7 is an example of word dictionary information to be referred to in Japanese language analysis processing 1, and includes information on word stems, regular words, and keyword attributes.

FIG. 8 is an overview PA of input sentence meaning and content confirmation sentence generation processing 3.
Fig. D shows the structure of input sentence meaning/content confirmation sentence generation processing, which is composed of the following processing units.

First, it consists of an intermediate expression expansion process 301, and then a confirmation sentence generation process 302.

FIG. 9 is a PAD diagram of intermediate representation expansion processing 301,
In order to generate a confirmation sentence, the intermediate expression is tabulated, and this process will be explained below using a PAD diagram (see also FIGS. 14 and 19, which will be described later). The following processing is performed for the intermediate expression search conditions (30101). Refer to the clause table, extract the regular word of the clause corresponding to section A from the clause table,
Register as the attribute item name of the intermediate expression expansion table (3
0102). Next, refer to the logical directive relationship table,
The logical directive corresponding to the A term is extracted and registered as a logical directive in the intermediate representation expansion table (30103). Next, refer to the phrase table, extract the regular word of the phrase corresponding to the D term corresponding to the A term from the phrase table, and register it as attribute item data in the intermediate expression expansion table (30
104).

Next, it is necessary to determine whether the condition exists within the curly braces "(")" and is not an it condition within the curly braces "(")" (30105). If so, the integer "1" (the next ) is registered as the conditional modification range of the intermediate expression expansion table (30106). 3010
If the answer is 5, the integer "0" (indicating that the next line is not modified) is registered as the range of condition modification in the intermediate expression expansion table (30107). After 30105, it is difficult to determine whether a logical operator exists immediately after the condition (30108)
, in the case of response, the logical operator is "and" (30109), and in the case of response, the string "&" representing and.
" is registered as a logical operator in the intermediate representation expansion table (30110). If the result in 30109 is negative, the character string "!" representing or is registered. ” is registered as a logical operator in the intermediate expression expansion table (30111). If 30108 is negative, check if there is a brace “ )” immediately after the condition.3
0112), if not, register the character string “E” representing the end as a logical operator in the intermediate expression expansion table (30
113). 30101 to 30113 generate the intermediate expression development table shown in FIG. 14 from the intermediate expression shown in FIG. 13. Similarly, the intermediate representation expansion table shown in FIG. 19 is generated from the intermediate representation shown in FIG.

In addition, in FIG. 14 and FIG. 19, as mentioned above, "=="
is affirmative (logically equal), r>=J is greater than or equal to, "&" is A n d % r! J is or, r=J is other than,
rEJ represents End. The attribute item name shown in the intermediate representation is a character string that corresponds to attribute in the database, and the attribute item data is a character string that corresponds to attribute in the database.
The logical indicator 1 is a character string corresponding to the Value corresponding to e, and the logical indicator 1 is a logical indicator character string representing the logical indicator relationship between the attribute item name and the attribute item data, and these three items are collectively called a condition. In addition, the logical operator is a logical operator that represents a logical operation between conditions, and the range of condition modification represents the range of logical operations between conditions.
”, the logical operation between the condition and the next condition is given priority, and in the case of rOJ, the normal logical operation is performed. In addition, logical directive 2 is a logical directive representing a logical directive relationship among the entire set of conditions expressed in the range of condition modification (for example, in FIG. 19, the range rlJ of condition modification is This means that the condition ``='' is added to the set in the second and third lines.)

FIG. 10 is a PAD diagram of the confirmation sentence generation process (302), showing the process of generating a value sentence from an intermediate expression expansion table in which intermediate expressions are expanded.
The explanation will be given according to the diagram.

First, character string conversion s e n L (Sentenc
e) is cleared (30201), and then the integer conversion FLG,n is initialized (30202). Next, the following processing is performed until the n-th logical operator of the intermediate expression expansion table becomes "E" (30203).

First, it is necessary to determine whether the character string corresponding to logical indicator 2 on the nth line of the intermediate expression expansion table is registered as a logical indicator in the logical indicator relation table (30204), and if so, set the integer rloOJ to the variable FLG. Please substitute 3020
5) Assign the string "(" to the variable 5ent.302
06), Add the contents of variable n to variable i (30207
). Set the string corresponding to the attribute item data to the string variable se
nt (30208), and immediately after that, the case particle "ga" is substituted (30209). After 30204,
A character string corresponding to the attribute item data in the n-th row of the intermediate expression expansion table is added to the variable 5ent (30210).

