JPS63109573A - Processing system for adjunct interrogative - Google Patents

Abstract

PURPOSE:To simplify the processing of an anlysis by finding the part of an adjunct interrogative of English as a morphological pattern and separating and processing the part of the adjunct interrogative from an object to be analyzed. CONSTITUTION:When there is the adjunct interrogative in an English text 2, it is understood morphologically as the adjunct interrogative and the part of the adjunct interogative is separated from the part to be analyzed, then the analysis processing is simplified. In the processing of the adjunct interrogative when ',' comes and below mentioned word groups succeed, ',' to '?' is deleted as an unit and a flag is stands as block information. A) As the form of the adjunct interrogative, there are, (auxiliary verb)+(personal pronoun)? or the like. b) As the type of the auxiliary verb, there are am or the like. c) As the type of the personal pronoun, there are I or the like.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a language analysis device, and more particularly, to a method for processing additional questions in natural language analysis in a machine translation device or the like.

Although the forms of additional questions in English are very limited, in ordinary analysis methods, the usage of the auxiliary verbs and pronouns that make up the additional questions, the subdivision of parts of speech, etc. are wide-ranging. This results in complicated processing. For example, if additional questions are also handled during syntactic analysis, (1) It is difficult to recognize partially the same structure as an ordinary question as an additional question.

(2) Even if recognition is possible, there is a risk that a large number of useless partial analyzes will be performed during the analysis.

(3) (a) A+, can't you'7;
(b) A', can you7 (c) A+, can
you not'7; (d)～, don't
You7 etc. are recognized as different structures, but they are actually interpreted as indicating doubts about the main sentence.
Difficult (because they are structurally different).

There are drawbacks such as.

Purpose The present invention was made in view of the above-mentioned circumstances.
In particular, this was done with the aim of making the analysis process easier by grasping the fact that it is an additional question in terms of form and separating the additional question part from the part to be analyzed.

Structure In order to achieve the above object, the present invention is characterized in that the portion of the additional question in English is found as a morphological pattern, and the portion of the additional question is separated from the analysis target and processed. Hereinafter, the present invention will be explained based on examples.

FIG. 7 is a diagram for explaining the overall configuration of an English-Japanese automatic translation device to which the additional question processing method according to the present invention is applied. Needless to say, the present invention can be effectively applied not only to the device but also to any language analysis device that analyzes sentences in an input language when translating one language into another language.

In Fig. 7, 1 is the input section, 2 is the English text, 3 is the pre-editing section, 4 is the morphological analysis section, 5 is the syntax analysis section 1, 6 is the syntax analysis section, 7 is the operation display section, and 8 is the word Dictionary, 9 is an analysis rule file, 10 is a control section, 11 is a structure conversion section,
12 is a translation generation unit, 13 is a post-editing unit, 14 is an output unit, 1
5 is a Japanese sentence, and as shown in the figure, this translation device has an input section 1 through which an English text 2 to be translated into 1 Japanese is input. The input unit 1 includes, for example, a keyboard having character keys such as alphanumeric keys and function keys, and an optical character reading device (optical character reading device) that reads English text recorded on paper.
(OCR) and/or a file storage device for reading English text recorded on a storage medium such as a magnetic disk.

The English text input by the input unit 1 is sent to the pre-editing unit 3
, and pre-processing for translation is performed. here,
It mainly processes sentences and unknown words. This functions as part of morphological analysis.

The pre-edited English text data is transferred to the morphological analysis unit 4 together with the information obtained in the pre-editing. The morphological analysis unit 4 indexes the word dictionary 8, divides it into sentences, analyzes the morphemes of the English sentence, processes unknown words, performs various summarizations such as proper nouns, time table 2, number expressions, etc., and adds Processes entire sentences such as questions and recognition of apposition. The morphological analysis rules are stored in the analysis rule file 9.

The morphologically analyzed English data is transferred to the syntactic analysis unit 5 together with the dictionary information obtained through the morphological analysis. The syntactic analysis unit 5 is a functional unit that applies grammar rules to English data to analyze the surface structure of a sentence and find all syntactic possibilities.

The English data parsed by the parsing unit 1 5 is sent to the parsing unit 6 together with the analysis information.

Here, a structural description is applied to select a solution from the superficial syntactic analysis results obtained by syntactic analysis I. Through this, a reliable parse tree of English description is created and its structure is created. These parsing rules are still stored in parsing rule file 9.
is stored in.

The parsed English data is transferred to the structure conversion unit 11 as parse tree data. The structure conversion unit 11 creates a syntax tree of a corresponding Japanese sentence from a syntax tree that is an intermediate structure of an English sentence, and converts one Japanese sentence into a Japanese basic structure that is easy to translate.

The syntax tree data indicating the basic structure of the Japanese language whose structure has been converted in this way is sent to the translation generation unit 12, and the latter generates a translation. This is a functional unit that generates Japanese sentences from the tree structure of a Japanese syntax tree.

