JP2005135086A - Dictionary data compression apparatus, electronic dictionary device, compressed dictionary data production method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To increase the compression efficiency of a dictionary coding method by discretely selecting entry unit data to create a reference part and changing the coding method in dependence on a character string. <P>SOLUTION: A CPU 10 discretely selects entry unit data from original English-Japanese dictionary data 200 as a processed English-Japanese reference part 302, and handles the remaining entry unit data as main data part intermediate data 306. The CPU 10, for character strings of a predetermined word length or longer, executes a global search program 214 to decide character strings to be coded, and for the other character strings, executes a local search program 216 to decide character strings to be coded. A coding program 218 is then executed to code the main data part intermediate data 306 into a coded English-Japanese main data part 304. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a dictionary data compression device that compresses dictionary data, an electronic dictionary device that expands and uses compressed dictionary data, and a compressed dictionary data manufacturing method and program for manufacturing compressed dictionary data.

  Various data compression algorithms are known, but compression of dictionary data used in an electronic dictionary device or the like is “random access (decoding / decoding for each headword) compared with compression of general sentence data”. There is a great difference in two points: “requires expansion” and “encoding (also referred to as compression) is performed only once during product development, so there is no problem even if sufficient time is spent”.

  “Random access for each headword” means decoding / decompression of dictionary data in units of headwords. The dictionary data has the same contents as a paper dictionary. The dictionary data before encoding is generally a series of text data in which characters printed on a paper dictionary are character codes. That is, the text data is a series of characters (text) printed in a paper dictionary in order from the first headword. If this dictionary data is simply compressed as a whole, information of any headword (descriptive text related to the headword) cannot be decoded. For this reason, it is necessary to divide the continuous text data into headwords (headword units) and compress them.

  “Encoding is performed only once at the time of product development” means that the manufacturer can literally encode dictionary data over a sufficient time. That is, the electronic dictionary device only decompresses the compressed dictionary data, and does not compress (encode) it. Therefore, there is an advantage that the most efficient compression method can be adopted after trying all possibilities by a high-speed computer.

  Among such features, for example, dictionary data compression methods such as Patent Document 1, Patent Document 2, and Patent Document 3 have been proposed.

  For example, the entire dictionary data is divided into about 10% part (hereinafter referred to as “reference part” as appropriate) and the remaining part (hereinafter referred to as “main data part” as appropriate). Then, it is determined whether or not the character string in the main data part is included in the reference part, and if it is included, encoding is performed based on the position and word length in the reference part included. A method for compressing the main data portion is known.

  Specifically, the character string to be encoded in the main data part is determined and encoded at which position in the reference part. A method of encoding with reference to a reference portion including such a character string is known as a dictionary-type encoding method. There are LZ77 method, LZ78 method, etc. in the dictionary type encoding method, but the feature of applying the dictionary type encoding method to the dictionary data is to enable the “random access for each headword”, the reference part It may be fixed.

If there is no matching character string in the reference portion, the character string to be encoded is compressed by directly encoding it with a variable length using a Huffman code or the like. As described above, two types of encoding are used: a dictionary-type encoding method that encodes by referring to the position of the reference portion that includes the character string, and a method that encodes directly using a variable-length code such as a Huffman code. To do.
JP-A-6-251070 JP-A-8-314960 JP 11-96186

  Here, of the two compression methods described above, the method of directly encoding with a variable length does not use the reference part, and it is necessary to uniquely assign a code to all the used character strings. The compression rate (also referred to as compression efficiency) is worse than the type coding method. Therefore, in order to encode a larger number of character strings using the dictionary-type encoding method, it is conceivable to increase the ratio of the reference portion in the dictionary data.

  However, when the ratio occupied by the reference portion is increased, the address indicating the position of the character string in the reference portion is increased according to the size of the reference portion. Therefore, when the information amount (code) representing the address is increased, the information amount of the encoded main data portion is increased, and the compression efficiency of the entire dictionary data is deteriorated.

  Also, the compression efficiency is better as the character string matches the longer character string. Furthermore, in the case of a short character string, there are cases where compression efficiency is better when directly encoded with a variable length and compressed than when encoded and compressed by a dictionary-type encoding method. However, conventionally, since it has only been determined from the head of the main data portion whether or not it sequentially matches the reference portion, the compressed character string cannot always be said to have good compression efficiency.

  Further, conventionally, a predetermined ratio of the entire dictionary data is divided from the head and used as a reference portion. However, the dictionary data is usually arranged and saved in the order of abc or aiueo. Therefore, when divided from the head according to a predetermined ratio, the character strings included in the reference part often overlap, and the character string matching the main data part is not efficiently included.

  The present invention has been made in view of the above problems, and an object of the present invention is to create a reference section by discretely selecting headword unit data in a dictionary-type encoding method, and according to a character string. Changing the encoding method to increase the compression efficiency. Furthermore, when the character string of the main data portion appears frequently in the reference portion, the object is to increase the compression efficiency of the dictionary data by representing the position of the character string in the reference portion with a code.

In order to solve the above problems, the dictionary data compression device according to the first aspect of the present invention provides:
Classifying means (for example, hard disk 20 in FIG. 2; original English-Japanese dictionary data 200) for classifying dictionary data in which character strings are described in series in headword units, into reference parts and data parts;
A position memory that selects a plurality of positions of characters that appear at a predetermined frequency from the reference section as the frequent reference positions based on the appearance frequency of the characters, and stores a plurality of positions in association with codes for identifying each frequent reference position Means (eg, RAM 30 in FIG. 2; sub-table 314);
When the reference position is the frequent reference position, the code of the reference position and the word length from the reference position stored in the position storage means are used. When the reference position is not the frequent reference position, the reference position and Main data portion encoding means (for example, CPU 10 in FIG. 2; step in FIG. 5) that encodes the main data portion so that it can be decoded in units of headwords by a dictionary-type encoding method using the word length from the reference position. A18)
It is characterized by providing.

The compression dictionary data manufacturing method of the invention according to claim 6 is:
On the computer,
A classification step (for example, hard disk 20 in FIG. 2; original English-Japanese dictionary data 200) for classifying dictionary data in which character strings are described in series in headword units into a reference portion and a main data portion;
A position memory for selecting a plurality of positions of characters appearing at a predetermined frequency as the frequent reference positions from the reference portion based on the appearance frequency of the characters, and storing a plurality of positions in association with codes for identifying each frequent reference position A process (eg, RAM 30 in FIG. 2; sub-table 314);
When the reference position is the frequent reference position, the code of the reference position stored in the position storage step and the word length from the reference position are used. When the reference position is not the frequent reference position, the reference position and the reference position A main data portion output step (for example, CPU 10 in FIG. 2; FIG. 5) that outputs the main data portion in such a way that it can be decoded in units of headwords by the dictionary-type encoding method using the word length from the reference position. Step A18)
It is characterized by including.

The program of the invention according to claim 10 is:
On the computer,
A classification function (for example, hard disk 20 in FIG. 2; original English-Japanese dictionary data 200) for classifying dictionary data in which character strings are described in series in headword units into a reference part and a main data part;
A position memory for selecting a plurality of positions of characters appearing at a predetermined frequency as the frequent reference positions from the reference portion based on the appearance frequency of the characters, and storing a plurality of positions in association with codes for identifying each frequent reference position Functions (eg, RAM 30 in FIG. 2; sub-table 314);
When the reference position is the frequent reference position, the code of the reference position and the word length from the reference position stored in the position storage function are used. When the reference position is not the frequent reference position, the reference position and Main data part encoding function (for example, CPU 10 in FIG. 2; step in FIG. 5) that encodes the main data part so that it can be decoded in units of headwords by a dictionary-type encoding method using the word length from the reference position. A18)
It is characterized by realizing.

