JPH05233701A - Dictionary storage device - Google Patents

Abstract

PURPOSE: To execute the high-speed access of a character by making the inner codes of two corresponding middle sentence characters index values to be a secondary coordinate place to occupy continuously storage space while shifting these through an indexing device. CONSTITUTION: In the indexing device 103, middle sentence indexed values to occupy one continuous space are generated by obtaining the inner code shift arithmetic operation of an inputted KANJI (Chines character). Then, the first character and the second character of one compound word of KANJI in the dictionary format (archaigue) of a magnetic disk 7 are read in from a reader 101, and are converted into the indexed values X, Y for these two characters through one indexing device 103, and these two indexed values X, Y are made the coordinates of an axis S and the axis Y. An element value corresponding to it shows the connection relation of these two characters (this element value is made one to plural bits). This element value is recorded and stored in a word and phrase connection relation storage device 104, and further, it is compressed through a corresponding device, and is stored in a compressing and storage device 105.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Chinese dictionary storage device or a method for compressing and storing a Chinese dictionary.

[0002]

2. Description of the Related Art A product for processing Chinese data using a computer usually needs to store a Chinese dictionary in memory to prepare for information (question). A Chinese sentence requires 2 bytes for one character, a compound word is composed of several Kanji characters, and the Chinese sentence dictionary contains tens of thousands of Kanji characters. It requires a large amount of storage space, is burdensome, and slows down the character calling speed.

[0003] The usual methods used traditionally are the binary interrogation method and the Hash method. Although the character search is faster when using the hash method, it requires a hash table in a large storage space, and when using the binary interrogation method, it is possible to find out after a number of interrogations. Moreover, the large dictionary cannot be stored in the RAM and must be stored in the hard disk in the form of stock, and reading the hard disk several times by the dichotomy takes a very long time. From a computer perspective, one generally has to choose between space and speed.

[0004]

DISCLOSURE OF THE INVENTION The present invention stores a dictionary in a special manner and stores it in a RAM after compressing it to a very small size. In addition, the interrogation speed of the post-compression dictionary can be increased to the hash level by the method of placing the "draw line" (leg lead wire) and the memory at the contrast position.

The operation of the present invention extends from a missing character in a Chinese sentence, a typographical error check (spell check) to a system for processing after the OCR of a Chinese sentence or a speech sound identification.

[0006]

[Means for Solving the Problems] The internal code of the input Chinese sentence
Do one shift operation through the indexing device
It becomes a value of Chinese indexing that occupies space continuously,
The letter A is the coordinate value of the X axis and the second letter B is the coordinate value of the Y axis.
To do. Whether or not these two characters form a idiom depends on the lexical connection function.
It is set every element as the element value of the storage device.
And the values are stored in the memory so that the values of the elements correspond.
Idioms are not composed of the combination of the first letter A and the second letter B (6
02), the number is recorded in the counter. So
Has a value of 2 ^-14If it reaches -1, the counter value is the element
A compressed word that is equal to the value and merges into compressed storage
First letter A, second letter written inside and corresponding to this element value
B is considered to be a two-letter word and part of a multiple lexical
Instead, the value of the counter and the element value are combined and stored in the compressed storage device.
Write in compressed word, the first letter A to which this element value corresponds
And the second character B is a two-letter word and part of a multi-letter word.
To do. Compressed storage device that captures the counter value and element value together
Write in one compressed word in storage, then top A and B
The index value after the second character that corresponds to
Entered in the text, the second and subsequent characters and the connection relationship between the characters
Is written in the next element value. In addition, the first letter A and the second letter B are necked
The second to nth tokens are defined as the second and subsequent letters.
The connection relationship of characters is also written in the next element value.

According to the present invention, each element value is placed in line with the position of the lexical connection relation storage device and the position in the compression storage device, and the value is about 1.
Two control positions for every 0000 element values are written in one storage device to achieve the purpose of high-speed character calling.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT FIG. 1 shows the structure of the present invention, and a reader 10
From 1, the magnetic disk dictionary format (archiving) 102
The first character A and the second character B of the middle-phrase word are read and converted into indexing values X and Y for these two characters through one of the indexing devices 103, and these two index values are the X and Y axis coordinates. And The element value corresponding to this is a memory of the connection relationship of these two characters (this element value is made from one to multiple bits). This element value is recorded and stored in the lexical connection relation storage device 104, further compressed via a compression device, and stored in the compression storage device 105.