Next, it is determined whether the character string corresponding to the logical indicator l in the n-th row of the intermediate expression expansion table is registered as a logical indicator character string in the logical indicator relation table (30211), and if yes, the logical indicator relation Add the logical instruction string corresponding to the logical indicator of the table to the variable 5ent (302
12).

After 30211, it is necessary to determine whether the range of the conditional modification in the nth row of the intermediate expression expansion table is equal to the integer "0".3021
3) In the case of 30214), it is necessary to determine whether the variable FLG is equal to the integer r100J (see 30204 above).
In this case, add the string “ ) to the variable 5ent.
215), adds the logical instruction character string corresponding to the i-th logical indicator 2 of the intermediate expression expansion table to the variable 5ent (30216). Next, change the string “de” to variable 5.
ent (30217), if no in 30214, determine whether the logical operator on the nth row of the intermediate expression expansion table is equal to the string "E" (30218), and if yes, the string " ” to the variable 5ent 302
19) If 30218 is negative, add the character string "de" to the variable 5ent (30220). 3021
If 3 is negative, determine whether the n-th logical operator of the intermediate expression expansion table is equal to the character string “E” (30221),
If not, determine whether the logical operator in the nth row of the intermediate expression expansion table is equal to the character string “&” (30222),
If yes, the conjunction string “and” is added to the variable 5ent (30223), and if no, the conjunction string “or” is added to the converted 5ent (30224).

After 30213, add the integer "1" to the variable n (30
225). After 30203, add the character string registered as the intermediate expression search target to the variable 5ent (30
226). Next, search for the string "in the variable sen["
Add (30227).

30201 to 30227, the mirror sentence shown in FIG. 15 can be generated from the intermediate expression expansion table shown in FIG. 14. Similarly, the confirmation sentence shown in FIG. 20 can be generated from the intermediate expression expansion table shown in FIG. 19.

That is, by using 30201 to 30227, it is possible to generate a confirmation sentence indicating the meaning of the Japanese sentence representing the input operation instruction for information retrieval.

FIG. 11 shows an example of a Japanese character string separated into clauses representing an operation instruction for information retrieval input by the user, and FIG. 16 is similar.

FIG. 12 is an example of a clause table showing the recognition result of clause-separated Japanese character strings representing operation instructions for information retrieval input by the user, and FIG. 17 is similar.

FIG. 13 is an example of an intermediate expression representing the logical meaning content of the Japanese character string shown in FIG. 12, and FIG. 18 is similar.

FIG. 14 is an example of an intermediate expression expansion table that expands the intermediate expression shown in FIG. 13, and FIG. 19 is the same.

FIG. 15 shows an example of a confirmation sentence generated using the intermediate expression expansion table shown in FIG. 14, and FIG. 20 is similar.

FIG. 21 shows an example of the logical directive word relationship table.

FIG. 22 is an example of a logical operator relationship table.

According to this embodiment, since each functional block is realized by program logic, L
It can be integrated into SI to increase speed, and since the processing information to be used is stored separately, the information content can be replaced depending on the field of application and can be used for general purposes.

The operation instruction content (content of the operation instruction command language 7a) of the question sentence expressing the operation instruction is written in Backus notation (Backus No.
It can be expressed by the intermediate representation shown in FIG. 23 using ur Form ). A word representing a logical indicator is prepared as a logical instruction character string, a word representing a logical operator is prepared as a logical operation character string, and a confirmation sentence is generated according to the intermediate expression format. Therefore, a confirmation sentence equivalent to the operation instruction content in the intermediate representation format is generated.

Embodiments other than those described above that expand the present invention will be described below.

(1) How does the system understand and execute the search in terms of the means for parsing the Japanese text representing the operation instructions for the input information search and converting it into an operation instruction command language (e.g. 5QL)? The following confirmation table is generated as one that can confirm the meaning and content of the input Japanese sentence expressing . That is, immediately before generating the confirmation sentence shown in the above example, the word corresponding to the Δ term is used as the attribute item name, and the word corresponding to the A term, the word expressing logical instructions, and the word expressing logical operation are used as attributes. As item data, a confirmation table is generated by dividing it into search conditions and search targets. According to this, the first
The confirmation table shown in FIG. 24 is generated from the input Japanese sentence shown in FIG.