  According to the invention described in claim 1, 6 or 10, when the character string to be referred to frequently appears when the main data portion is encoded by the dictionary type encoding method, the position of the frequently appearing character string is determined. It can be represented by a code. Therefore, it is possible to encode with a smaller amount of information than in the case of direct encoding using the position of the character string.

The dictionary data compression device according to the second aspect of the present invention provides:
A plurality of data in units of headwords are discretely extracted from the entire dictionary data in which character strings are described in series in units of headwords, and a reference unit (for example, RAM 30 in FIG. 2; post-processing English-Japanese reference unit 302). And a sorting means (for example, CPU 10 in FIG. 2; step A10 in FIG. 5) for classifying the remaining portion as a main data portion (for example, RAM 30 in FIG. 2; main data portion intermediate data 306);
Main data portion encoding means for encoding the main data portion so that it can be decoded in units of headwords by a dictionary-type encoding method using the reference portion as a reference source (for example, CPU 10 in FIG. 2; step A18 in FIG. 5). When,
It is characterized by providing.

Further, the compression dictionary data manufacturing method of the invention according to claim 7 is provided in a computer.
A classification process in which multiple pieces of data in terms of terms are discretely extracted from the entire dictionary data in which character strings are described in series in terms of terms, and used as a reference part, and the remaining part is classified as a main data part (For example, CPU 10 in FIG. 2; Step A10 in FIG. 5);
A main data part output step (for example, the CPU 10 in FIG. 2; FIG. 5) for encoding and outputting the main data part so as to be decodable in units of headwords by a dictionary-type encoding method using the reference part as a reference source. Step A18)
It is characterized by including.

The program according to claim 11 is:
On the computer,
A classification function that separates the data of the headword unit from the entire dictionary data in which character strings are described in series in the unit of headword, and uses it as a reference part and the rest as the main data part (For example, CPU 10 in FIG. 2; Step A10 in FIG. 5);
Main data portion encoding function for encoding the main data portion so that it can be decoded in units of headwords by a dictionary-type encoding method using the reference portion as a reference source (for example, CPU 10 in FIG. 2; step A18 in FIG. 5) When,
It is characterized by realizing.

  According to the second, seventh, or eleventh aspect of the present invention, it is possible to discretely extract headword unit data from the entire dictionary data and use it as a reference unit. Therefore, it is possible to create the reference portion without being biased toward a part of the dictionary data.

The dictionary data compression device according to the invention of claim 3
Dictionary data in which character strings are described in a series of headwords is divided into a reference part and a main data part, and a matching character string that matches the character string in the main data part exists in the reference part In this case, in a dictionary data compression apparatus that compresses dictionary data by a dictionary-type encoding method that encodes a character string of the main data portion with a code representing a copy of the matching character string,
More than a predetermined word length compression means (for example, CPU 10 in FIG. 2; step A14 in FIG. 5) for compressing the character string in the main data portion by using the dictionary type encoding method for a character string having a predetermined word length or more. When,
Storage means (for example, the hard disk 20 of FIG. 2; encoding efficiency prediction value table 206) that stores a plurality of character strings and the encoding efficiency prediction values of the character strings in association with each other;
Among the character strings included in the main data portion after being compressed by the compression means over the predetermined word length, the dictionary type encoding is performed in order from the character string having the highest encoding efficiency prediction value stored in the storage means. Evaluation value order compression means (for example, CPU 10 in FIG. 2; step A16 in FIG. 5) for compressing by the method;
It is characterized by providing.

A compression dictionary data manufacturing method according to an eighth aspect of the present invention provides a computer,
Dictionary data in which character strings are described in a series of headwords is divided into a reference part and a main data part, and a matching character string that matches the character string in the main data part exists in the reference part In this case, in a compressed dictionary data manufacturing method for compressing dictionary data by a dictionary-type encoding method for encoding a character string of the main data portion by a code representing a copy of the matched character string,
In the computer,
More than a predetermined word length compression process (for example, CPU 10 in FIG. 2; step A14 in FIG. 5) for compressing the character string in the main data portion by using the dictionary-type encoding method for a character string having a predetermined word length or more. When,
Among the character strings included in the main data portion after being compressed by the compression step with a predetermined word length or more, the evaluation value order is compressed by the dictionary encoding method in order from the character string having the highest encoding efficiency prediction value. A compression step (for example, CPU 10 in FIG. 2; step A16 in FIG. 5);
An output step (for example, CPU 10 in FIG. 2; step A18 in FIG. 5) for outputting the compressed character string data as compressed character string data in the evaluation value order compression process;
It is characterized by including.

The program of the invention according to claim 12 is
Dictionary data in which character strings are described in a series of headwords is divided into a reference part and a main data part, and a matching character string that matches the character string in the main data part exists in the reference part A computer that compresses dictionary data by a dictionary-type encoding method that encodes a character string of the main data portion with a code representing a copy of the matching character string,
More than a predetermined word length compression function (for example, CPU 10 in FIG. 2; step A14 in FIG. 5) that compresses the character string in the main data portion by using the dictionary-type encoding method for a character string having a predetermined word length or more. When,
A storage function (for example, the hard disk 20 of FIG. 2; encoding efficiency prediction value table 206) that stores a plurality of character strings and the encoding efficiency prediction values of the character strings in association with each other;
Among the character strings included in the main data portion after being compressed by the compression function with a predetermined word length or longer, the dictionary type encoding is performed in order from the character string having the highest encoding efficiency prediction value stored in the storage function. Evaluation value order compression function (for example, CPU 10 in FIG. 2; step A16 in FIG. 5) for compressing by the method;
It is characterized by realizing.

  According to the invention of claim 3, 8 or 12, when compressing a character string longer than a predetermined word length, compressing by a dictionary type encoding method and compressing a character string not longer than a predetermined word length Can be compressed by a dictionary-type encoding method from a character string having a high encoding efficiency prediction value. Therefore, efficient compression can be performed according to the character string.

The dictionary data compression apparatus according to claim 4 is provided.
Storage means (for example, the hard disk 20 in FIG. 2) for storing dictionary data composed of character string data;
Selecting means (for example, CPU 10 in FIG. 2; step B10 in FIG. 7) for selecting any one-byte character code from the standardized one-byte character codes;
A two-byte character code conversion table is created in which the first byte is a blank one-byte character code for which no corresponding character is specified from the standardized one-byte character code. Conversion table creation means (for example, CPU 10 in FIG. 2; step B14 in FIG. 7) for setting the selected 1-byte character code to an unset 2-byte character code;
Detecting means (for example, CPU 10 in FIG. 2; step B16 in FIG. 7) for detecting a 2-byte character code having an appearance frequency of a predetermined frequency from the character string data;
Corresponding character changing means for changing the corresponding character represented by the selected one-byte character code to the corresponding character represented by the detected two-byte character code (for example, the CPU 10 in FIG. 2; step B18 in FIG. 7). )When,
Of the character string data, first replacement means for replacing the selected 1-byte character code with the corresponding 2-byte character code set in the 2-byte character code conversion table (for example, CPU 10 in FIG. 2; Step B22) in FIG.
Second replacement means (for example, CPU 10 in FIG. 2; step B24 in FIG. 7) for replacing the detected 2-byte character code in the character string data with the selected 1-byte character code;
It is characterized by providing.

