JPH08263478A - Single/linked chinese character document converting device - Google Patents

Abstract

PURPOSE: To provide the device for exactly converting a document written in characters with reduced number of strokes (simplified characters) and a written in characters with full numbers of stroked (unabridged characters) document in Chinese. CONSTITUTION: In addition to information showing the character and reading of Chinese characters (KANJI), a word conversion dictionary is referred to. Specifically, a word converting part 30 records a source document and a language flag inputted from an input part 10 into a buffer 40 and refers to the language flag and a word conversion dictionary 20 and when there is any KANJI or word matched with the retrieved word of the word conversion dictionary 20, this KANJI or word is reloaded into a correspondent word. A character/reading converting part 60 refers to the language flag recorded in the buffer 40 and a system dictionary 50 (including plosive character dictionary) and converts the character of the source document to a reading symbol. A reading/ character converting part 70 refers to the language flag, system dictionary 50 and a simplified/unabridged character correspondence dictionary 80 and converts the reading symbol of the source document to the character of a target document according to the algorithm of conversion from reading to character, and the result is outputted from an output part 90.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for converting a simplified Chinese document and a traditional Chinese document.

[0002]

2. Description of the Related Art In recent years, exchanges between the Chinese mainland and Taiwan have become frequent, and along with this, the exchange of documents between them has also increased. However, since there has been no interaction for 40 years, not only the fonts of the Kanji used by both parties, but also the words and terms (single or multiple words that are often used in everyday life and their usages) have become quite different. There is. This makes it difficult to understand each other and the documents used on the other. For example, in Taiwan, a laser printer in Japan is called (a term used for) a "lightning machine," but in the continent it is called a "glow light stamping machine." Therefore, the demand for mutual conversion between simplified Chinese documents used in the continent and traditional Chinese documents used in Taiwan is rapidly increasing. However,
The following technical and linguistic difficulties must be solved for the mutual conversion between simplified and traditional documents.

(1) The number of common characters of simplified characters is about 8,000
However, the number of traditional Chinese characters used is much more than that of Simplified Chinese characters. For example, even in traditional Chinese on a computer 13,
There are 053 pieces. Therefore, some simplified characters may correspond to one simplified character. As an example, the simplified Chinese character “after” corresponds to the traditional Chinese characters “after” and “after”. Therefore, when converting a simplified document into a traditional document, it is necessary to select an appropriate traditional character.

(2) Due to differences in society and culture, words and their usages are also different. For example, the "level (technical level)" used to express the degree of quality in Japan is
It is called "horizontal" in mainland China, but "level" in Taiwan. For this reason, a conversion device for documents written in simplified characters and documents written in traditional characters has been studied and developed.

An example of a conventional Simplified Chinese document conversion device is, for example, pages 187 to 201 of a collection of papers by the 7th computer linguistics study group of the 1994 Republic of China.
Page to "A Text Conversion System Between Simp
lified and Complex Chinese Characters Based on OCR
Approaches ". This paper describes an example of a device for converting a simplified document into a traditional document. Fig. 1 shows this configuration.
In the figure, reference numeral 100 denotes a source document input means capable of inputting a simplified document or a traditional document as a source document. Reference numeral 300 is a character usage frequency table that stores the usage frequency of each character. Reference numeral 350 is a feature database that stores the feature value of each character. Reference numeral 200 denotes a character cutout unit that cuts out a character from the image data. Reference numeral 210 denotes a feature extraction unit that extracts and calculates the feature value of the character cut out from the image data. Reference numeral 220 is a collation unit that collates characters by referring to the character usage frequency table 300 and the feature database 350. Reference numeral 500 is a word conversion dictionary that stores words. Reference numeral 510 is a basic dictionary that stores general character information. Reference numeral 520 is a code correspondence table that stores character codes of simplified characters and traditional characters. Reference numeral 530 is a BIGRAM table that stores the connection frequencies of adjacent characters, which are obtained in advance by statistics or the like. Reference numeral 400 denotes a Chinese language module for converting candidate kanji or words into word lattices (word lattice, which refers to a lattice formed by each candidate kanji, word, etc. being processed). It has a modification module. Reference numeral 420 is a kanji or word conversion unit that finds an optimum conversion path.
Reference numeral 410 is a character correction unit that manually corrects the erroneously recognized character. An output unit 600 performs conversion and outputs the obtained target document.