This makes it a little difficult for novice end users to understand, but for more experienced users it is more intuitive (intuitive) than in the example where they read the confirmation text and understand the meaning. be able to understand meaning and content (in terms of meaning). Therefore, using only one of the confirmation text and the confirmation table has a considerable effect, but if both can be generated and displayed, it will be easier for users to understand the meaning and content, regardless of whether they are beginners or experts. It is advantageous because it can be done.

(2) An input that represents how the system understands and attempts to execute a search, in a means of parsing the input Japanese text representing operation instructions for information retrieval and converting it into operation instruction command language. The meaning of the Japanese sentence is shown to the user in an intermediate representation as shown in FIG. 13, before the confirmation sentence shown in the above embodiment is generated. In this case, a user who understands logical expressions can understand the meaning by looking at the - eye. In this case as well, if intermediate expressions can be displayed together with the confirmation sentence, users of various proficiency levels can easily understand the meaning.

(3) The confirmation sentence generated by the above embodiment is used as an input sentence after the input sentence.

According to this embodiment, when correcting the input sentence, the time spent on input is shortened, and the usability of the system is improved.

(4) The confirmation sentences generated according to the above embodiment are stored as input sentences after the input sentence.

According to this embodiment, after the input sentence, if the sentence to be input and the input sentence have similar meanings, the stored confirmation sentences are retrieved and corrected, thereby reducing the time spent inputting. This has the effect of improving the usability of the system, and when the confirmation sentence is used as an input sentence, since the confirmation sentence is in ideal Japanese, accurate Japanese analysis is performed, and the processing time spent on Japanese language analysis is reduced. This has the effect of reducing time.

(5) Refer to a word dictionary in which word information is defined in advance, and if the keyword attribute represents a logical directive word, use that regular word as a logical directive character string in the logical directive word relationship table,
Register automatically. Further, by referring to the word dictionary, if the keyword attribute is a logical product operator or a logical sum operator, the regular word is automatically registered as a logical operation character string in the logical operator relationship table.

According to this embodiment, when creating a logical directive word relation table or a logical operator relation table, there is no need to register all information, and the table can be easily created in a short time.

(6) After showing the user the confirmation text generated according to the present invention, the user decides whether to execute the search or not. In addition, if the meaning of the input Japanese sentence differs from the meaning shown in the confirmation sentence, the confirmation sentence is corrected, added, or deleted at the user's discretion. According to this embodiment, when the user searches This has the effect that it becomes possible to easily operate whether or not to execute the confirmation statement, and it becomes possible to easily modify, add, or delete the confirmation statement.

(7) In the above embodiment, the case where the present invention is mainly applied to an information retrieval system has been described, but the present invention is not limited to this and can be applied to other Japanese language analysis systems and natural language interfaces.

〔Effect of the invention〕

According to the Japanese analysis result confirmation sentence generation method of the present invention, the confirmation sentence generation and display means that expresses the meaning of the input Japanese sentence understood by the system is provided. Based on the analysis results of the Japanese sentences expressing operation instructions, the user can easily check the analysis results of the input Japanese sentences by using confirmation sentences that clearly express the meaning and content of the input Japanese sentences, and at the same time, it also makes it easier to perform searches. As a result, the user can perform an accurate information search and obtain the search results, which is an excellent effect.

[Brief explanation of drawings]