According to a ninth aspect of the present invention, there is provided a compressed dictionary data manufacturing method in a computer,
A compressed dictionary data production method for producing compressed dictionary data obtained by compressing dictionary data composed of character string data,
In the computer,
A selection step (for example, CPU 10 in FIG. 2; step B10 in FIG. 7) for selecting any one-byte character code from the standardized one-byte character codes;
A two-byte character code conversion table is created in which a blank one-byte character code for which no corresponding character is specified is specified as the first byte from the standardized one-byte character code. A conversion table creation step (for example, CPU 10 in FIG. 2; step B14 in FIG. 7) for setting the selected 1-byte character code to an unset 2-byte character code;
A detection step (for example, CPU 10 in FIG. 2; step B16 in FIG. 7) for detecting a 2-byte character code having an appearance frequency of a predetermined frequency from the character string data;
Corresponding character changing step for changing the corresponding character represented by the selected one-byte character code to the corresponding character represented by the detected two-byte character code. For example, the CPU 10 in FIG. 2; Step B18 in FIG. 7) When,
A first replacement step (for example, CPU 10 in FIG. 2) of replacing the selected one-byte character code in the character string data with a corresponding two-byte character code set in the two-byte character code conversion table. Step B22) in FIG.
A second replacement step (for example, CPU 10 in FIG. 2) that replaces the detected 2-byte character code with the selected single-byte character code in the character string data after the replacement in the first replacement step. Step B24) of FIG.
An output step (for example, CPU 10 in FIG. 2; step B24 in FIG. 7) for outputting the character string data after replacement in the second replacement step as compressed character string data;
It is characterized by including.

The program of the invention described in claim 13 is:
On the computer,
A storage function (for example, the hard disk 20 in FIG. 2) for storing dictionary data composed of character string data;
A selection function (for example, CPU 10 in FIG. 2; step B10 in FIG. 7) for selecting any one-byte character code from the standardized one-byte character codes;
A two-byte character code conversion table is created in which the first byte is a blank one-byte character code for which no corresponding character is specified from the standardized one-byte character code. A conversion table creation function (for example, CPU 10 in FIG. 2; step B14 in FIG. 7) for setting the selected 1-byte character code to an unset 2-byte character code;
A detection function (for example, CPU 10 in FIG. 2; step B16 in FIG. 7) for detecting a 2-byte character code having an appearance frequency of a predetermined frequency from the character string data;
Corresponding character change function for changing the corresponding character represented by the selected one-byte character code to the corresponding character represented by the detected two-byte character code (for example, CPU 10 in FIG. 2; step B18 in FIG. 7) )When,
A first replacement function (for example, CPU 10 in FIG. 2) that replaces the selected one-byte character code in the character string data with a corresponding two-byte character code set in the two-byte character code conversion table. Step B22) in FIG.
A second replacement function (for example, CPU 10 in FIG. 2; step B24 in FIG. 7) that replaces the detected 2-byte character code in the character string data with the selected 1-byte character code;
It is characterized by realizing.

  According to the invention described in claim 4, 9 or 13, it is possible to replace a 2-byte character having a predetermined appearance frequency as a 1-byte character and replace a 1-byte character as a 2-byte character. Therefore, by representing a 2-byte character having a high appearance frequency as a 1-byte character, dictionary data can be compressed with a small amount of information.

The electronic dictionary device of the invention according to claim 5 is:
Storage means for storing dictionary data consisting of character string data (for example, EEPROM 140 in FIG. 14; compressed English-Japanese dictionary data 1400);
Character code storage means (for example, EEPROM 140 in FIG. 14; character code conversion table 1408) for storing character codes and standardized character codes in association with each other;
Extraction means (for example, CPU 110 in FIG. 14; step G10 in FIG. 16) for extracting the first byte of the character code representing the character included in the character string data;
Based on the first byte extracted by this extraction means, character determination means for determining whether the character code is a 1-byte character code or a 2-byte character code (for example, CPU 110 in FIG. 14; step in FIG. 16) G14)
When the character determining unit determines that the character code is a one-byte character code, a standardized character code corresponding to the determined one-byte character code is read from the character code storage unit, and the standardized character is read. Replacement means for replacing with a code (for example, CPU 110 in FIG. 14; step G26 in FIG. 16);
It is characterized by providing.

The program of the invention according to claim 14 is:
On the computer,
A storage function for storing dictionary data composed of character string data (for example, EEPROM 140 in FIG. 14; English-Japanese dictionary data after compression 1400);
A character code storage function (for example, EEPROM 140; character code conversion table 1408 in FIG. 14) that stores character codes and standardized character codes in association with each other;
An extraction function (for example, CPU 110 in FIG. 14; step G10 in FIG. 16) for extracting the first byte of a character code representing a character included in the character string data;
Based on the first byte extracted by this extraction function, a character determination function for determining whether the character code is a 1-byte character code or a 2-byte character code (for example, CPU 110 in FIG. 14; step in FIG. 16) G14)
When the character determination function determines that the character code is a 1-byte character code, the standardized character code corresponding to the determined 1-byte character code is read from the character code storage function, A replacement function (for example, CPU 110 in FIG. 14; step G26 in FIG. 16) for replacing with a code;
It is characterized by realizing.

  According to the invention of claim 5 or 14, even if the character of the character string data is converted, the first byte of the character is extracted and converted to the original character according to the extracted character. It becomes possible to do.

  According to the invention described in claim 1, 6 or 10, when the character string to be referred to frequently appears when the main data portion is encoded by the dictionary type encoding method, the position of the frequently appearing character string is determined. It can be represented by a code. Therefore, it is possible to encode with a smaller amount of information than in the case of direct encoding using the position of the character string.

  According to the second, seventh, or eleventh aspect of the present invention, it is possible to discretely extract headword unit data from the entire dictionary data and use it as a reference unit. Therefore, it is possible to create the reference portion without being biased toward a part of the dictionary data.

  According to the invention of claim 3, 8 or 12, when compressing a character string longer than a predetermined word length, compressing by a dictionary type encoding method and compressing a character string not longer than a predetermined word length Can be compressed by a dictionary-type encoding method from a character string having a high encoding efficiency prediction value. Therefore, efficient compression can be performed according to the character string.

  According to the invention described in claim 4, 9 or 13, it is possible to replace a 2-byte character having a predetermined appearance frequency as a 1-byte character and replace a 1-byte character as a 2-byte character. Therefore, by representing a 2-byte character having a high appearance frequency as a 1-byte character, dictionary data can be compressed with a small amount of information.

  According to the invention of claim 5 or 14, even if the character of the character string data is converted, the first byte of the character is extracted and converted to the original character according to the extracted character. It becomes possible to do.

[1. overall structure]
FIG. 1 is an overview of a computer 1 and an electronic dictionary device 100 to which the present invention is applied. The computer 1 is usually installed in a manufacturer or the like of the electronic dictionary device 100, and is a kind of dictionary compression device that is used for compression of dictionary data. The dictionary data compressed by the computer 1 is stored in the EEPROM 107, and the electronic dictionary device 100 on which the EEPROM 107 is mounted is manufactured. In the electronic dictionary device 100, the compressed dictionary data is expanded and the contents of the dictionary data (such as headwords and explanation information) are displayed.