Next, the procedure for converting simplified Chinese characters to traditional Chinese characters in the above-mentioned Chinese simplified Chinese character document conversion device will be described by taking the simplified Chinese character document shown in FIG. 2A as an example. When the simplified document shown in FIG. 2A is input from the source document input unit 100, an image as a figure is read into the OCR module by an OCR (optical character reading device). Then, after the image of each character is cut out by the character cutting unit 200, the characteristic extraction unit 210 calculates the characteristic value of each character. The collation unit 220 is used to refer to the character usage frequency table 300 and the feature database 350 to detect candidate simplified characters. Then, the operation of the character conversion and post-processing module is started. First, with reference to the code correspondence table 520 in the Chinese language module 400, each corresponding candidate character as shown in FIG. The word conversion dictionary 500 and the basic dictionary 510 are searched using the combination of the target Kanji of each extracted candidate as a search key,
Candidate words as shown in FIG. 2C are extracted. The extracted candidate words are arranged like the word lattice shown in FIG. The word conversion unit 420 refers to the BIGRAM table 530 (the frequency of use of two adjacent characters or words based on the corpus), and statistically bigram Markov Lang
An optimal or most probable conversion path as shown in FIG. 2E is extracted from the word lattice by the uage Model (Markov language model based on statistical BIGRAM), and then output by the output unit 600.

Next, a method of extracting the optimum conversion path will be briefly described. BIGRAM statistical data (P
(C _i | C _i−1 ) and P (for C _i | S _i )) are used to find a path that can maximize the value of the function as follows. MaxP (C | S) = P (C 1, C 2 .... C n | S 1,
S ₂ .... S _n ) → Sum. P (C _i | C _i-1 ) (C _i |
S _i ) (i = 1, ..., N) S: A candidate kanji set, for example, the first reading candidate set is S1.

C: Set of elements of candidate set S, eg C ₁
Represents the first candidate kanji for S ₁ . _{P (C i | C i-} 1): on certain letter is C _i, representing the probability of occurrence of BIGRAM when the previous character of the character is a C _i-1. In the same way, apply the above equation to the following equation. → Sum. P (W _i | W _i-1 ) P (C _i | S _i ) (i =
1, ..., N) P (W _i | W _i-1 ): represents the appearance probability of BIGRAM when a word is W _i and the preceding character of this word is W _i-1 .

However, the above contents are (EJYannakouda
kis and PJ Hutton, "n-Gramsand their impliment to
natural language understanding ", Pattern Recognitio
n, Vol.23.no.5, pp.509 to 528, 1990, and the algorithm of Markov Process is described in, for example, Hyundai Mathematics Co., Ltd., 1981, "Introduction O |
All of them are well-known techniques as described on page 96 of "R seminar". Therefore, further explanation is omitted.

The flow of the above procedure is shown in FIG.

[0011]

However, the above conventional techniques have the following problems. (1) For example, a simplified Chinese character "us" is often mistakenly recognized as "dry", but a character that is incorrectly recognized by OCR is difficult to correct by post-processing. (2) There is a certain limit to the accuracy of conversion because only the character information is used and effective processing is not performed for Chinese phonetic characters. It should be noted that the word "Hyakuto" refers to a Kanji that has multiple readings. To give a specific example, taking "another dry and dry type work" of a simplified document as an example, "dry" is "gan".
There are readings of "1" and "gan4", and therefore the traditional characters "dry" and "dry"("gan1") and "stem"("gan")
4)), the conventional technique remains as "drying" is "(gan1) (gan1)", so it is generally converted to "other dry and dry work", which is correct. I can't get the "other trunk dry work".