Figure 1 is a block diagram showing an overview of the configuration of the Japanese language analysis result confirmation sentence generation method of the present invention, Figure 2 is a block diagram showing the hardware configuration of an embodiment of the present invention, and Figure 3 is a Japanese language analysis process. Figure 4 is a PAD diagram of term recognition processing, Figure 5 is PAD diagram of modified expression recognition processing, Figure 6 is PAD diagram of intermediate expression expansion processing, and Figure 7 is Japanese language analysis. A configuration diagram of an example of word dictionary information for reference in processing,
Figure 8 is a schematic PAD diagram showing input sentence semantic content rJfi recognition sentence generation processing, and Figure 9 is a PAD diagram of intermediate expression expansion processing.
Figure 10 is a PAD (PAD) diagram of confirmation sentence generation processing, Figure 11 is a configuration diagram of an example of a segmented Japanese sentence expressing operation instructions for information retrieval, and Figure 12 is a recognition result of the input Japanese sentence. Configuration diagram of an example of a bunsetsu table representing 13th
Figure 14 is a configuration diagram of an example of an intermediate expression, Figure 14 is a configuration diagram of an example of an intermediate expression expansion table, Figure 15 is a configuration diagram of an example of a confirmation sentence, and Figure 16 is a clause representing an operation instruction for information retrieval. FIG. 17 is a configuration diagram of another example of a segmentation table showing the setting result of an input Japanese sentence; FIG. 18 is a configuration diagram of another example of intermediate expression; 1st
Figure 9 is a configuration diagram of another example of the intermediate representation expansion table, and the 20th
Figure 21 is a configuration diagram of another example of a confirmation sentence, Figure 21 is a configuration diagram of an example of a logical instruction word relationship table, Figure 22 is a configuration diagram of an example of a logical operator relationship table, and Figure 23 is a configuration diagram of an example of a logical instruction word relationship table. FIG. 24 is a block diagram of an example of a command language to be expressed. FIG. 24 is a block diagram of an example of a confirmation table. ■・・・・・・Japanese Analysis Department, 2・・・・・・・
...Word dictionary, 3...Input sentence meaning and content confirmation sentence generation processing unit, 4...Logic indicator relation table, 5... - Logical operator relationship table, 6... User confirmation section, 7...
...Operation instruction command language conversion processing unit, 8...
...input/output device, 9...processor,
10... Japanese language analysis processing storage area, 11
......Storage device, 12...
Working area, 20... Word dictionary storage area, 30... Input sentence meaning/content confirmation sentence generation processing unit storage area, 40... Logical directive relation table storage area, 50...
- Logical operator relation table lure. Figure 1 Figure 3 Figure 2 Figure 4@otot Figure 5 Figure 7 Figure 6 Figure 8 Figure 11 Figure 12 Figure 1O Figure 13 Figure 14 Figure 15 Ward 16 Figure 18 Figure 19 Figure 17 District Figure 20 Figure 21 Figure 22 Figure 23 Figure $24

Claims (1)