  Dictionary data is data consisting of headwords and explanatory information for explaining the headwords. For example, multiple types such as Japanese dictionary, English-Japanese dictionary, Japanese-English dictionary, English-English dictionary, Katakana dictionary, etc. There is. However, for the sake of simplicity, in the present embodiment, the dictionary data compressed by the computer 1 and stored in the electronic dictionary device 100 will be described as only dictionary data of an English-Japanese dictionary. In order to distinguish dictionary data before compression (encoding) from dictionary data after compression, the dictionary data before compression is hereinafter referred to as “original dictionary data”. The compressed dictionary data is referred to as “compressed dictionary data”.

  As shown in FIG. 1, the computer 1 includes hardware such as a general-purpose server computer including a display 3 such as a CRT (Cathode Ray Tube), a keyboard 5 and a memory 7 such as a RAM or a hard disk. The The electronic dictionary device 100 includes a display 103 such as an LCD (Liquid Crystal Display), various key groups 105 such as character input keys and dictionary type selection keys, and an EEPROM 107.

  The basic functions of the electronic dictionary device 100 are as follows. That is, when a dictionary is selected by a user and a character to be a search word is input (hereinafter, the input character is referred to as “input character”), the electronic dictionary device 100 searches for a headword that matches the input character. Search from dictionary data and display as a list of headword candidates. And the explanatory information corresponding to the searched headword is displayed.

[2. Dictionary data compression device]
[2.1 Configuration]
First, processing when the dictionary data is compressed in the computer 1 will be described. FIG. 2 is a block diagram showing the computer 1. As shown in FIG. 1, a computer 1 includes a CPU (Central Processing Unit) 10, a hard disk 20, a RAM (Random Access Memory) 30, a ROM (Read Only Memory) 40, an input unit 50, and a display unit 60. And.

[2.1.1 Storage area]
The hard disk 20 stores an operating system, necessary programs, data files, and the like. The hard disk 20 also includes the original English-Japanese dictionary data 200, the encoding efficiency prediction value table 206, the dictionary compression program 210, the character code compression program 212, the global search program 214, the local search program 216, and the encoding. A program 218 is stored.

  The original English-Japanese dictionary data 200 is dictionary data containing data before compression of the content of the “English-Japanese dictionary”. FIG. 3A shows an outline of the original English-Japanese dictionary data 200. In FIG. 3A, the part indicated by “XXX” represents a headword, and the part indicated by “...” explains the headword (explains the headword). Character constituting a sentence). As shown in FIG. 3A, the original English-Japanese dictionary data 200 is a series of text data in which characters printed on a paper dictionary are character codes.

  FIG. 3B is a conceptual diagram schematically illustrating the original English-Japanese dictionary data 202 divided into headword units for convenience of explanation. According to FIG. 3B, for example, the explanation information of the headwords “CD-R” and “CD-R” (hereinafter, one headword and the explanation information of the headword are combined together. "Word-unit data") is described from the "100" th byte starting from the "1" th byte of the original English-Japanese dictionary data 200. The headword unit data of the headword “CD-RW” is described from the “500” byte of the original English-Japanese dictionary data 200.

  The encoding efficiency prediction value table 206 is a table that stores the size when a character string is encoded and compressed by a dictionary-type encoding method as a prediction value. FIG. 4A shows an example of the data structure of the encoding efficiency prediction value table 206. The encoding efficiency prediction value table 206 associates a character string (for example, “[noun”) with a predicted number of bits (for example, “28”) when the character string is encoded by a dictionary-type encoding method. Is saved.

  The RAM 30 includes a memory area that temporarily holds various programs executed by the CPU 10, data related to the execution of these programs, and the like. In this embodiment, the compressed English-Japanese dictionary data 300, the main data part intermediate data 306, the 1-byte character conversion table 308, the character code conversion table 310, the character avoidance table 312, the sub table 314, and the Huffman code table 316, a reserve array storage area 318, and a headword table 320 are provided. The RAM 30 corresponds to the memory 7 in FIG.

  The post-compression English-Japanese dictionary data 300 is dictionary data obtained by compressing the original English-Japanese dictionary data 200 by the CPU 10 executing a dictionary compression process based on the dictionary compression program 210. Although details will be described later, the post-compression English-Japanese dictionary data 300 is divided into a post-processing English-Japanese reference unit 302 and an encoded English-Japanese main data unit 304.

  The main data portion intermediate data 306 is dictionary data created from the original English-Japanese dictionary data 200 when the CPU 10 executes post-processing English-Japanese reference portion creation processing. The CPU 10 stores the remaining dictionary data that is not stored as the post-processing English-Japanese reference unit 302 from the original English-Japanese dictionary data 200 in the RAM 30 as main data unit intermediate data.

  Here, the character code will be described. Generally, text characters used in a computer or the like are specified by a character code. Specifically, the half-width alphanumeric character “0” is not handled by the computer as the character “0” itself, but the character code “30h” (h is a hexadecimal number; hereinafter omitted). Is processed with the data. Character codes include a 1-byte character code and a 2-byte character code, and there are various standards. Any standard may be used, but in the present embodiment, for example, a character code according to the Shift JIS standard is referred to as a general-purpose character code. Moreover, although the original English-Japanese dictionary data 200 is described by a general-purpose character code, a part of the general-purpose character code is changed and rewritten in the character code compression processing. Hereinafter, the 1-byte character conversion table 308, the character code conversion table 310, and the character saving table 312 used in the character code compression process will be described.

  The 1-byte character conversion table 308 is a table of character codes relating to 1-byte characters, and is a table in which 256 characters are associated with the character codes. Since the character code in the 1-byte character conversion table 308 is a 1-byte character, it has two digits, the upper digit from “0” to “F” and the lower digit from “0” to “F”. It is configured. FIG. 8A shows an example of the 1-byte character conversion table 308. For example, the character code of the character “#” is “23”.

  The character code conversion table 310 is a table in which character codes of 1-byte characters are associated with general-purpose character codes. FIG. 8B shows an example of the character code conversion table 310. The character code conversion table 310 stores, for example, “89BD” as a general-purpose character code corresponding to the character code “21” of a 1-byte character.

  The character saving table 312 is a table that defines a code of the second byte of a new 2-byte character that is not included in the general-purpose character code. The new 2-byte character defined in the character saving table 312 is a general-purpose character code. For example, the character “!” Is assigned with the general-purpose character code “21” as a one-byte character in the general-purpose character code. In FIG. The character code “1300” is assigned with the code of the first byte being “13” and the code of the second byte being “00”.

  The sub-table 314 is a table that stores codes corresponding to addresses in the post-processing English-Japanese reference unit 302. FIG. 4B is a diagram illustrating an example of a data configuration of the sub table 314. For example, the address “00000000000000000111” is stored in association with the code “01”. By using the sub-table 314, the position of a character string to be referred to is expressed by a code instead of an address when encoding by a dictionary-type encoding method described later.

  The Huffman code table 316 is a table that stores a character string and a Huffman code when the character string is Huffman-coded. A unique code is assigned to each character string included in the original English-Japanese dictionary data 200.

  FIG. 4C is a diagram showing an example of the data configuration of the Huffman code table 316. For example, the Huffman code table 316 stores a character string “[noun]” and a Huffman code “101100.