(3) Since a large and well-balanced Chinese compass cannot be easily obtained due to differences in copyright and the contents of the text to be created, it takes a lot of man-hours and a large amount of cost to extract BIGRAM. Need. Also, if the Chinese copas are imbalanced, the extracted BIGRAM table affects the accuracy of word conversion. The term "CORPUS" as used herein refers to a database of occurrence probabilities of adjacent characters and words.

Therefore, the advent of an inexpensive and accurate mutual conversion device for Simplified Chinese characters and Traditional Chinese characters is desired.
The present invention has been made for the purpose of solving such problems.

[0014]

In order to achieve the above object, in the invention of claim 1, the source document written in Simplified Chinese or Traditional Chinese in Chinese is used for the other by utilizing the information of characters and reading. In a device for converting a simplified Chinese character document into a target document described in characters, a simplified Chinese term (including single or plural Chinese characters, idioms, phrases, etc.) and a corresponding traditional Chinese term are combined (including: (Substantial set) a term conversion dictionary to be stored, a term conversion unit that searches the term conversion dictionary for terms in the source document written in simplified or traditional characters, finds a corresponding word, and rewrites the term in the source document, A system for storing phonetic symbols (including, phonetic symbols, syllables, and also plural) and corresponding Simplified and traditional Chinese characters (single or plural) or words (including, collocations, idioms, etc.) Calligraphy, a character-to-reading conversion unit for converting characters of a source document into reading symbols (including various glyph symbols) by referring to the system dictionary, and a simplified unit for storing Simplified characters and corresponding Traditional characters as a set. A character pair dictionary, the system dictionary, and a reading pair that converts the above-mentioned reading symbols into the characters of the target document of other typeface characters by a predetermined conversion algorithm from reading symbols to characters by referring to the simplified character corresponding dictionary. It is characterized by including a character conversion unit.

In the invention of claim 2, the simplified Chinese characters and the traditional Chinese characters are, for example, "00" for the former and "0" for the latter.
It is distinguished by a predetermined mark (language flag) indicating that, such as "1", and the system dictionary stores Chinese characters and words in non-breaking letters in non-breaking letters, regardless of whether they are simplified or traditional Chinese. There is a non-transliterated character section means and a transliterated character section means for storing a transliterated character in the transliterated character section (including distinction by both flags in the same memory), and the reading-to-character conversion section as a conversion algorithm, First, the longest match method that prioritizes the number of syllables (Chinese characters) to be configured and secondly prioritizes the syllable that exists first (including, in addition to this, other methods such as so-called morphological analysis is adopted. It has a longest matching method reflection conversion means.

In the invention of claim 3, the reading-to-character conversion unit has a frequency-of-use reflection conversion means for converting a character or a word that is frequently used as a conversion algorithm by giving priority to it (including other words). It has a means). In the invention of claim 4, the use frequency reflection conversion means may use, for example, for each field, the conversion algorithm or the frequency table used by the algorithm for conversion according to the content of the field such as a technical document or a literary book. It is characterized in that it has a document content reflection use frequency switching control means which can be switched by exchanging it with the above.

[0017]

With the above structure, in the invention of claim 1,
After the user inputs a source document and a language flag (including substantially the same instruction or input) indicating the simplified document or the traditional document by the OCR, the disk driver, etc., the term conversion unit determines whether the source document is the simplified (traditional) source document. In terms of terms, the term conversion dictionary that stores in advance the simplified Chinese terms and the corresponding Traditional Chinese terms is searched, the corresponding corresponding terms are found, and the terms of the source document are rewritten. The system dictionary pre-stores phonetic symbols and their corresponding simplified and traditional Chinese characters or words. The character-to-reading conversion unit converts the characters of the source document into reading symbols by referring to the system dictionary. The simplified-traditional dictionary stores sets of simplified characters and corresponding traditional characters in advance. The phonetic-to-character conversion unit converts the phonetic symbols into characters of the target document of traditional (simplified) characters by a predetermined conversion algorithm from the phonetic symbols to characters by referring to the system dictionary and the simplified character dictionary.