The translated Japanese text data, that is, the translated text data, is sent to the post-editing section 13. The post-editing unit 13 uses the information used in the translation process to index the dictionary 8 and correct the translated data to complete a more natural Japanese sentence. This Japanese sentence data is transferred to the output section 14, and the translated Japanese sentence data is transferred to the output section 14, and outputted from the output section 14 as a translated Japanese sentence 15. The output section 14 is
Examples include printers, displays, and/or file storage devices such as magnetic disks.

The flow of a series of these translation processes is controlled by a control unit 10 that controls the entire control of this apparatus.

In the case of the illustrated example, dictionary data for English and Japanese words is stored in the word dictionary v8, and not only word 當, but also dictionary data for English and Japanese words are stored.
Various information such as dependency relationships, that is, co-occurrence relationships, meanings, singularity, plurality, parts of speech, etc. are described.

Further, the analysis rule file 9 stores rule data for morphological analysis and English sentence analysis.6 The control unit 10 is connected to the operation display unit 7. The operation display section 7 is used by an operator to give various instructions to the apparatus. For example, operation keys such as translation instruction keys, cursor keys, and input English text.

It has a display and indicators that visually display the translated Japanese text, intermediate data such as dictionary information, and various instructions for the operator. Note that many of these operation display functions may be configured to be included in the keyboard if the input unit 1 is equipped with a keyboard, or in the display if the output unit 14 is equipped with a display. Further, if the output section 14 is equipped with a display, it may be configured to be included in the display.

The present invention relates to a machine translation device as described above.

When there is an additional question in English text 2, this method simplifies the analysis process by recognizing the additional question in its form and separating the additional question from the part to be analyzed. .

In addition, in the processing of additional questions, when the following word group continues when "," is received, the following words from "," to "?" are deleted as a unit, and a flag is set as block information.

b) The form of an additional interrogative sentence is 1. (auxiliary verb) + (personal pronoun)? , (auxiliary verb) n't + (personal pronoun)? , (auxiliary verb) +
(Personal pronoun) + not'7 etc.

口) Types of auxiliary verbs.

an+ is are was wer
edo does have has
had will 5halls
could not could
ay might must won'
5han't can't need dare
There are used etc. Also, c) types of personal pronouns are: I you he she it we
They et al.

FIG. 1 is an overall block diagram for explaining one embodiment of the present invention, and FIG. 2 is a flowchart for explaining the operation of FIG. 1. In the figure, 2o is an input device, 21 is a text The unit division section 22 is a division dictionary.

23 is a word extraction part, 24 is a primitive (deliI
Ilit) table, 25 is a dictionary search unit, 26 is a reference dictionary, 27 is an additional question processing unit, 28 is a sentence information buffer, 2
Reference numeral 9 denotes a dictionary information storage table. First, the input processing section 20 takes in the sentence information to be processed. Next, the sentence unit division section 21 refers to the division dictionary 22 and divides the input sentence based on the division information. Divide into sentences. Next, the unit extraction unit 23 performs extraction in dictionary lookup units.
mit) Take out boundary information from the table 24 and cut out sentence units as dictionary lookup units. As a result of cutting out the dictionary lookup unit, if there is no additional question in the cutout unit, the dictionary is searched and written to the dictionary information storage table, and if there is an additional question, the additional question is processed. In addition, how to divide into sentence units.

Any type of string is fine, but for example, a group of sentence-end delimiter strings is provided, and the sequence immediately after the previous sentence-end delimiter group or from the beginning of the input to the first sentence-end delimiter string that appears is recognized as one sentence. do. The sentence-final delimiter group is 1. period,! (Exclamation point),? (question mark), = (colon), etc. These can be used in conjunction with quotation marks, but quotation marks are optional and may or may not be present b).

The dictionary search unit extraction is a character string starting from the character immediately after the dictionary search delimiter character string group, and the dictionary search delimiter character string group includes punctuation marks.

There are quotation marks, spaces, etc. Also, dictionary searches.

An entry exists (the search was successful) when a substring from the beginning of the string matches an entry in the dictionary, and immediately after the substring is a dictionary search string group. do. If the delimiter character is a punctuation mark or quotation mark,
Register independently in the dictionary information storage table. In addition, a one-sentence information buffer is provided separately from the dictionary information table, and additional question information is written in this buffer and saved in the sentence section.

FIG. 3 is a block diagram for explaining details of additional question processing according to the present invention, and FIG. 4 is a flowchart for explaining the operation of FIG. 3. In the figure, 30 is a pointer setting section; is a comma determination unit, 32 is a determination unit for the α group word following the comma, 33 is a determination unit for the β group word following the comma, 34
35 is a determination unit for the next pronoun after the word in the α group, 36 is a determination unit for the question mark following the pronoun, 37 is a determination unit for not after the pronoun, 38 is a determination unit for the question mark following not, 39 is a determination unit for the question mark following the pronoun, and 40 is a deletion processing unit. First, a pointer is set to the first word of the word information, and if the pointer has not been set yet, determines whether it is a comma or not. Note that in this specification and drawings, words in the α group refer to words that include an auxiliary verb or a be auxiliary verb in the part of speech (those that are not negative).