Indexing device: The indexing device yields a Chinese sentence indexing value occupying one continuous space through an inner code shift operation of the input Chinese character. A Chinese character forms one character with two bytes, and the memory position range of the high-order byte in the BIG5 code is A4 to F9. 40 low order bytes
It is divided into two stages, -7E and A1-FE. Any Kanji index value X, Y (X is a high-order byte, Y is a low-order byte), and if the Y value is A1 to FE, Y = Y-2
Through the operation of 2, the low-order bytes become a continuous range.
FIG. 2A shows a storage space distribution diagram of original BIG-5 Chinese inner code. Y for Y value of A1 ≦ Y ≦ FE
= Y-22 and X = X-A4, Y = Y-40, the range of X is set to 0-55, the range of Y becomes 0-9C, and X ^* (Y length 9π) + Y After the calculation,
The index values X and Y of an arbitrary Chinese character are reflected in one value within the range of 0 to 13501. Fig. 2 (B) shows a structural diagram of the continuous space occupied by the Kanji after indexing. This indexing method also operates in other Chinese internal code systems, but is converted into continuous space through different shift conversion. Can be used.

Lexical connection relation storage device: The traditional Chinese dictionary (see FIG. 3 (A)) has a very large space occupation amount, and it is necessary to put it on the magnetic disk. The number of times of entry and exit increases, and the processing speed inevitably slows.

FIG. 3B is a structural diagram of a lexical connection relation storage device according to the present invention. In this embodiment, 1 is used as a 2-bit element value, and the connection relation is stored for the first two characters in the middle character phrase. If this middle character phrase is a compound word consisting of two characters or a single character, the second bit in the element value of FIG. 3B is set to "1". If the middle character phrase is three or more characters, the first bit in the element value of FIG. 3B is set to "1".

FIG. 4 shows a flow for forming a lexical connection relationship.

One medium character phrase 401 is read, and if it is judged as a single character 402, one indexing device (10 in FIG. 1) is read.
After 3), the index value 403 of this character is calculated to be the X-axis coordinate value, the Y-axis coordinate value is set to 5401, and the lexical connection relationship is the (X + 1) ^* 5402 ^* 2 + 1 position in the storage device. (That is, the element value of (X, 5401) is set to "1". If the read middle character phrase is not the single character 402, the first and second characters are passed through the indexing device (103 in FIG. 1). Calculate the index value 405 (calculate X and Y respectively), which is two letters and sets the element value at (X ^* 5402 ^* 2) + Y ^* 2 + 1 location to "1" 406. or ( ^* 5402 ^* 2)
+ Y ^* 2 Set the element value at the location to “1” 407. According to this method, all the Chinese characters can be written in the lexical connection relation storage device.

Compressor: A large space matrix is formed by element values indicating a lexical connection relationship, and occupies a relatively large memory space. Non-lexical element values can be compressed using a compression device to reduce memory space. The compression device used in this embodiment is

[0015]

[Equation 3]

A compression device.

Compressed storage device: The element value after compression is stored in the compressed storage device as 2 zones (16 bits) as one zone (sector), and the token after being compressed to 14 bits in the front part is stored. The number of element values that are not formed is loaded, and the element values themselves are loaded in the latter two bits to form a compressed word. FIG. 5A shows the configuration diagram. Among them, when the element value is '00', the Chinese character corresponding to this element value does not form a token in the table. If the element value is '01', the two Chinese characters to which the element value corresponds form exactly two words. Element value is '10'
Then, the two Chinese characters corresponding to this element value form three or more words. When the element value is '11', the two Chinese characters themselves corresponding to this element value in the table are two-letter words, or three or more letters.

When the last 2 bits (that is, the element value) in one word described above are "10" or "1", the corresponding first letter (A) and second letter (B) are displayed, and three letters are displayed. The above words can be formed. At this time, the index value of the third character is loaded into the previous 13 bits of the next 1 word, and the character relationship between the third character and the subsequent characters is loaded with the next 3 bits of the next element value.

FIG. 5B shows a memory system of a phrase having two or more characters in the compression storage device. When the second bit in the element value is set to "1", the character is a medium phrase. Indicating that it is the last character, the next one word loads the third character of the middle phrase beginning with another index value (X, Y). When the third bit in the next element value is set to "1", it means that the following one word and this word belong to the same middle character phrase, and the next element value (element)
The middle first bit is reserved and is not used in this embodiment.

FIG. 6 shows the process in which the compressed and memorized element values reach the compressed storage.

1-element value 60 in lexical connection relation storage device
1 is read, and the first character A and the second character B corresponding to this element value
When it is determined 602 that the phrase is not formed, if the value of the measurement counter has already reached 2 ¹⁴ -1 (605 in FIG. 7A), the counter value is shifted by 2 bits and linked to the element value. , Is also written in one compressed word 606 in the compressed storage device, and the counter returns 607 to “0”. If it is determined that the first letter A and the second letter B corresponding to this element value are two letters and are not part of the plural tokens 603, the value of the counter is shifted by 2 bits and the compression value is stored together with the element value. It is written to one of the compressed words 608 and the counter becomes 0 (609).

The first character A and the second character B to which this element value corresponds
Is a two-character lexical part and it is judged that it is a part 604 of the plural lexical parts, the counter value is shifted by 2 bits and written together with the element value into one compressed word 610 in the compressed storage device, and at the same time, the counter returns to 0. (611). The index value of the second and subsequent characters of the lexical beginning with A and B is shifted by 3 bits and written in 612 in the next one compressed word, and the second and subsequent characters and the connection relationship between the characters also continue. 6 out of element values
Write in 13. FIG. 8 shows a memory configuration diagram of words of two or more characters in the compressed storage device.

Lexical calling device: contrast position storage device To achieve the purpose of calling and calling characters by searching at high speed (to prevent decompression from the front row of the compression storage device, Hold until searched),
Each element value is a combination of a position in the lexical connection relation storage device and a position in the compression storage device. The present invention writes two types of reference positions into one memory device for every 10,000 element values.

FIG. 9 shows the control position setup and memory.

An element value position index, a compression position index, and an initial value 801 of a lead wire point value are set, and one integer is read from the compression storage device (802). This number represents the number of element values which do not form a token, and when cumulatively added to reach the element value position index 803, it increases to the compression position index 804. If the element value after the integer represents two words and there are two or more characters, the integer after that is read and the next element value is "00".
It continues until it becomes 0 ″ (805). Next, it is measured whether the position index of the element value is larger than the point value 806 of the lead wire. Writing during 807, and when the accumulation of the lead wire points reaches 1000 (808), 1 is added to the position index of the element value (809), and other integers are completely processed.

Process of lexical search (interrogation): (AB [CD
...]) is an idiom 901, and the index value (X, Y) of (AB) [Note: if it is a single character, the index value is (X, 54).
01)] is calculated. The position (LOC) (903) of the element value of (X, Y) is calculated to search for the element of the [element value position / 10,000] item in the reference position storage device, and one lead wire closest to the LOC To get That is, it is an element value index and an index of the compression position reflected on it. 90
4 starts reading one integer 905 in order from the position of the compression position index, and adds up until the accumulation reaches the “lexical connection table” index 906. LOC> element value position index 907
At the time of, it displays the middle character phrase that reflects relative to the element value to be searched from the one that is not decompressed,
Continue decompressing. The element value of this integer is "10"
Or "11" is measured, the integer of the compression storage device is continuously read, and the element value next to the end number of the integer is "000" (90
Read until 8) is displayed. Next, the position index of the element value increases by 1, and the compression of the dictionary is continuously released (909). LO
If the position index is C <element value, it indicates that the characters (A, B) do not exist, that is, there is no single word (AB [CD ...]) (910). If LOC = index,
(A, B) It is displayed that the position of the lexical is searched for. In that case, if there are only two letters (A, B), it is measured whether the integer element value corresponding to it is "01" or "11". If (A, B) does not indicate the presence of a lexical (A,
It indicates that the phrase B) does not exist. In the challenge of words with three or more letters (ABCD ...), the integer is "1".
It measures 0 "or" 11 "and continuously compares the input characters after the second character with the Chinese characters that follow in the compression storage device, and continues until the comparison succeeds. Continuing the comparison until the element of becomes "000" (911).