According to the second aspect of the present invention, the simplified and traditional Chinese characters of the Chinese traditional Chinese document converter are distinguished by the language flag. The system dictionary stores Chinese kanji and words, regardless of whether they are simple or traditional, by dividing non-breaking characters and breaks into non-breaking and breaks, respectively (including, arranging in alphabetical order , Separated by a separate flag). The phonetic-to-character conversion unit adopts the longest matching method as a conversion algorithm for conversion.

In the third aspect of the invention, the reading-to-character conversion section of the Chinese simplified Chinese document conversion device preferentially converts Kanji and words that are frequently used. In the invention of claim 4, the document content reflection use frequency switching control means in the use frequency reflection conversion means can switch the conversion algorithm or the frequency table used for the conversion according to the content of the source document. .

[0020]

EXAMPLES The present invention will be described below based on examples.
FIG. 4 is a block diagram of an embodiment of a Simplified Chinese character document conversion device according to the present invention. In the figure, reference numeral 10 is an input unit for inputting a source document and a language flag by a communication line OCR connected to the outside, a disk driver or the like. 20
Is a term conversion dictionary storing a correspondence table of term differences between mainland China and Taiwan as shown in FIG. Reference numeral 30 is a term conversion unit that refers to the language flag and the term conversion dictionary and rewrites the corresponding term of the source document that matches the search term into the corresponding term.
The processing flow of the term conversion unit 30 is shown in FIG. Note that this figure will be described later in detail. Reference numeral 50 is a system dictionary that stores each phonetic symbol and its corresponding simplified Chinese characters and words and traditional Chinese characters and words (including deaf letters). As shown in FIG. 7, the system dictionary 50 stores Chinese characters and words in simplified and traditional forms. Reference numeral 60 is a character-to-reading conversion unit that converts the characters of the source document into reading symbols by referring to the language flag and the system dictionary 50. The processing flow of the conversion unit 60 is shown in FIG. This figure will also be described in detail later. Reference numeral 80 denotes a simplified-traditional character dictionary that stores the correspondence between simplified characters and traditional characters. FIG. 9 shows a configuration of the simplified Chinese character dictionary 80. Reference numeral 70 is a reading-to-character conversion unit that converts the reading symbol of the source document into the character of the target document by referring to the language flag, the system dictionary 50, and the simplified Chinese character dictionary 80.
FIG. 10 shows a processing flow of the reading-to-character conversion unit 70. This figure will also be described in detail later. An output unit 90 outputs the target document obtained as a result of the conversion process. 40
Is a buffer for temporarily recording the processing result.

In addition to the above, a character correction unit and the like to be finally checked by the translator are provided, but since these are obvious and well-known techniques, illustration and description thereof will be omitted. The operation of this embodiment will be described below with reference to the flow chart. First, Fig. 6
The operation of the term conversion unit 30 shown in will be described. (A1) A language flag and a source document are input by the input unit 10 and recorded in the buffer 40.

(A2) With reference to the language flag recorded in the buffer 40, the search word and the replacement word corresponding thereto are taken out in order from the term conversion dictionary 20. (A3) It is determined whether all the search terms have been extracted. If the search word still exists, the process of (a4) is started, the characters of the source document are matched with the search word as a search key, and the process returns to (a2).

When there is no search word, the processing of the term conversion unit 30 is finished and the processing of the character-to-reading conversion unit 60 is started. Next, the operation of the character-to-phonetic conversion unit 60 shown in FIG. 8 will be described. (B1) Input the characters of the source document converted from the term conversion unit 30. (B2) Characters of the source document are cut out for each syllable by a delimiter (for example, a reading point or a punctuation mark in a sentence) and then recorded in the buffer 40.