Words in the β group are negative forms like auxiliary verbs or b in the part of speech.
e Refers to things that include the negative form of auxiliary verbs. If it is a comma, keep the pointer still and determine whether the word next to the comma is a word in the α group. If it is a word in the α group, the next word in the α group is a pronoun. If it is a pronoun, determine whether the next word after the pronoun is not, and n
When it's ot.

Determine whether the word following not is a question mark or not.

If it is a question mark, information indicating that it is an additional question is written into the sentence information buffer. Further, if the word following the pronoun is not "not", it is determined whether the word following the pronoun is a question mark or not, and if it is a question mark, information indicating that there is an additional question is written in the sentence information buffer.

When the information that is an additional question is written to the one-sentence information buffer in this way, the ~? is deleted from the word information table and the pointer is advanced by one. Also, if the word following the comma is not a word in the α group, the word following the comma is β
Determine whether it is a group word or not, and if it is not a group word, if the word next to the pronoun is not a question mark, and not
The pointer is advanced by one in the same way as if the next word is not a question mark.

When the word following the comma is a word in the β group, determine whether the next word in the β group is a pronoun, and if it is a pronoun, determine whether the word following the pronoun is a question mark or not. If it is a question mark, information indicating that it is an additional question is written to the sentence information buffer.

FIG. 5 shows an embodiment in which information is exchanged between the dictionary storage table 29 and each determination section 31 to 39 in FIG. 3 via the bus control device 50. The figure is an overall detailed block diagram when the circuits shown in Figures 1 and 5 are combined, but since the details of each part in Figures 5 and 6 have already been explained, they have the same effect. The same reference numerals are used to refer to the same parts.

Tables 1 to 4 are I vent to 5chool, d
This table shows an example of processing the additional question mark idn't I'7. Table 1 shows the contents of the reference dictionary, Table 2, and Table 3.
Table 4 represents the changing state of the contents of the dictionary information storage table, and Table 4 represents the contents of the sentence information buffer.
First, an input sentence is extracted into sentence units using a segmentation dictionary. (Leaving space below) Table 1 Table 2 Table 3 Table 4 Next, words are extracted based on the delimit information of the table containing the primitive table boundary line data. A dictionary search is performed by comparison, and each search data is recorded in the dictionary information storage table (Table 2). Once all the searches have been completed for the units divided into sentence units, processing of additional questions is next performed.

Additional question information is searched and determined in order from the first word of the extracted word information. While referring to the dictionary information storage table shown in Table 2, determine the functional relationship and existence of commas, alpha group 8, beta group words, pronouns, not, and if there is a question mark next to one pronoun + not. Next, if there is a question mark, write information that there was an additional question to the sentence information buffer (Table 4).6 Then, delete the data recognized as the additional question from the contents of the dictionary information storage table (Table 2). Rewrite it as shown in Table 3 (
p didn't I '7 in Table 2 is deleted from the dictionary information storage table). These data are output for each sentence, transferred to the dictionary information storage table and the sentence information buffer, and recorded. The analysis processing side receives the data necessary for analysis in a form that distinguishes the processing data of additional questions from other information.

Effects---As is clear from the above explanation, according to the present invention, the fact that it is an additional question is understood morphologically, and the additional question part is separated from the part to be analyzed, which simplifies the analysis process. It can be done.

[Brief explanation of the drawing]

FIG. 1 is an overall block diagram for explaining one embodiment of the present invention, FIG. 2 is a flowchart for explaining the operation of FIG. 1, and FIG. 3 is an additional block diagram for explaining an embodiment of the present invention. 4 is a detailed block diagram of the question processing unit 27, FIG. 4 is a flowchart for explaining the operation of FIG. 3, and FIG. A block diagram illustrating an example of a case where information is exchanged through a bus control device, FIG. 6 is a combination of the block diagram shown in FIG. 1 and the block diagram shown in FIG. 5. FIG. 7 is a block diagram for explaining an example of a translation device to which the present invention is applied. 2o... Input device, 21... Sentence unit division section, 22.
...Divided dictionary, 23...Word extraction section, 24...
Primitive table, 25...Dictionary search section, 26...
・Reference dictionary. 27... Additional question processing unit, 28... Sentence information buffer. 29... Dictionary information storage table, 30... Pointer setting section, 31... Comma judgment section, 32... Judgment section for the word in the alpha group next to the comma, 33... Next β after the comma Group word determination unit, 34 . . . a pronoun determination unit next to the α group word. 35... Judgment unit for the next pronoun after the word in the β group, 36...
A determination unit for the question mark following the pronoun, 37... A determination unit for the not next to the pronoun, 38... A determination unit for the question mark following the not. 39... Judgment unit for the question mark next to the pronoun, 40... Deletion processing unit, 50... Bus control device.

Claims (1)

[Claims] A method for processing additional questions that is characterized by finding the additional questions in English as a morphological pattern and processing the additional questions separately from the analysis target.