  The reserve array storage area 318 is an area for storing a reserve array that is an array used when the main data portion intermediate data 306 is encoded. Here, the subscript of the reserve array indicates the number of characters from the beginning of the main data portion intermediate data 306, and the word length of the character string including the character is used as the value of the element of the array. (Hereinafter, the values of the elements of the reserve array corresponding to the character are referred to as “character-reserved reserve array values” as appropriate).

  FIG. 10B is a diagram showing an example of the reserve array stored in the reserve array storage area 318. “0” or any other positive integer value is assigned to the character correspondence reserved array value. Here, when “0” is assigned to the character correspondence reserved array value, the CPU 10 encodes the character string of the element to which “0” is assigned based on the Huffman encoding method. If a numerical value other than “0” is assigned to the character-corresponding reserved array value, the CPU 10 encodes the character string based on the dictionary-type encoding method. For example, in FIG. 10B and FIG. 10B, since “10” is assigned to the reserve array [54], the CPU 10 uses the dictionary-type encoding method for “10” characters from the “54” character. Encode.

  The headword table 320 is a table in which the start position (start byte) of the compressed English-Japanese dictionary data 300 of each headword included in the compressed English-Japanese dictionary data 300 is stored. After executing the dictionary compression process, the CPU 10 stores the start position of the headword unit data stored in the post-compression English-Japanese dictionary data 300 in the headword table 320.

  FIG. 4D is a diagram illustrating an example of the data structure of the headword table 320. The entry word table 320 stores the start byte position (for example, “49”) of the entry word unit data after encoding included in the compressed English-Japanese dictionary data 300 in ascending order.

  The ROM 40 stores an initial program (for example, a basic input / output system (BIOS)) for performing various initial settings, hardware inspection, or loading a necessary program. The CPU 10 sets the operating environment of the computer 1 by executing this initial program when the computer 1 is turned on.

[2.1.2 CPU]
The CPU 10 executes processing based on a predetermined program in accordance with an input instruction, and transfers instructions and data to each function unit. Specifically, the CPU 10 reads a program stored in the hard disk 20 in response to an operation signal input from the input unit 50, and executes processing according to the program. Then, a display control signal is appropriately output to the display unit 60 to display the processing result.

  Further, in this embodiment, the CPU 10 executes a dictionary compression process (see FIG. 5) according to the dictionary compression program 210 of the hard disk 20, and in this dictionary compression process, the character code according to the character code compression program 212 is executed. The compression process, the global search process according to the global search program 214, the local search process according to the local search program 216, and the encoding process according to the encoding program 218 are executed as subroutines.

  Specifically, in the post-processing English-Japanese reference section processing in the dictionary compression processing, the CPU 10 discretely selects headword unit data from the entire original English-Japanese dictionary data 200, and collects the selected headword unit data. A post-processing English-Japanese reference unit 302 is created. Then, the CPU 10 creates the main data portion intermediate data 306 by collecting the entry word unit data that have not been selected. Then, the CPU 10 executes a character code compression process on the characters included in the post-processing English-Japanese reference unit 302 and the main data unit intermediate data 306. Further, the CPU 10 executes a global search process and a local search process for the main data part intermediate data, and creates a reserve array that is referred to when encoding is performed by the dictionary-type encoding method. Then, the CPU 10 creates an encoded English-Japanese main data section 304 by executing an encoding process with reference to the created reserved array. Hereinafter, although details will be described later, an outline of each process will be described.

  First, the CPU 10 determines one empty code from the 1-byte character conversion table 308 in the character code compression process. Further, the CPU 10 determines a 1-byte character to be saved from the 1-byte character conversion table 308, and assigns the determined 1-byte character to the character saving table 312. Next, the CPU 10 determines two-byte characters that frequently appear in the original English-Japanese dictionary data 200 by the number of saved characters, and assigns the saved one-byte character code to the decided two-byte characters. The 1-byte character conversion table 308 is changed. Then, the CPU 10 saves the general-purpose character code of the 2-byte character determined as the frequent character and the newly assigned character code in the character code conversion table 310. Then, the CPU 10 replaces the character code of each character included in the post-processing English-Japanese reference unit 302 and the main data portion intermediate data 306 with the converted character code.

  Next, in the global search process, the CPU 10 searches whether or not a character string longer than a predetermined word length included in the main data portion intermediate data 306 is included in the post-processing English-Japanese reference portion 302 and corresponds to the character. Assign the word length to the reserved array value.

  Next, in the local search process, the CPU 10 searches the post-process English-Japanese reference unit 302 for whether or not a character string having a predetermined word length or less included in the main data part intermediate data 306 is included. The character length of the character string that is predicted to have the highest encoding efficiency among the character strings of the word length or less is assigned to the character-corresponding reserved array value.

  Next, the CPU 10 extracts a character correspondence reserved array value in the encoding process. If the character correspondence reserved array value is “0”, the CPU 10 encodes the character string with the Huffman code. Further, when the character correspondence reserved array value is other than “0”, the CPU 10 extracts the address of the character string referred to in the post-processing English-Japanese reference unit 302. Here, when the extracted address is stored in the sub-table 314, the CPU 10 performs encoding based on the sub-table 314. When the extracted address is not stored in the sub-table 314, the post-processing English-Japanese reference unit 302 is processed. Encoding is performed based on the address. Then, all character strings of the main data portion intermediate data are encoded and stored in the encoded English-Japanese main data portion 304.

[2.1.3 Input / output unit]
The input unit 50 is an input device that includes a key group necessary for character input such as kana and alphabets, function selection, and the like, and outputs a signal of a pressed key to the CPU 10. A control command input means for instructing execution of processing is realized by key input in the input unit 50. The input unit 50 corresponds to the keyboard 5 shown in FIG. 1, but is not limited to the keyboard, and may be, for example, a mouse.

  The display unit 60 displays various screens based on display signals output from the CPU 10, and is configured by a CRT (Cathode Ray Tube) or the like. The display unit 60 corresponds to the display 3 shown in FIG.

[2.2 Operation]
[2.2.1 Dictionary compression processing]
First, dictionary compression processing executed by the CPU 10 in the present embodiment will be described. FIG. 5 is a flowchart for explaining the operation of the computer 1 related to the dictionary compression processing. This dictionary compression process is a process realized by the CPU 10 executing the dictionary compression program 210 stored in the hard disk 20.

  In the dictionary compression process, the CPU 10 performs various processes to compress the original English-Japanese dictionary data 200 and create post-compression English-Japanese dictionary data 300. Hereinafter, each process included in the dictionary compression process will be described.

[2.2.2 Post-processing English-Japanese reference part creation processing]
First, the CPU 10 reads the original English-Japanese dictionary data 200 and creates a post-processing English-Japanese reference unit 302 and main data unit intermediate data 306. Specifically, the word-word unit data included in the original English-Japanese dictionary data 200 is discretely selected from the word-word unit data and added in order to be stored as the post-processing English-Japanese reference unit 302. The headword unit data not selected is stored as the main data portion intermediate data 306.

  This will be specifically described with reference to FIG. FIG. 6A shows the original English-Japanese dictionary data 200. As shown in the figure, the original English-Japanese dictionary data 200 stores a plurality of headword unit data. In the figure, only the headword portion in the headword unit data is displayed for convenience.

  FIG. 6B is a diagram when the post-processing English-Japanese reference section creation process is executed. The CPU 10 discretely selects headword unit data from the original English-Japanese dictionary data 200. Then, the CPU 10 stores the selected headword unit data in the RAM 30 as the processed English-Japanese reference unit 302. Further, the entry unit data not selected is stored in the RAM 30 as the main data portion intermediate data 306.