(B3) Each syllable recorded in the buffer 40 is extracted into each syllable, and after referring to the system dictionary 50, non-fictional characters (characters without ambiguity take precedence) are converted into reading symbols. Record in buffer 40. (B4) The character with the diacritical character recorded in the buffer 40 is converted into the corresponding phonetic symbol by referring to the diacritical character section of the system dictionary 50.

(B5) With reference to the characters of the source document in the buffer 40, the reading symbol of each character in the buffer 40 is corrected based on the Chinese idiom. For example, the reading of "Ma" is "m
"a1", but the tone of the second "Ma" is 1
Light voice (the highest among the four voices) without reading (sounds are weak and light pronunciation because the syllable is continuous and loses its specific tone)
Since it is "ma0" that should be read in, correct the reading mark of the second "ma".

The conversion of character-to-reading is a well-known technique disclosed in, for example, Japanese Unexamined Patent Publication No. 4-238397, August 26, 1992, and therefore further description will be omitted. With the above, the processing of the character-to-reading conversion unit 60 ends, and the processing of the reading-to-character conversion unit 70 shown in FIG. 10 is started. The contents will be described below.

(C1) The phonetic symbol of the source document obtained by the character-to-phonetic conversion unit 60 is input. (C2) With reference to the system dictionary 50, all syllables that can possibly become syllables of the reading symbol are cut out.
Then, candidate kanji and words corresponding to each syllable are recorded in the buffer 40.

(C3) By referring to the characters of the source document recorded in the buffer 40, a character corresponding to each character is extracted from the simplified Chinese character dictionary 80 by using each character as a search key and recorded in the buffer 40. (C4) The kanji or word of each syllable candidate having a possibility of becoming a syllable is taken out from the buffer 40, the character corresponding to each reading symbol in the buffer 40 is referred to, and an unsuitable candidate,
In other words, delete Kanji and words that are unlikely to occur.

(C5) Using the longest match method, an appropriate character is selected from candidate kanji and words. With the above, the process of the reading-to-character conversion unit 70 ends. Next, the operation of the present embodiment will be described more specifically by taking a simplified Chinese document as a source document and converting it into a traditional document as a target document.

When a source document "another unused ruan3 (external 1) case" and a language flag "simplified" are input from the input unit 10, the term conversion unit 30 causes a term as shown in FIG. Refer to the conversion dictionary 20. "Ruan3 (1 outside)
Since there is a search word "case", replace "ruan3 (outer 1) case" in the source document with the corresponding word "ruan3 (outer 1) body",
The source document is rewritten into the “other unused body”, and the replaced source document and language flag are recorded in the buffer 40. Next, the character-to-phonetic conversion unit 60 uses the language flag recorded in the buffer 40 and the system dictionary 50 shown in FIG.
Convert this rewritten character into the reading symbol "talshang4uei4sh3yueng4ruan3ti3" as shown below.
Then, the processing of the reading-to-character conversion unit 70 as shown in FIG. 10 is started. In (c1) above, if you input the above phonetic symbols, (c2) cuts out the syllable and converts it into one kanji,
Extract syllables that may correspond to words or terms. The extracted syllables and candidate Chinese characters and words corresponding to them are as shown in FIG.

[0031]

[Outside 1]

In FIG. 10 (c3), referring to each character in the source document in the simplified character dictionary 80 and the buffer 40 shown in FIG. 7, it corresponds to reading each kanji as shown in FIG. Take out a character. Then, in (c4), by referring to the corresponding Chinese character or word extracted from the simplified Chinese character dictionary 80, by deleting an inappropriate candidate single Chinese character or word, the possibility as shown in FIG. 13 is obtained. You can get the kanji and words that are highly likely candidates.

