JPS63163571A - Sentence inspecting device - Google Patents

Abstract

PURPOSE:To automatically detect a part to be corrected by extracting a word in which a KANJI (Chinese character) is included in a notation of the extracted words and indicating the extracted word in a display method different from other word. CONSTITUTION:A specific part of speed extracting means 3 extract the word of the specific part of speech from an inputted Japanese sentence. A KANJI word extracting means 4 extracts the word in which the KANJI is included in the notation of the words extracted by the specific part of speech extracting means 3. Further, a word indicating means 5 indicates the word extracted by the KANJI word extracting means according to the display method different from other word. Accordingly, a part for which it seems to be preferable that a KANJI notation is changed to KANA (Japanese syllabary) notation is automatically detected by considering the part of speech of the word. Thereby, the including ratio of the KANJI is reduced, thereby, the efficiency in a correcting operation for easily reading the sentence is attained.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a text inspection device that automatically detects points to be proofread from input Japanese text.

(Prior Art) Conventional text inspection devices that automatically detect points to be proofread from input Japanese text include the ``Sentence Abnormality Detection Device'' (Japanese Patent Laid-open Publication No. 127138/1983) and the ``Text Abnormality Inspection Device'' Automatic text abnormality inspection and correction device J
Publication No. 40). These methods detect incorrect words and terms in sentences.

In other words, [Text Anomaly Inspection Apparatus J
27130 (Japanese Patent Laid-open No. 127140) detects parts of words that cannot be considered as correct, and the ``Automatic Sentence Abnormality Inspection and Correction Device'' (Japanese Unexamined Patent Publication No. 127140/1982) detects character strings that may be used incorrectly. are memorized in advance and detected from the text.

Now, if we consider the purpose of proofreading a text, it is not only necessary to detect and correct incorrect words and terms in the text as described above, but also to make the text easier to read.

It is known that the content of kanji in a Japanese text (the ratio of the number of kanji to the total number of characters in the text) has a large effect on the readability of the text. Generally, it is best to limit the use of kanji to around 40% (Yasunori Yasumoto: ``Writing Techniques for Persuasion'', Kodansha Gendai Shinsho).

Therefore, Japanese Patent Application Laid-open No. 59-94132 (Japanese Patent Application Laid-Open No. 59-94132) has introduced a Japanese word processor to limit the types of kanji to be used. In other words, this “10 Japanese words”
In "Setsusa", kanji are stratified into multiple levels, and only kanji within the specified level range are used as the result of kana-kanji conversion.

(Problem that the invention aims to solve) However, if the types of kanji used are simply restricted in order to reduce the content of kanji in a text, the notation will become unnatural and the text will become difficult to read. Sometimes. An example is shown below.

The following example sentence (1) has a kanji content rate of 42% (total number of characters: 69, number of kanji characters: 29), and the amount of kanji is rather large, giving the impression that it is difficult to read.

"Everyone knows the fact that objects fall to the ground. However, it is by no means easy to discover the law of nature called gravity from this fact."
(1) In order to reduce the number of kanji from this example sentence (1), for example, if we restrict the use of kanji such as ``mode,''``ran,''``ji,'' and ``easy j,'' we get the following (2+).

``The fact that objects fall to the ground is something that everyone knows.However, it is by no means easy to discover the natural law of gravity from this fact.
...(By restricting the use of 21 kanji, the content rate of this sentence + 21 kanji is 33% (total number of characters 7)
2, the number of kanji decreased to 24). However, there are some places where it has become easier to read, such as when ``butashi'' has been changed to ``but J,'' but the same word is written in both kanji and kana, such as ``narihado'' and ``oi''. or “
Kanji such as "shizen" are written in kana, making the text generally more difficult to read.

In order to make texts easier to read by reducing the number of kanji, we should not simply limit the types of kanji used.
It is necessary to take into account the part of speech of the word and convert the kanji notation to kana notation. For example, formal nouns, auxiliary verbs, pronouns, adverbs,
If you write conjunctions in kana, the example sentence given earlier (
1) becomes as follows.

"Everyone knows that objects fall to the ground. However, it is by no means easy to discover the natural law of gravity from this fact." (3) The content rate of kanji in this sentence (3) is 2
8% (total number of characters: 76, number of kanji: 20), giving an impression of being softer and easier to read than the first sentence (1).