  As described above, according to the post-process English-Japanese reference section creation process, the post-process English-Japanese reference section 302 can be created by discretely extracting data in units of headwords from the original English-Japanese dictionary data 200. Therefore, the post-processing English-Japanese reference unit 302 can be created without being biased to a part of the original English-Japanese dictionary data 200.

[2.2.3 Character code compression processing]
[2.2.3 (a) Process flow]
FIG. 7 is a flowchart for explaining the operation of the computer 1 related to the character code compression processing. This character code compression processing is realized by the CPU 10 executing the character code compression program 212 stored in the hard disk 20.

  First, the CPU 10 determines one character code that is not used in the 1-byte character conversion table 308 (hereinafter referred to as “empty code” as appropriate) (step B10). Next, the CPU 10 determines, from the 1-byte character conversion table 308, a 1-byte code character to be saved in the character save table 312 (hereinafter, referred to as “a save character” as appropriate) (step B12). Here, as the escape character, a predetermined number is determined in order from the least frequently used characters in the original English-Japanese dictionary data 200.

  Next, the CPU 10 assigns the determined save character to the character save table 312 (step B14). Then, the CPU 10 determines as many frequent characters as the number of escape characters (step B16), and assigns the frequent characters to the 1-byte code of the 1-byte character conversion table 308 to which the escape characters have been assigned (step B18). Then, the CPU 10 associates and saves the assigned 1-byte character code and the general-purpose character code of the frequent characters, and creates the character code conversion table 310 (step B20). Then, the CPU 10 replaces the character code of the escape character included in the post-processing English-Japanese reference unit 302 and the main data portion intermediate data 306 with a 2-byte character code (step B22), and changes the character code of the frequent character to a 1-byte character code. Replace (step B24).

[2.2.3 (b) Specific example]
The character code compression process will be specifically described with reference to FIG. FIG. 8A is an example showing a part of the 1-byte character conversion table 308. One empty code “13” is determined from the one-byte character conversion table 308 (step B10). Next, the CPU 10 determines one save character from the 1-byte character conversion table 308. Here, it is assumed that “!” Is determined as the escape character. Then, the CPU 10 assigns the character “!” Assigned to the character code “21” to the position of the character code “00” in the character save table 312 as a save character. In this case, the character code of the character “!” Is “1300”.

  Next, the CPU 10 determines the same number of frequently used characters as the number of escape characters (step B16). Here, it is assumed that “what” is determined as a frequent character. Then, the CPU 10 assigns the character “what” to the place (1-byte character code) where the save character “!” Determined in step B14 was assigned (step B18). Then, the CPU 10 stores the general-purpose character code “89BD” of the character “what” in the character code conversion table 310 in association with the code “21” (step B20). In this case, the character code of the character “what” is “21”.

  FIG. 8D is a diagram in which the character code is replaced. Here, the upper “(original)” is a general-purpose character code, and the lower “(after)” is a character code after replacement by executing character code compression processing. The character code of the character “what” is replaced by “21” from the general-purpose character code “89BD”, and the character code of the character “!” Is replaced by “1300” from the general-purpose character code.

  As described above, according to the character code compression process, a 2-byte character having a high appearance frequency is replaced with a 1-byte character, a 1-byte character having a low appearance frequency is replaced with a 2-byte character, and the post-processing English-Japanese reference unit 302 and The main data part intermediate data 306 can be created. Therefore, it is possible to compress dictionary data (reduce the amount of data) by expressing 2-byte characters with high appearance frequency as 1-byte characters.

[2.2.4 Global search processing]
[2.2.4 (a) Process flow]
FIG. 9 is a flowchart for explaining the operation of the computer 1 related to the global search process. This global search process is realized by the CPU 10 executing the global search program 214 described in the hard disk 20.

  First, the search character length is substituted for the variable i (step C10). Here, the search character length is the length of the character string that matches the character string included in the post-process English-Japanese reference unit 302 and the character string included in the main data intermediate data 306 in the global search. The length of the maximum character string. In this embodiment, it is “10” characters.

  Next, the CPU 10 substitutes “0” as an initial value for the value of the variable j (step C12). Then, the CPU 10 determines whether or not the value of the reserve array [j], which is the j-th reserved array, is 0 with j as a subscript (step C14). If the value of the reserve array [j] is other than “0” (step C14; No), the CPU 10 adds “1” to the value of the variable j (step C24).

  When the value of the reserve array [j] is “0” (step C14; Yes), a character string having a word length i starting from the current position (jth character) is extracted (step C16). Then, the CPU 10 determines whether or not the same character string as the extracted character string exists in the post-processing English-Japanese reference unit 302 (step C18). If it is determined that the same character string as the extracted character string is not included in the processed English-Japanese reference unit 302 (step C18; No), the CPU 10 adds “1” to the value of the variable j and performs processing. (Step C24).

  Here, if the CPU 10 determines that the character string is included in the post-processing English-Japanese reference unit 302 (step C18; Yes), the value of the reserve array [j + i-1] from the value of the reserve array [j]. Is substituted as the value of the variable i. Then, the CPU 10 adds the value of the variable i to the value of the variable j (Step C22).

  Then, the CPU 10 determines whether or not all character strings included in the main data portion intermediate data 306 have been examined. If the determination has not been completed yet (step C26; No), the CPU 10 repeats the processing from step C14 to step C26 in the same manner.

  If the determination has been completed (step C26; Yes), the CPU 10 determines whether or not the variable i is greater than the minimum character length (step C28). Here, the minimum character length is a value serving as a threshold for the word length to be subjected to the global search, and is set to “5” in the present embodiment. Therefore, when the value of the variable i is larger than “5” (step C28; Yes), 1 is subtracted from the value of the variable i (step C30), and the processing is executed from step C12.

[2.2.4 (b) Specific example]
The global search process will be specifically described with reference to FIG. FIG. 10A is an example showing a part of the post-processing English-Japanese reference unit 302. “[Noun] [Computer ...” indicates that the 10th character starts. FIG. 10B shows a part of the main data part intermediate data 306 and a part of the corresponding reserve array in order to show the global search processing procedure. Further, “[noun ...” indicates that the main data portion intermediate data 306 starts from the 50th character.

  For example, the global search process is executed with the current value of the variable i being “10” and the value of the variable j being “50”. First, FIG. 10B and FIG. 10A are reserved array states when the global search process is not executed. All the values of the reserve array are assigned “0”.

  Next, since the reserved array [j] is 0 when the value of the variable j is “50” (step C14 in FIG. 9; Yes), the CPU 10 starts the word length “10” starting from the “50” character. The character string “[noun] [computer” is extracted (step C16). Then, it is determined whether or not the extracted character string is included in the post-processing English-Japanese reference unit 302 (step C18 in FIG. 9). Here, since the character string “[noun] [computer” is not included in the post-processing English-Japanese reference unit 302 (step 18; No), the CPU 10 adds “1” to the value of the variable j, and the variable j Is set to “51” (step C24). Similarly, the CPU 10 determines that the post-processing English-Japanese reference unit 302 is not included even when the value of j is 51 to 53.

  Next, when the value of the variable j is “54”, the CPU 10 extracts the character string “[computer term]” (step C16). Then, the CPU 10 determines that the extracted character string is included in the post-processing English-Japanese reference unit 302 (step C18; Yes). Then, the CPU 10 substitutes “10” that is the value of the variable i from the reserve array [54] to the reserve array [63] (FIG. 10 (b) (b)). Then, the CPU 10 adds the value “10” of the variable i to the value “54” of the variable j (step C22), and executes the process from the reserved array [64].

  Similarly, when the global search process is executed and all the character strings included in the main data portion intermediate data 306 are completed, the process is executed with the value of the variable i set to “9”. At this time, each element value of the reserve array does not change (FIGS. 10B and 10C). Further, when the value of the variable i is “8”, the character string “CDR” is included in the post-processing English-Japanese reference unit 302, and thus the CPU 10 changes from the reserve array [64] to the reserve array [71]. “8” is substituted (FIG. 10 (b) (d)).

  As described above, according to the global search process, it is possible to preferentially encode a character string having a long word length by using the dictionary-type encoding method and compress it. Therefore, encoding can be performed more efficiently than encoding based on a character string having a short word length.

[2.2.5 Local search processing]
[2.2.5 (a) Process flow]
FIG. 11 is a flowchart for explaining the operation of the computer 1 related to the local search process. This local search process is realized by the CPU 10 executing the local search program 216 stored in the hard disk 20.

  First, the CPU 10 substitutes “0” for the value of the variable j (step D10). Further, the CPU 10 substitutes the maximum value that can be taken for the variable min (step D12).

  Next, the CPU 10 determines whether or not the value of the reserve array [j] is “0” (step D14). Here, when the value of the reserve array [j] is other than “0” (step D14; No), the CPU 10 adds “1” to the value of the variable j and shifts the processing to step D40 (step D38). ).

  When the value of the reserve array [j] is “0” (step D14; Yes), the CPU 10 determines whether the character string starting from the j-th character and the character string included in the post-processing English-Japanese reference unit 302 are included. The character string m having the longest word length is extracted (step D16). Specifically, the CPU 10 extracts a longest character string that is a character string starting from the j-th character and whose value of each element (reserved array) of the character string is “0”. Then, it is searched whether or not the extracted character string is included in the post-process English-Japanese reference unit 302. If not, the last character is deleted and the post-process English-Japanese reference unit 302 is similarly searched. To do. By repeating this process, the character string m is extracted.

  Next, the CPU 10 reads the predicted encoding efficiency value est (m) corresponding to the extracted character string from the predicted encoding efficiency value table 206 (step D18). Then, the read encoding efficiency predicted value est (m) is compared with the value of the variable min (step D20).

  Here, when the CPU 10 determines that the encoding efficiency predicted value est (m) is smaller than the variable min (step D20; Yes), the length len (m) of the character string m is The position loc (m) of the character string m is calculated (step D22). Then, the value of est (m) is substituted for the variable min, the value of len (m) is substituted for the variable len, and the value of loc (m) is substituted for the variable loc (step D24). Further, the character string m is stored in the RAM 30 as a save character string (step D26).

  Next, when the CPU 10 determines that the value of len (m) is greater than 1 (step D28; Yes), the CPU 10 deletes the last character of the character string m and updates the character string m (step D30). D18 to step D26 are similarly repeated.

  Further, the CPU 10 determines whether or not the variable min is smaller than the Huffman code length (step D32). Here, the Huffman code length is obtained by reading the Huffman code corresponding to the saved character string from the Huffman code table 316 and calculating the code length of the read Huffman code. Then, the CPU 10 compares the value of the variable min with the Huffman code length, and when the Huffman code length is equal to or greater than the value of the variable min (step D32; No), 1 is added to the value of the variable j (step D38). . If the value of the variable min is smaller than the Huffman code length (step D32; Yes), the CPU 10 substitutes the value of the variable len from the reserve array [j] to the reserve array “j + len−1” (step D34). ). Then, the value of the variable len is added to the variable j (step D36).

  Then, the CPU 10 determines whether or not processing has been completed for all character strings included in the main data portion intermediate data (step D40). If the CPU 10 determines that the processing has not been completed for all the character strings, the processing is performed again from step D12.

[2.2.5 (b) Specific example]
The local search process will be specifically described with reference to FIG. The lower two stages of (e) and (f) in FIG. 10B are diagrams showing the reserve arrangement when the local search process is executed. First, when the value of the variable j is “49”, the reserved array when the local search processing is executed is shown in FIGS. 10B and 10E, and when the value of the variable j is “50”. FIG. 10B and FIG. 10F show the reserved array when the local search process is executed.

  First, when the value of the variable j is “49”, the value of the reserve array corresponding to the 50th to 53rd characters is “0”. Next, since the value of the reserve array [50] is 0 when the value of the variable j is “50” (step D14; Yes), the CPU 10 matches the reference portion with the character string starting from the 50th character. The longest character string “[noun]” is extracted (step D16). Then, the encoding efficiency prediction value est ([noun]) of the character string “[noun]” is read from the encoding efficiency prediction value table 206. Then, the encoding efficiency prediction value est ([noun]) is “20”. Next, “∞” is assigned as the current minimum value min, and the encoding efficiency predicted value est ([noun]) is smaller, so the CPU 10 substitutes each value. Specifically, the encoding efficiency prediction value “20” is substituted into the variable min, the length “4” of the character string “[noun]” is substituted into the variable len, and the position “ 50 ”is substituted into the variable loc. Further, the character string “[noun]” is stored in the RAM 30 as a saved character string.

  Next, the CPU 10 deletes the last character of the character string “[noun]” and executes the character string “[noun]” from step D18. In this case, the predicted encoding efficiency value of the character string “[noun” is “28”, which is larger than “20”. Similarly, a predicted encoding efficiency value is similarly calculated for “[name] from which the final character of the character string“ [noun ”is deleted, and is compared with the value of the variable min.

  When the length rec (m) of the character string m becomes 1 (step D28; No), the variable min is compared with the Huffman code length of the escape character string. At this time, if the encoding efficiency prediction value is the smallest in the case of the character string “[noun]”, the value of the variable min is “20”. Further, the CPU 10 obtains the Huffman code length of the character string “[noun]” from the Huffman code table 316. At this time, if the Huffman code length is “32” bits, since the variable min is smaller, “4” is substituted into the value of the reserve array [53] from the reserve array [50].

  Thus, according to the local search process, when a character string is encoded, it is possible to encode with a character string having a smaller number of codes. Further, by comparing the case of encoding a short character string with the Huffman code, the main data part intermediate data 306 can be compressed based on a more appropriate encoding method.

[2.2.6 Encoding process]
[2.2.6 (a) Process flow]
FIG. 12 is a flowchart for explaining the operation of the computer 1 related to the encoding process. This encoding process is realized by the CPU 10 executing the encoding program 218 stored in the hard disk 20.

  First, the CPU 10 selects a character from the main data portion intermediate data 306 (step E10). Next, the CPU 10 extracts a character correspondence reserved array value corresponding to the selected character (step E12). If the extracted value is “0” (step E14; Yes), the CPU 10 encodes the character string having the character correspondence reserved array value “0” including the selected character with the Huffman code ( Step E24).

  If the character correspondence reserved array value is other than “0” (step E14; No), the address of the reference portion of the character string to be referenced is extracted (step E16). Specifically, the character string corresponding to the value extracted from the character correspondence reserved array value is extracted from the main data part intermediate data 306. Then, the CPU 10 extracts an address (position) including the same character string as the extracted character string from the processed English-Japanese reference unit 302.