[Claims] 1. means for inputting a Japanese sentence representing an operation instruction; means for analyzing the input Japanese sentence and converting it into an operation instruction command language; A Japanese language processing system comprising: a confirmation sentence generation and display means for generating and displaying a confirmation sentence representing the meaning of an input Japanese sentence understood by the processing system. Japanese sentence analysis result confirmation sentence generation method. 2. Claim 1, further comprising means for determining whether or not to execute the command depending on whether or not the displayed confirmation sentence and the input Japanese sentence have the same meaning or content. Japanese sentence analysis result confirmation sentence generation method described. 3. The Japanese sentence according to claim 1, further comprising updating means for modifying, deleting, or adding a confirmation sentence when the displayed confirmation sentence and the input Japanese sentence have different meanings. Analysis result confirmation sentence generation method. 4. Claim 1, wherein the confirmation sentence generation means includes a means for analyzing the input Japanese sentence to generate an intermediate expression, and a means for generating a confirmation sentence by referring to the intermediate expression. Japanese sentence analysis result confirmation sentence generation method described. 5. The intermediate expression is expressed by a logical directive relationship and a logical relationship between a word representing the attribute item name of the database and the corresponding attribute item data, and the confirmation sentence is generated based on the intermediate expression. 5. The Japanese sentence analysis result confirmation sentence generation method according to claim 4, wherein the Japanese sentence analysis result confirmation sentence generation method is configured as follows. 6. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical instruction word relationship table that defines words expressing logical instructions. 7. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical directive word relationship table that defines logical directive symbols corresponding to words expressing logical directives. 8. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical directive word relationship table that simultaneously defines words representing logical directives and logical directive symbols corresponding to the words. 9. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical operator relationship table defining words representing logical operations. 10. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical operator relationship table defining logical operators corresponding to words expressing logical operations. 11. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical operator relationship table in which words expressing logical operations and logical operators corresponding to the words are defined at the same time. 12. A logical instruction word relationship table that defines one or both of a word representing a logical instruction and a logical instruction symbol corresponding to the word, and either a word representing a logical operation and a logical operator corresponding to the word. 2. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, further comprising a logical operator relation table defining both of the above and the logical operator relation table. 13. The Japanese sentence analysis result according to claim 4, wherein the means for generating a confirmation sentence by referring to the intermediate expression includes nominative case generating means for connecting a case particle to a word representing an attribute item name in the database. Confirmation sentence generation method. 14. The means for generating a confirmation sentence by referring to the intermediate expression is configured to refer to the intermediate expression and generate a word representing an attribute item name and a word representing a logical indication of attribute item data (or a logical indication word relation table using a logical indication symbol). 5. The Japanese sentence analysis result confirmation sentence according to claim 4, characterized in that the Japanese sentence analysis result confirmation sentence comprises means for extracting a word representing a corresponding logical instruction), and complement generation means for linking the extracted word representing the logical instruction to attribute item data. Generation method. 15. Generate the smallest unit of confirmation sentence by combining a nominative generation means that connects a case particle to a word representing an attribute item in the database and a complement generation means that connects a word representing a logical instruction to attribute item data. 5. The Japanese sentence analysis result confirmation sentence generation method according to claim 4. 16. The means for generating a confirmation sentence by referring to the intermediate expression refers to the intermediate expression and extracts a word expressing a logical operation (or a word expressing a corresponding logical operation in the logical operator relation table using a logical operator). 5. The Japanese sentence analysis result confirmation sentence generation method according to claim 4, further comprising a conjunction generation means for generating a conjunction. 17. By combining the nominative generating means and the complementary generating means, a minimum unit confirmation sentence is generated, and by referring to the intermediate expression, a word expressing a logical operation (or a corresponding word in the logical operator relation table using a logical operator) is generated. The logical instruction and the meaning of the logical operation are expressed as a confirmation sentence by extracting a word expressing a logical operation and using it as a conjunction, and executing the combination of the minimum unit confirmation sentence and the conjunction. 16. The Japanese sentence analysis result confirmation sentence generation method according to claim 15. 18. The Japanese sentence analysis according to claim 1, wherein the confirmation sentence generation and display means includes means for displaying a table of words representing the meaning of the input Japanese sentence understood by the processing system. Result confirmation statement generation method. 19. The Japanese sentence analysis result according to claim 1, further comprising means for displaying an intermediate expression representing the meaning of the input Japanese sentence understood by the processing system, in addition to the confirmation sentence generation and display means. Confirmation sentence generation method. 20. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, characterized in that the confirmation sentence expressing the meaning of the input Japanese sentence is used as the input Japanese sentence subsequent to the input Japanese sentence. 21, comprising means for storing a confirmation sentence representing the meaning content of the input Japanese sentence as an input Japanese sentence subsequent to the input Japanese sentence;
2. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, wherein the stored Japanese sentence is used as the input Japanese sentence after the input Japanese sentence. 22. The Japanese sentence analysis result confirmation sentence generation method according to claim 7, further comprising means for creating a logical directive word relationship table by referring to a word dictionary in which information possessed by words is defined in advance. 23. The Japanese sentence analysis result confirmation sentence generation method according to claim 9, further comprising means for creating a logical operator relationship table by referring to a word dictionary in which information possessed by words is defined in advance. 24. The Japanese sentence analysis result confirmation sentence generation method according to claim 1, wherein the Japanese sentence representing the operation instruction is for information retrieval, and is applied to an information retrieval system in Japanese. 25. Japanese sentence analysis result confirmation sentence generation according to claim 1, characterized in that it is applicable to a natural language interface method that analyzes a Japanese sentence representing an operation instruction inputted into a computer and provides operation instructions to the computer. method. 26. In a Japanese language analysis system having means for inputting Japanese and means for analyzing the input Japanese, a confirmation sentence representing the meaning of the input Japanese analyzed by the Japanese language analysis system is generated. A Japanese sentence analysis result confirmation sentence generation method characterized by comprising confirmation sentence generation and display means for displaying the confirmation sentence.