Therefore, when proofreading the text, from the text, such as the above,
If a machine could automatically detect places where it would be better to change the kanji notation to kana by taking into account the parts of speech, it would help make proofreading more efficient.

The purpose of the present invention is to make the text easier to read by reducing the content of kanji in the text. Rather than simply restricting the kanji used, the purpose of the present invention is to consider the parts of speech of words and write them in kanji. To provide a text inspection device that detects a part where it is considered preferable to change it to kana notation.

(Means for Solving the Problem) The present invention provides a text inspection device that automatically detects a part of proofreading from an input Japanese text. a specific part-of-speech extracting means for extracting a specific part-of-speech; a kanji-word extracting means for extracting words whose notation includes kanji from among the words extracted by the specific part-of-speech extracting means; This is a text inspection device characterized by comprising a word presentation means for presenting words in a different display method.

(Example) The structure and operation of the present invention will be explained using the drawings.

FIG. 1 is a block diagram showing the structure of an embodiment of the text inspection device of the present invention.

In FIG. 1, text input means 1 is a means for manually inputting a character string containing kanji and kana in a Japanese text, and includes a kana-kanji conversion device equipped with a keyboard, a Bentouch keyboard, a character recognition device, and the like.

The text storage means 2 is a means for storing a character string including kanji and kana of the Japanese text inputted by the text input means 1, and uses an IC memory, a magnetic disk device, a magnetic tape device, etc.

The specific part of speech extraction means 3 is a means for extracting words of a specific part of speech, preferably written in kana, from the character string containing kanji and kana stored in the text storage means 2. Specific parts of speech that are preferably written in kana include pronouns, adjectives, adverbs, conjunctions, formal nouns, auxiliary verbs, particles,
They are auxiliary verbs, etc. Conventionally, the word dictionary has been used as described in ``Memory of a Japanese dictionary and automatic segmentation of Japanese sentences'' (Nagaya, Tatsuben, Yamagami, Renbe, ``Information Processing'' Vol. 19, No. 6, 197B). A known method is to divide a character string containing kanji and kana into words and identify the part of speech of each word.

Therefore, using this method, a string of characters containing kanji and kana is divided into words, and once the part of speech of each word has been determined, pronouns, adnominals, adverbs, conjunctions, formal nouns, auxiliary verbs, particles,
By extracting auxiliary verbs, etc., the specific part-of-speech extraction means 3 can be realized.

FIG. 3 shows a block diagram of a configuration example of the specific part of speech extraction means 3. In FIG. 3, the input string buffer 310 is
The character string read from the text storage means 2 is temporarily stored. The word dictionary 320 stores the notation, part of speech, and pronunciation of words. The word search unit 300 compares the character string in the input character string buffer 310 with the word dictionary 320 and outputs the found words. The word candidate buffer 311 temporarily stores the words output by the word search section 300. The connection table 321 is a table that describes whether or not two parts of speech can be connected. The grammar analysis unit 301 uses the connection table 321
The connection of words in the word candidate buffer 311 is checked with reference to , and clause candidates are formed and output. Clause evaluation section 30
2 is selected from among the phrase candidates in the phrase candidate buffer 312,
Determine and output the most likely phrase. The analysis result buffer 313 temporarily stores a series of clauses output by the clause evaluation unit 302. The specific part-of-speech table 323 stores part-of-speech codes for pronouns, adnominals, adverbs, conjunctions, formal nouns, auxiliary verbs, particles, and auxiliary verbs as parts of speech that are preferably written in kana. The part-of-speech search unit 303 compares the part-of-speech of the word forming the clause in the analysis result buffer 313 with the part-of-speech in the specific part-of-speech table 323, and outputs matching words. The output word buffer 314 stores the output of the part of speech search unit 303.

The kanji word extraction means 4 determines whether the words extracted by the specific part of speech extraction means 3 include kanji in the word notation, and extracts only those words in which the kanji is included in the word notation. It is a means. Whether or not a word's notation includes kanji can be easily determined by checking each character code of the word's notation (for example, JI
In the code system defined by S, codes larger than a certain value can be determined to be Kanji.)

The word presentation means 5 is means for presenting the words extracted by the Kanji word extraction means 4 in a different way from other words. For example, when a sentence is displayed on a CRT display, a presentation method different from other words may include displaying the character string corresponding to the word extracted by the kanji word extraction means 4 in reverse video, or changing the color. Possible methods include displaying the text differently or displaying it with an underline added.