In step (c5), the longest matching method is used to convert to "other unused soft shavings". Finally, the output unit 90 outputs the converted character string. Although the present invention has been described above based on the embodiments, the present invention is not limited to the above embodiments, and it is needless to say that the invention may be appropriately modified and implemented without departing from the scope of the invention. That is, for example, (1) the system dictionary does not store kanji and words in simplified and traditional forms, but stores them in the order of reading symbols by distinguishing simplified and traditional forms according to the symbol to that effect.

(2) The algorithm used for the conversion of the reading-to-character conversion section is not limited to the longest match method, but morphological analysis such as considering the number of unconfirmed characters is adopted, and other information such as grammar and frequency of use is also acquired. It is used to judge the accuracy of the conversion result. Specifically, frequency of use and OC
The candidate characters are narrowed down (specified) or excluded from the recognition probability of R.

(3) For the convenience of manufacturing or the like, a plurality of essential constituent elements (constituent elements, matters, parts) of the present invention may be provided in a plural number, or conversely, a plurality of them may be provided in a single element, or a combination thereof may be appropriately used. I am. (4) The same configuration as that of the present invention is obtained by loading a necessary program into an existing word processor or conversion device or adding a disk storing a dictionary.

(5) Characters that are frequently used are expensive, but are stored in a faster storage means to improve the embodiment. (6) The input unit is a unit that is read by an L / E type reader that is separate from the conversion device main body and is input (read) from a disc that stores bit information. Further, the output means outputs the document coded on the disc.

(7) The frequency with which words and characters are used varies considerably depending on the field of the document. For example, in documents related to animals, letters such as “moving”, “shellfish”, “horse”, “cat (simplified)”, and “bird” are frequently used, and in patent relations, words such as “kairan” are used. Frequent. Therefore, when the frequency of use is adopted or reflected as the conversion algorithm of the reading-to-character conversion unit, the purpose of the document input by the user of this device,
I try to use the frequency table created according to the content area. Alternatively, the frequency of use is determined based on the frequency of use of a specific word such as "kairei", and the frequency table automatically referred to thereafter is changed.

(8) When a sentence input by OCR is to be changed, it is judged that a Kanji character or the like not originally listed in the dictionary is an OCR reading error, and a function to output that effect is added. . (9) For characters that are known to have many similar characters, many strokes, and other low OCR reading accuracy based on experience, weight the word containing the character and the correlation with the preceding and following characters. Functions such as raising and alerting the translator to that effect are added.

[0039]

As described above, according to the Chinese Simplified Chinese document compatible device of the present invention, the conventional problems can be solved,
The following effects can be obtained. (1) Since the information of characters and readings is used, when converting the readings into characters, if the word is not in the dictionary, or there is a problem of conversion between two adjacent words (for example, "Yuichiten" The corresponding phonetic string has the words "beneficial" and "one heaven" before and after, so there is an erroneous conversion of "beneficial heaven") and the problem of the selection of homonyms (for example, "upper", "still"),
It is possible to effectively solve the problems caused by the "three homonyms" of "Captain".

Further, it is possible to reduce erroneous conversions of phonetic characters by using the reading information such as the grammar and grammar, and the accuracy rate of conversion between simplified and traditional characters is improved. Also, the size of the dictionary can be reduced. (2) Since the BIGRAM information is not used, the accuracy of conversion is improved without being affected by the contents of the copass. Also, there is no inaccuracy of the conversion due to the difficulty of collecting the corpus. In addition,
It is needless to say that if an accurate one suitable for each field can be obtained, the accuracy can be further improved by using these.

(3) As a Chinese word processor or the like, the same system dictionary can be used for converting syllabary symbols into Chinese characters when inputting Chinese documents and for converting between simplified and traditional characters. Therefore, it is not necessary to provide another dictionary when it is also used as a Chinese word processor. As a result, it is possible to reduce the number of man-hours required to create a dictionary, which leads to cost reduction.

(4) Since the term conversion dictionary is described in one language, as long as this dictionary is used, there is no need to switch the simplified system or the traditional system intricately for use in the continent and Taiwan, and it is economical. And labor-saving.
Therefore, the practicability of the present invention is very high.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a configuration of a conventional Chinese simplified Chinese document conversion device.

FIG. 2 is a diagram showing a hardware flow of a process of a conventional Chinese simplified document conversion device.

FIG. 3 is a flow chart of processing of a conventional Chinese simplified character document conversion device.

FIG. 4 is a block diagram of an embodiment of a simplified Chinese character document conversion device according to the present invention.

FIG. 5 is a conceptual diagram of a data structure of a term conversion dictionary in the above embodiment. In the figure, (a) is used for conversion from a simplified character document to a traditional character document. (B)
Is used to convert a traditional document into a simplified document.

FIG. 6 is a flowchart mainly showing the operation of the term conversion unit in the above embodiment.

FIG. 7 is a conceptual diagram of a data structure of a system dictionary in the above embodiment.

FIG. 8 is a flowchart mainly showing the operation of the character-to-reading conversion unit in the above embodiment.

FIG. 9 is a diagram conceptually showing a data structure of a simplified Chinese character dictionary in the above embodiment.

FIG. 10 is a flowchart mainly showing the operation of the reading-to-character conversion unit in the above embodiment.

FIG. 11 is a reading-to-character conversion unit in a specific example of conversion from a simplified character document to a traditional character document in the above embodiment.
It is the figure which showed the Chinese character and the word extracted as a result of the process of the step (c2) of.

FIG. 12 is a diagram showing characters corresponding to the reading of each Chinese character extracted as a result of the step (c3) of FIG.

FIG. 13 is a diagram showing characters corresponding to the reading of each Chinese character extracted as a result of the step (c4) of FIG. 10;

[Explanation of symbols]

 10 input unit 20 term conversion dictionary 30 term conversion unit 40 buffer 50 system dictionary 60 character-to-reading conversion unit 70 reading-to-character conversion unit 80 simplified Chinese character dictionary 90 output unit

Claims (4)

[Claims] 1. A simplified Chinese character document conversion device for converting a source document written in Simplified Chinese or traditional Chinese into a target document written in the other character using information on characters and readings. A term conversion dictionary that stores terms and their corresponding traditional Chinese words as a set, and a term in the original document that is written in Simplified or Traditional Chinese, searches the term conversion dictionary to find the corresponding word, and finds the corresponding original word A term conversion unit that rewrites, a system dictionary that stores reading symbols and corresponding Simplified and Traditional Chinese characters or words, and a character-to-reading conversion unit that refers to the system dictionary and converts the characters of the source document into reading symbols. , A simplified dictionary corresponding to the simplified Chinese characters and the traditional Chinese characters corresponding to the simplified Chinese characters, and the system dictionary, referring to the simplified Chinese characters corresponding dictionary, a predetermined symbol from the reading symbol to the character. A Chinese-Simplified Chinese character document conversion device, comprising: a reading-to-character conversion unit for converting the above-mentioned reading symbol into a character of a target document of another typeface character by a conversion algorithm. 2. Simplified Chinese characters and traditional Chinese characters are distinguished by a mark to that effect, and the system dictionary defines Chinese characters or words in Simplified or Traditional Chinese characters as non-breaking characters. The longest-matching method adopts the longest-matching method as the conversion algorithm. The Chinese simplified Chinese document conversion device according to claim 1, further comprising reflection conversion means. 3. The reading-to-character conversion unit has a usage frequency reflection conversion unit for converting a character or a word that is frequently used as a conversion algorithm with priority. 2. A simplified Chinese document conversion device according to 2. 4. The use frequency reflection conversion means has a document content reflection use frequency switching control means capable of switching the conversion algorithm or the frequency table used for the conversion according to the content of the source document. The Chinese-Simplified Chinese character document conversion device according to claim 3.