JPS6320564A - Natural language processor - Google Patents

Abstract

PURPOSE:To evade waste in case of analyzing a sentence structure, so that a natural language processing can be executed efficiently and quickly, by truncating and executing a sentence structure analysis processing by unit information of a phrase and a clause which are held in a sentence structure analysis unit information holding means. CONSTITUTION:Whether a sentence which is inputted from an input part 1 has been held in an input sentence holding part 2 or not is decided, and when the input sentence has been held, a sentence structure analysis processing part 3 executes a sentence structure analysis processing by using a grammar rule which is held in a sentence structure analyzing information holding part 4, and sentence structure analyzing information of a word dictionary, etc. Subsequently, whether names of a 'noun phrase' and a 'verb phrase', etc. of a sentence structure tree which is generated by the sentence structure analysis processing part 3 are contained in a sentence structure analysis unit holding part 5 or not is checked. Whether a symbol for showing the end of a sentence, such as a 'punctuation mark' and a 'question mark' (?) is contained in a part where the sentence structure analysis processing is completed by the sentence structure analysis processing part 3 or not is decided. Subsequently, the part where the sentence structure analysis processing is completed by the sentence structure analysis processing part 3 is deleted from the input sentence holding part 2, and added to a sentence structure analysis result holding part 6 as a sentence structure analysis result.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a processing device for natural languages such as Japanese and English.

[Prior Art] Conventional natural language processing devices process sentences inputted from input means such as a keyboard, microphone, or input tablet and pen from syntactic analysis information holding means that holds information such as word dictionaries and grammar rules. The syntax is parsed based on the information, the result of this syntax analysis is temporarily held, and after this, the modification and semantic analysis are performed.

In such syntactic analysis, the syntactic analysis is normally continued until the entire input sentence is constructed as one syntax tree.

An example is "The white bird ran away from the cage" as shown in Figure 3.
When parsing the example sentence, as shown in FIG. 4, the parsing is performed sequentially from the beginning of the sentence, and the sentences are stacked to form a single parse tree.

However, even with the syntax tree created in this way, the subsequent processing steps such as dependency analysis and semantic analysis will be deemed unsuccessful, and the drawing analysis will have to be restarted again.
Or you may have no choice but to ignore the syntax tree itself.

That is, to explain the above-mentioned example sentence, the semantic network considered from the meaning of the example sentence is shown in FIG. The adjective ``white'' directly modifies the noun ``bird'' and becomes the subject, and the verb ``ran away'' becomes the predicate.

However, in the stacked syntax tree shown in Figure 4, "white" and "bird" are analyzed separately, so the original semantic network of the example sentence shown in Figure 5, &-fifi''-disjunction 814.: v J:5 tt 'j%.

This has the disadvantage that it is necessary to reconfigure it into the form of the syntax tree shown in FIG. 6, which takes time and effort.

[Problems to be Solved by the Invention] Therefore, the purpose of the present invention is to eliminate the above-mentioned drawbacks, eliminate waste in syntactic analysis, and improve the processing speed of the natural 35-place Ori neck. It's about sharing.

[Means for Solving the Problems] In order to achieve the above object, the present invention provides an input means for inputting a sentence, a lateral sentence analysis means for analyzing the syntax of the sentence input by the input means, and an input means used for syntactic analysis. a syntactic analysis information holding means that holds information such as a word dictionary and grammar rules; a syntactic analysis result holding means that holds the results of syntactic analysis by the syntactic analysis means; and a syntactic analysis unit information such as phrases and clauses. The present invention is characterized in that the syntax analysis unit information is provided to the syntax analysis means to perform syntax analysis.

[Effect] The operation of the present invention having the above structure will be explained below.

The sentence inputted from the input means is parsed by the parsing means using the parsing information held in the parsing information holding means.

During this syntactic analysis process, the information content held in the syntactic analysis unit information holding means that holds syntactic analysis unit information such as phrase 1 clause is searched, and the "noun phrase" of the syntactic tree generated as a result of the syntactic analysis process is searched. '' or the name of a ``verb phrase'' is checked to see if it is included in the retained content. If it is included, this "noun phrase" etc. is sent as is to the parsing result holding means.

In this way, the syntax is analyzed for each unit of information such as "noun phrase" or "verb phrase", thereby speeding up the processing.

[Example] Embodiments of the present invention will be described below with reference to the accompanying drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

Here, 1 is an input unit for inputting sentences in various forms such as a keyboard, a microphone, etc.

Reference numeral 2 denotes an input sentence holding section that holds sentences input from the input section 1.

Reference numeral 3 denotes a syntactic analysis processing unit that parses the sentence held in the input sentence holding unit 2. The syntax analysis processing unit 3 is provided with a ROM 3A having an area in which a control program according to the flowchart of FIG. 2 is stored, for example.

Reference numeral 4 denotes a syntactic analysis information holding unit that holds information such as grammar rules and word dictionaries necessary for such syntactic analysis.

Reference numeral 5 denotes a syntactic analysis result holding unit that holds syntactic analysis unit information such as "phrase" or "clause j" used in syntactic analysis.

Reference numeral 6 denotes a parsing result holding unit that holds the parsing results of the parsing processing unit 2.

Next, an example of the control procedure of this embodiment having the above configuration will be explained in a second example.
This will be explained with reference to the flowchart shown in the figure.

First, when the program starts, it is determined in step S1 whether the sentence input from the input section 1 is held in the input sentence holding section 2 or not. If it is not held, this step Sl is repeated until it is held.

When the input wheel is held, the process proceeds to step S2, where the syntactic analysis process is performed using the syntactic analysis information such as grammar rules and word dictionaries held in the syntactic analysis information storage unit 4. Section 3 performs horizontal text analysis processing.

Next, in step S3, it is checked whether the names of "noun phrase", "verb phrase", etc. of the syntax tree generated by the syntax analysis processing section 3 are included in the syntax analysis result holding section 5. If it is not included, the process returns to step S2 and the parsing process is repeated. If it is included, the process advances to step S4.

In step S4, it is determined whether or not the portion for which the syntactic analysis process has been completed by the syntactic analysis processing unit 3 includes a symbol representing the end of a sentence, such as a "full stop" or "question mark" (?). If it is not included, the process proceeds to step S5; if it is included, the process proceeds to step S6.

In both steps S5 and S6, the part for which the syntax analysis process has been completed by the syntax analysis processing section 3 is deleted from the input sentence holding section 2, and added to the syntax analysis result holding section 6 as a result of the syntax analysis.

After the above processing, in the case of step S5, since syntax analysis is in progress, the process returns to step S2.

On the other hand, in step S6, since the syntactic analysis of one sentence has been completed, the process moves to the end and ends the process.

The syntax analysis result holding unit 6 sends the processing results to a necessary location for subsequent semantic analysis processing, etc.

As an example, "noun phrase" is stored in the parsing unit information holding unit 4,
A case will be described in which parsing unit information such as "adjective phrase Jtr verb phrase" and "terminal symbol" is held and the previously mentioned example sentence (see FIG. 3) is parsed.

First, an example sentence "The white bird ran away from the cage" is stored in the input sentence storage unit 2, and then parsing processing is performed (step S2).
is performed, but since "kago" + "kara" is a "noun phrase", as a result of comparison with the parsing unit information holding unit 4,
If the match is found (step S3), the parsing is interrupted, the part "from the basket" is deleted from the input sentence holding section 2, and the parsing result "from the basket" is held in the parsing result holding section 6 (step S3). S5).

Hereinafter, if you return to step S2 and continue the process in the same way, since "white" + "i" is an "adjective phrase", as is clear from the above explanation, this "white" will be stored in the input sentence holding section. 2 and is retained in the parsing result retaining unit 6. This process is repeated, and in the end, the syntax tree shown by the solid line in FIG. 7 is obtained.

Thereafter, by performing semantic analysis of this syntax tree, a semantic network as shown in FIG. 5 is obtained.

In short, as is clear from comparing the above-mentioned syntax tree in Figure 6 with the syntax tree in Figure 7, in the syntax tree in Figure 7, the parts indicated by broken lines are Since no syntax tree is generated, the processing speed of only that sentence is increased.

In the above embodiment, an example has been described in which the input sentence is temporarily held in the input sentence holding section, but it may be sent directly to the parsing section and step S1 shown in FIG. 2 may be omitted. It goes without saying that the syntactic analysis information may include knowledge expression information in addition to grammar rules and word dictionaries.

[Effects of the Invention] As is clear from the above description, according to the present invention, the horizontal sentence analysis process is performed while terminating using the unit information of phrases and clauses held in the syntactic analysis unit information holding means, so that the lateral sentence Natural language processing is performed efficiently and quickly by eliminating waste in analysis.

[Brief explanation of the drawing]

FIG. 1 is a block configuration diagram showing an embodiment of the present invention, FIG. 2 is a flowchart showing an example of a control procedure of the embodiment, and FIG. 3 is a diagram showing an "example sentence" used to explain syntax analysis processing. , Figure 4 shows one of the conventional syntax trees for the "example sentence", Figure 5 shows the semantic network of the "example sentence", and Figure 6 shows the conventional syntax tree for the "example sentence". Another diagram, FIG. 7, is a diagram showing a syntax tree used in the present invention for the above-mentioned "example sentence". DESCRIPTION OF SYMBOLS 1...Input unit, 2...Input sentence holding unit, 3...Syntax analysis processing unit, 3A...ROM having a control program storage area. 4... Syntactic analysis information holding unit, 5... Syntactic analysis unit information holding unit, 6... Syntactic analysis result holding unit. 4-Invention 1.77 Figure 1 Winter Invention Actual Flow Chart Figure 2 9 Column Text Figure 3 Figure 5 shows the old and taste network of the 粂 Dao corner - Bunne 1.

Claims (1)

[Scope of Claims] An input means for inputting a sentence, a syntactic analysis means for analyzing the syntax of the sentence input by the input means, and a syntactic analysis device for holding information such as a word dictionary and grammar rules used for the syntactic analysis. a syntax analysis result retention means for retaining the results of syntax analysis by the syntax analysis means; and a syntax analysis unit information retention means for retaining syntactic analysis unit information such as phrases and clauses, A natural language processing device characterized in that parsing is performed by supplying analysis unit information to the parsing means.