[Brief description of drawings]

FIG. 1 is a configuration diagram of the present invention.

2A and 2B are a continuous storage space decomposition diagram of a Chinese inner code and a continuous storage space diagram of an indexed Chinese inner code.

3A and 3B are diagrams illustrating a traditional dictionary storage format and a memory system for lexical connection in the present invention.

FIG. 4 is a flowchart illustrating a lexical connection establishment and memory method.

5A and 5B are diagrams for explaining an element value memory system in a compression storage device and a compound word memory system of two or more characters in the compression storage device.

FIG. 6 is a flow chart illustrating a method of compression and reaching from a memory element value to a compression dictionary storage device.

7 (A), (B), and (C) are diagrams for explaining a method of similarly performing compression and a memory element value to reach a compression dictionary storage device.

FIG. 8 is a memory exploded view for a phrase having two or more characters in a compressed storage device.

FIG. 9 is a flowchart illustrating a memory method of a reference position.

FIG. 10 is a flowchart illustrating a character calling method.

[Explanation of symbols]

 101 Reading Device 103 Indexing Device 104 Lexical Connection Relation Storage Device 105 Compressed Storage Device

Claims (10)

[Claims] 1. An inner code of two corresponding Chinese characters, which stores a connection relationship between Chinese characters and can display an element value of 1 to multiple bits in the connection relationship between the above characters. The dictionary storage device is a binary index value that continuously occupies the storage space while shifting through the indexing device. 2. A one-character connection relation storage device for storing element values of 1 to a plurality of bits, that is, a connection relation between Chinese characters and a character, and A compression dictionary storage device comprising: a compression device that compresses those that do not form the above-mentioned element value middle-word, and a compression storage device that stores the number of element values that do not form the above-mentioned token, and two or more characters. .. 3. The compression dictionary storage device according to claim 2, wherein the compression dictionary storage device includes one indexing device, calculates an inner code of Chinese characters into one, and continuously converts a memory space of the characters. 4. The two-character memory of the compressed storage device comprises:
3. The compression dictionary storage device according to claim 2, wherein the element value of the connection relation is stored in a memory. 5. A memory for a plurality of characters in the compressed storage device stores an index value of the second and subsequent characters, and displays the next element value of the connection relationship between the characters of the plurality of characters. The compression dictionary storage device according to claim 2, wherein 6. The element value is 2 bits, according to claim 1 or 2.
The described compressed dictionary storage device. 7. The compression dictionary storage device according to claim 1, wherein the secondary element value is 3 bits. 8. A single lexical phrase is read, the first and second letters of the lexical phrase are taken, an inner code index value of the above-mentioned two characters is calculated, and a search is performed by a reference position device, and the value of the compression position and the lexical connection position are calculated. The value of is obtained, starting from the location of the value of the above-mentioned position in the above-mentioned compressed storage device, searching for the closest element value,
3. The compression dictionary storage device according to claim 2, further comprising one character search method comprising the steps of obtaining a connection relation between two characters, reading the next element value in order, and obtaining a connection relation of three or more characters. 9. A connection relation in which a first character and a second character are input by inputting a lexical character and switching the inner code of the first character and the second character to convert into an indexed value of a continuous space. Set the element value and compress the connection related element value of the above character, that is,
Compress the part that does not form a lexical record, record the number, the number of those that do not form the lexical, the character index after the second character and the next element value that indicates the continuous relationship between the characters with the next element value. A method of compressing and storing a Chinese dictionary that consists of each step of memory. 10. The equation 2 The method of compressing and storing a medium language dictionary according to claim 9, wherein a method of operating coding is adopted.