Next, the operation of this embodiment will be explained using an example.

First, it is assumed that example sentence (1) is input by the text input means l and stored in the text storage means 2.

Subsequently, the example sentence (1) stored in the sentence storage means 2 is analyzed by the specific part of speech extraction means 3, divided into the following word strings, and the part of speech of each word is identified. / indicates a word break, and the part of speech is indicated in parentheses.

[object (noun) / ga (particle) / ground (noun) / to (particle), / fall (verb) / to (particle) / say (auxiliary verb)
/'fact (noun)/ha (particle)/, (comma)/progress (adverb) 7/, (comma)/'who (pronoun)/also (particle)/ga (particle)/verb to know) 7 / te (particle) / iru (auxiliary verb) / 'thing (formal noun) / de (auxiliary verb) 7 / aru (auxiliary verb) /. (punctuate) / deshiku conjunction > /, (comma) / kono (adjunctive) / fact (noun) / gara (particle) / attraction (noun) / to (particle) / say (auxiliary verb) / nature (noun) /
of (particle) / law (noun) 7 / to (particle) / 'discover (verb) / formal noun) / is (particle) /, (comma) /
Never (adverb) / 'easy (adjective) / 'particle)
/Not (auxiliary verb)/. The specific part of speech extraction means 3 extracts pronouns, adjectives, adverbs, conjunctions, formal nouns, auxiliary verbs, particles, and auxiliary verbs, and sends them to the Kanji word extraction means X1. The next word is sent to the Kanji word extraction means 4.

ga (particle) に (particle) と (particle) say (auxiliary verb) wa (particle) ri (adverb) anyone (pronoun) also (particle) ga (particle) te (particle) いる (auxiliary verb) (noun) de (auxiliary verb) aru (auxiliary verb) deshiku conjunction) this (adjunctive) kara (particle) toku particle) say (auxiliary verb) of (particle) wo (particle) thing (formal noun) haku particle> never ( Adverb) は (particle) No (auxiliary verb) Subsequently, the kanji word extraction means 4 extracts words that include kanji in their notation from among these words. As a result, the following words are extracted and sent to the word presentation means 5.

to say (auxiliary verb) to change (adverb) who (pronoun) thing (formal noun) but conjunction) say (auxiliary verb) thing (formal noun) never (adverb) The word presentation means 5 receives the information sent from the kanji word extraction means 4. Present these words that have been taught to you in a way that is different from other words.

For example, FIG. 2 shows an example of display with an underline added on a CRT display.

The text inspection device of the present invention can be implemented not only as a stand-alone device as described in the above embodiments, but also in a form incorporated into a word processor or the like. At that time, the text input means 1, text storage means 2, etc. can be shared with a word processor.

The previously mentioned ``Text Abnormality Inspection Device'' (Unexamined Japanese Patent Publication No. 59-1
Publication No. 27138), “Automatic abnormality inspection and correction device for text”
By combining it with a Japanese word processor J (Japanese Patent Application Laid-Open No. 59-127140) or Japanese Word Processor J (Japanese Patent Application Laid-open No. 59-94132), the text inspection function can be enhanced.

(Effects of the Invention) As described above, according to the present invention, locations where it would be preferable to change the kanji notation to the kana notation are automatically detected in consideration of the part of speech of the word. Therefore, by reducing the content of kanji characters, the efficiency of proofreading work for making texts easier to read can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, and FIG.
The figure shows an example of the display by the word presentation means 5, and FIG. 3 is a block diagram showing an example of the configuration of the specific part of speech extraction means 3. In the figure, 1... text input means 2... text storage means 3... specific part of speech extraction means 3 4... kanji word extraction means 4 5... word presentation means. 7. -1
,・A7:τ゛0, Agent Patent Attorney Susumu Uchiatsu,・−H−

Claims (1)

[Claims] A text inspection device that automatically detects a portion to be proofread from an input Japanese sentence, comprising: a specific part of speech extraction means for extracting a word of a specific part of speech from the input Japanese text; and a specific part of speech extraction means. Kanji word extraction means for extracting words whose notation includes Kanji characters from among the words extracted by the means; and word presentation means for presenting the words extracted by the Kanji word extraction means in a different display method from other words. A text inspection device characterized by: