JP2592995B2 - Phrase extraction device - Google Patents

Description

The present invention relates to a phrase extracting apparatus, and more particularly, to a phrase extracting apparatus suitable for a language processing device such as a word processor, a translation device, a proofreading device, and a device using a database. The present invention relates to a phrase extracting device.

(B) Conventional technology Conventionally, in order to extract a phrase for translation or proofreading from a converted Japanese document into which the information of the phrase delimiter is not inserted, an independent word dictionary, an auxiliary word dictionary, an affix dictionary, etc. Dictionaries and a table showing the connection relationship of those elements,
Segments are extracted by referring to tables such as a grammar table.

In addition, a method has been considered in which character type information is used and, for example, a change in character type is used as a break between phrases.

(C) Problems to be Solved by the Invention In such a conventional phrase extraction apparatus using dictionaries and tables, (1) a large amount of storage devices are required to store the dictionaries and tables. Become.
(2) The time required for the phrase extraction processing for performing a dictionary search or a table search becomes longer. (3) A control program for segment extraction becomes complicated. (4) When a phrase is cut using character type information, the phrase is cut out according to the character type mechanically. (5) There are various problems such as the possibility of mistakes in segmentation at words in kana notation.

The present invention has been made in view of the above circumstances, and provides a phrase extracting apparatus that can solve the above problem.

(D) Means for Solving the Problems FIG. 1 is a block diagram showing the basic configuration of the present invention. Referring to FIG. 1, the present invention provides a sentence storing means 1 for storing a sentence, an instructing means 2 for giving an instruction to read out a desired sentence from the sentence storing means 1, and a sentence storing unit instructed from the sentence storing means 1 in sentence units. Reading means 3 for reading the sentence and character type discriminating means 4 for discriminating the character type such as kanji, hiragana, katakana or the like and the punctuation point based on the unique character code for each character of the sentence and generating a discrimination symbol. A converting means 5 for replacing the discrimination symbol with one of two kinds of codes and outputting the discriminated symbol, and a discrimination result accumulating means 6 for sequentially accumulating the codes output from the converting means 5 in read text units.
And the code string stored in the determination result storage means 6,
A phrase extracting means 7 for detecting a predetermined code transition point and inserting a break in the phrase read based on the transition point, and a specific hiragana character including a noun, a pronoun, an adverb, a conjunction, etc. A specific character dictionary means 8 storing a large number of strings, and a phrase break is inserted by a phrase cutout means, and then a hiragana character is used to determine whether or not the specific hiragana character string exists in the sentence into which the phrase break has been inserted. Searching toward the beginning of the column, if there is a specific hiragana character string, insert a break in the sentence as the one hiragana character string in the sentence, and correct the segment cutout correction position of the segment break insertion Means 9 and storage means 10 for storing a sentence outputted from the phrase segmentation correcting means 9
Output means for visually outputting the sentence stored in the storage means 10
11 is a phrase segmentation device.

In the present invention, it is preferable that each of the hiragana character strings stored in the specific character dictionary means 8 is stored in the same order as the hiragana character extraction order in the search.

The phrase extraction device according to the present invention can be applied to a word processor, a translation device, a proofreading device, a device using a database, and the like. Further, a phrase extraction process is also required in a device that outputs a sentence by voice. Can be applied.

(E) Operation According to the present invention, when a unit of the extraction process, for example, one sentence is cut out from the sentences stored in the sentence storage unit 1, the character type of each character constituting the sentence is determined by the character type determination unit 4 And the result of the determination is
Any of the types of codes, for example, L or H, is replaced by the determination result storage means 6 sequentially. Next, when detecting a predetermined code transition point, for example, a transition from L to H, the phrase cutout means 7 inserts a break in the read sentence. Next, the phrase extraction correcting means 9 recognizes a matching Hiragana character as one phrase by comparing the hiragana character specified in the specific character dictionary means 8 with the input character string, going back from the break of the phrase. It acts to output the obtained result to the output means 11 so that the position can be corrected and the corrected result can be confirmed.

(F) Embodiment The present invention will be described in detail below based on an embodiment shown in the drawings.
The present invention is not limited by this.

FIG. 2 is a block diagram showing an embodiment in which the present invention is applied to a word processor. In the figure, reference numeral 20 denotes a word processor. Reference numeral 21 denotes a text storage device, which can be composed of, for example, a floppy disk or hard disk as an external storage device, for example, a RAM as an internal storage device, or a database, for example, as another storage device. Word documents are accumulated. Reference numeral 22 denotes a keyboard as instruction means, which is provided with character input keys for performing text editing, text proofreading, etc., various instruction keys, etc., for inputting text and for inputting a desired text from the text storage device 21. Input an instruction to read. Reference numeral 23 denotes a reading device that cooperates with the CPU 24, and reads a sentence specified by the keyboard 22 from the text storage device 21. Numeral 25 is a character type discriminating device which cooperates with the CPU 24. The character type such as kanji, hiragana, katakana, etc. is discriminated one character at a time in the sentence read out from the sentence storage device 21, and the punctuation mark is discriminated. Numeral 26 denotes a result accumulating device as a judgment result accumulating means and a memory means. The result accumulating device is constituted by a RAM, and sequentially accumulates the character type judgment results, and stores a sentence cut by a later-described phrase cutout device.

Numeral 27 is a character type digitizing device as a conversion means cooperating with the CPU 24. The first code, specifically "H", is used for the kanji and katakana as the character type discrimination result by the character type discriminating device 25. ", And a second code, specifically" L ", for the hiragana and the punctuation mark, thereby converting the character type discrimination result to one of the two codes" H "or" L ". replace. And the codes “H” and “L” which are the result of character type determination
Are stored in the result storage device 26 via the CPU 24.

The phrase extraction device 28 cooperates with the CPU 24 to
When a predetermined code transition point is detected for the code strings “H” and “L” stored in 6, a break in the sentence is inserted in the read sentence.

Reference numeral 29 denotes a specific character dictionary table composed of a ROM, which is composed of adnominals, pronouns, adverbs, conjunctions, etc., and stores a large number of specific hiragana character strings accompanied by phrases. Each hiragana character string stored in the specific character dictionary table 29 is stored in the same order as the hiragana character extraction order in the search.

Reference numeral 30 is a phrase extraction correction device that cooperates with the CPU 24, searches for whether the specific hiragana character string is present in the sentence in which the break in the phrase is inserted, and if a specific hiragana character string is present, Using a hiragana character string as one phrase, a break in the phrase is inserted into the sentence, and the position of the phrase break insertion is corrected.

Reference numeral 31 denotes a display device as an output unit connected to the CPU 24 via the output control unit 32, which is configured by a dot matrix type display device such as a CRT or an LCD, so that a break in a phrase can be confirmed. Displays the sentence where the break was inserted.

In such a configuration, the phrase extraction processing will be described according to the flowchart shown in FIG. 12 using the example sentence "Materials to be submitted to the Patent Office were created and electronically filed by this word processor."

FIG. 3 shows a state in which, for example, one sentence is cut out from the sentence storage device 21 in which a sentence file and other databases are stored in accordance with the processing unit, and the sentence is stored in the result storage device 26. As described above, when a desired sentence is cut out from the sentence storage device 21 in accordance with the processing unit, for example, one sentence, one paragraph, one chapter, and the like (step 40), the character type of the sentence is determined. The code is determined (step 41). Specifically, since each character of the read sentence is assigned a unique character code such as a JIS code, by comparing the character code with each condition of the code determination table shown in FIG. , To determine the character type. Cc in condition
Are the characters to be used for character type determination, a1 and b1 represent the beginning and end of the Kanji code, a2 and b2 represent the beginning and end of the Hiragana code, a3 and b3 represent the beginning and end of the Katakana code, and a4 Represents a period.

The determined character type code is stored in the result storage device 26 (step 42) and digitized by the character type digitizing device 27 (step 43). FIG. 5 shows a code digitization table referred to for digitizing the determined output code. That is, if the character type is determined to be kanji, it is converted to the code "H", and if it is determined to be hiragana, it is converted to the code "L".
, The phrase is replaced by the code "L" respectively. The result of digitizing the sentence by collation with the code digitization table is stored in the result storage device 26 as a code string shown in FIG.

Next, the digitized result is compared with a cut-out determination table shown in FIG. 7 to determine a segment break (step
44). The determination of a segment break is as follows: (1) A break of a keyword (phrase) is made at a transition point between “L” and “H” in the digital output. (2) Discrimination is made by inserting a break between keywords after a period.

Next, a cutout symbol, for example, “/” is inserted at the break based on the determination of the break by collation with the cutout determination table, and the sentence into which the cutout symbol is inserted is stored in the result storage device 26. FIG. 8 shows the phrase segmentation result obtained by the above processing.

When a segment is cut out using only the information of the character type change, as shown in FIG. 8, when a word is written in hiragana, there is a defect that a segment containing the word cannot be cut out. In the present example sentence, the pronoun "Kono" is absorbed in the previous phrase, and "Material is this" is one phrase. It is a feature of the present invention to correct such segmentation mistakes caused by Hiragana notation.

In order to realize this, a specific character dictionary table 29 in which word strings of representative hiragana are accumulated is used. A set of word strings in Hiragana notation can be created as a well-known fact from ordinary language life. For example, in terms of part-of-speech classification, there are many such hiragana words in adnominals, pronouns, adverbs, conjunctions, and the like.

FIG. 9 shows an example of a hiragana notation table for explaining the specific character dictionary table 29. The figure shows typical words for explanation, which are sorted in the order of character strings of ordinary words. FIG. 10 shows the contents of the specific character dictionary table 29 in which the above hiragana notation table is sorted so that it can be looked up in reverse.

Subsequent to step 44, by referring to the specific character dictionary table 29, a character string portion is searched for a hiragana character string portion from the character string whose phrase has been cut as shown in FIG. 8 (step 45). The order of words stored in the specific character dictionary table 29 is, as described above,
This is the same as the search order of the character strings, and the matching with the word can be released when the matching of one character fails, so that the time until the matching can be reduced. Further, since it is possible to confirm that the collation has failed halfway, the capacity of the buffer can be reduced. This process is performed between hiragana character strings in which the character type changes from hiragana to another character type.

In the case of this embodiment, "this" in the part of "this" succeeds in the collation (step 46). If the collation is successful, insert a break in the phrase so that the phrase starts at the beginning of the hiragana notation word, and correct the position of the phrase (step
47). As a result, the segment cutout position of the sentence in FIG. 8 is corrected as shown in FIG. Next, the correction result is stored in the result storage device 26 (step 48).

Next, if the end condition is no, that is, if there is a next sentence to be segmented, control is performed to read the next sentence from the sentence storage device 21 and to control the storage positions so as not to overlap (step 49). ).

If yes in step 49, that is, if there is no longer any object to be cut out, the necessary information is stored in the result storage device 26.
And the process ends (step 50).

(G) Effects of the Invention According to the present invention, (1) when extracting a phrase, a dictionary is not used, so that the configuration of the apparatus can be simplified.
Thus, the present invention can be applied not only to word processors and office computers, but also to other small devices, specifically, electronic notebooks and calculators with program functions. (2) The phrase extraction processing and the keyword search processing can be performed at high speed. (3) A control program for segment extraction can be simplified.
(4) When extracting a phrase, there is a drawback that the position of the phrase cannot be known because Japanese is written in solid, so it takes a lot of processing and time to determine where to start and end the phrase. I have. According to the present invention, the language processing can be performed after the phrase position is determined, so that the processing time can be significantly reduced.
(5) Since the result of character type discrimination is converted into a binary value by a digital circuit and processed, the processing speed is increased, the circuit is simplified, and a phrase segmentation device can be realized at low cost. (6) It is possible to reduce mistakes in segment extraction positions due to Hiragana written words. (7) The comparison and collation between the hiragana character string of the specific character dictionary and the input character string can be performed quickly, and the collation can be realized with a small capacity of the buffer memory required for the dictionary.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic configuration of the present invention, FIG. 2 is a block diagram showing a configuration of a word processor according to an embodiment of the present invention, and FIG. 3 is an example of a sentence stored in a text storage device. FIG. 4 is an explanatory diagram showing the contents of a character type discrimination table, FIG. 5 is an explanatory diagram showing the contents of a code digitization table, FIG. 6 is an explanatory diagram showing the character type discrimination results, and FIG. Is an explanatory diagram showing the contents of the phrase extraction discrimination table, FIG. 8 is an explanatory diagram showing the extraction result, FIG. 9 and FIG.
FIG. 10 is an explanatory diagram showing the contents of the specific character dictionary table, and FIG.
FIG. 12 is an explanatory diagram showing the cutout position correction result, and FIG. 12 is a flowchart showing the processing operation of the embodiment. 1 ... sentence storage means, 2 ... instruction means, 3 ... readout means, 4 ... character type determination means, 5 ... conversion means, 6 ... determination result storage means, 7 ... phrase extraction means, 8 ... ... Specific character dictionary means, 9 ... Phrase cutout correction means, 10 ... Storage means, 11 ... Output means.

Claims (2)

(57) [Claims] 1. A text storage means for storing text, an instruction means for instructing reading of a desired text from the text storage means, a reading means for reading the specified text from the text storage means in text units, Character type determining means for determining a character type and a punctuation mark, such as kanji, hiragana, and katakana, based on a unique character code for each character of the sentence, and generating a determination symbol; Conversion means for replacing and outputting the code, determination result storage means for sequentially storing the codes output from the conversion means in units of read sentences, and a predetermined code transition point for a code string stored in the determination result storage means. A phrase segmentation means that detects and inserts a segment break in the sentence read out based on this transition point, and a specific phrase with a phrase, consisting of adnominal, pronoun, adverb, conjunction, etc. After inserting a break of a phrase by a specific character dictionary means storing a large number of character strings and a phrase extracting means,
A search is performed toward the beginning of the Hiragana character string to determine whether the specific Hiragana character string is present in the sentence into which the segment break has been inserted, and if the specific Hiragana character string is present, the Hiragana character string is set to 1 A phrase cut-out correcting means for inserting a break of the phrase into the sentence as one phrase and correcting the insertion position of the phrase break, a storage means for storing a sentence outputted from the phrase cut-out correcting means, and a sentence stored in the storage means And an output unit for visually outputting a phrase. 2. The phrase extracting apparatus according to claim 1, wherein each of the hiragana character strings stored in the specific character dictionary means is stored in the same order as the hiragana character extraction order in the search. .