  Next, the CPU 110 determines whether or not the extracted address is included in the subtable 314. If the CPU 10 determines that it is included in the sub-table 314, it performs encoding based on the sub-table (step E20). If the CPU 10 determines that the sub-table 314 is not included, the CPU 10 encodes the character string based on the address of the post-processing English-Japanese reference unit 302 (step E22).

  Then, the CPU 10 determines whether or not encoding has been completed for all character strings included in the main data part intermediate data 306, and determines that there is a character string that has not been encoded. In such a case (step E26; No), a character string that has not been encoded is selected (step E28), and the process is executed from step E12. If it is determined that the encoding is complete (step E26; Yes), the encoding process is terminated.

[2.2.6 (b) Specific example]
The encoding process will be specifically described with reference to FIGS. 10 and 13. First, the case where the CPU 10 executes the encoding process for the character string “[noun]” will be described.

  First, when the character “[noun]” is selected and the value of the corresponding reserved array is extracted from the reserved array, “4” is obtained. Therefore, the CPU 10 extracts the position of the character string “[noun]” in the processed English-Japanese reference unit 302 (step E16). Now, the address of the character string “[noun]” in the post-processing English-Japanese reference unit 302 is “00000000000000010000”. Next, it is determined whether or not this address is stored in the subtable 314. Then, the address “00000000000000010000” is stored in association with the code “10” of the sub table 314. Then, the CPU 10 encodes the character string “[noun]” based on the sub-table and the word length.

  FIG. 13A illustrates the encoding of “[noun]”. The CPU 10 indicates that “1” indicating that the data is encoded by the dictionary-type encoding method and that the address value including the character string is stored in the sub-table 314 in the post-processing English-Japanese reference unit 302. A code string consisting of “1” shown, the code “10” of the sub-table 314 corresponding to the address of the post-processing English-Japanese reference unit 302, and “00100” representing the word length “4” of the character string “[noun]” It encodes to “111000100”. Therefore, “[noun]” is originally expressed in 64 bits, but according to the present embodiment, it can be expressed in 9 bits.

  FIG. 13B shows a state of encoding by executing the encoding process for the character string “CDR”. The character corresponding reserve array value corresponding to the character “shi” is “8”. Further, it is assumed that the post-process English-Japanese reference unit 302 including the character string “CDR” is not stored in the sub-table 314 in the post-process English-Japanese reference unit 302. When the CPU 10 executes the encoding process, the character string “CDR” is “1” indicating that the character string “CDR” is encoded by the dictionary type encoding method, and the post-processing English-Japanese reference including the character string “CDR” is referred to. “0” indicating that the address in the part 302 is not stored in the sub-table, and the first character “S” in the character string “CDR” in the post-processing English-Japanese reference unit 302 is sent to the post-processing English-Japanese reference unit 302. The encoded address “00000000000000011110” and “01000” representing the word length “8” are encoded into a code string “100000000000000001111001000”. Therefore, originally “CDR” was expressed by 128 bits, but according to the present embodiment, it can be expressed by 27 bits.

  Thus, according to the encoding process, when the main data portion intermediate data 306 is encoded by the dictionary-type encoding method, if the position of the frequently occurring character string is stored in the sub-table 314, the character string The position can be represented by a code. Therefore, it is possible to encode with a smaller amount of information than in the case of direct encoding using the position of the character string.

[3. Electronic dictionary device]
[3.1 Configuration]
FIG. 14 is a block diagram illustrating a configuration of the electronic dictionary device 100. As shown in the figure, an electronic dictionary device 100 includes a CPU (Central Processing Unit) 110, a ROM (Read Only Memory) 120, a RAM (Random Access Memory) 130, and an EEPROM (Electronically Erasable and Programmable Read Only Memory). 140, an input unit 150, and a display unit 160.

[3.1.1 Storage area]
The ROM 120 stores an initial program for performing various initial settings, hardware inspections, loading of necessary programs, and the like. The CPU 110 sets the operating environment of the electronic dictionary device 100 by executing this initial program when the electronic dictionary device 100 is powered on.

  The ROM 120 stores various programs related to the operation of the electronic dictionary device 100 such as menu display processing, various setting processing, various search processing, and programs for realizing various functions of the electronic dictionary device 100. A dictionary expansion program 1200 and a character code expansion program 1202 are provided.

  The RAM 130 includes a memory area that temporarily holds various programs executed by the CPU 110 and data related to the execution of these programs.

  The EEPROM 140 is a memory for storing various dictionary data referred to by the CPU 110 and various settings in the electronic dictionary device 100 even after the power is turned off. In this embodiment, post-compression English-Japanese dictionary data 1400, 1-byte character conversion table 1406, character code conversion table 1408, character save table 1410, sub-table 1412, Huffman code table 1414, and entry word table 1416 And. Here, the post-compression English-Japanese dictionary data 1400 is the same dictionary data as the post-compression English-Japanese dictionary data 300. The 1-byte character conversion table 1406 is the same table as the 1-byte character conversion table 308. The character code conversion table 1408 is the same table as the character code conversion table 310. The character saving table 1410 is the same table as the character saving table 312. The sub table 1412 is the same table as the sub table 314. The Huffman code table 1414 is the same table as the Huffman code table 316. The entry word table 1416 is the same table as the entry word table 320.

[3.1.2 CPU]
The CPU 110 executes processing based on a predetermined program in accordance with an input instruction, and transfers instructions and data to each function unit. Specifically, CPU 110 reads a program stored in ROM 120 in response to an operation signal input from input unit 150, and executes processing according to the program. Then, the CPU 110 appropriately outputs a display control signal to the display unit 160 to display the processing result.

  In this embodiment, the CPU 110 executes a dictionary expansion process (see FIG. 15) according to the dictionary expansion program 1200 stored in the ROM 120, and reads the character code expansion program 1202 in the dictionary expansion process. The character code expansion process is executed as a subroutine.

  Specifically, the CPU 110 searches for a headword corresponding to the input character and extracts a code string in the dictionary expansion process. Therefore, when the first bit of the extracted code string is 1, the CPU 110 determines that the code string is a code string encoded by the dictionary-type encoding method. Further, when the next 1 bit of the code string encoded by the dictionary type encoding method is 1, the CPU 110 determines that the position of the character string to be referenced is stored in the sub-table, and the character to be referred to Read column position from sub-table and decode. When the next 1 bit is 0 in the code string encoded by the dictionary-type encoding method, the CPU 110 extracts the position of the character string to be referenced from the code string, and decodes the code string. . Then, the CPU 110 executes a character code expansion process for the decoded character string, and displays the character string on the display unit 160.

  The CPU 110 extracts the first byte of the character code in the character code expansion process. Next, the CPU 110 extracts the character type of the first byte of the extracted character code from the 1-byte character conversion table 1406. In the case of a 2-byte character, it is displayed as a 2-byte character. Here, in the case of a 1-byte character, the CPU 110 determines whether or not it is a 1-byte character converted by the character code compression process (hereinafter referred to as “1-byte converted character” as appropriate). If it is a 1-byte conversion character, the CPU 110 reads the corresponding character code from the character code conversion table 1408 and displays the corresponding character.

[3.1.3 Input / output unit]
The input unit 150 is an input device including a key group necessary for inputting characters such as kana and alphabets and selecting functions, and outputs a signal of a pressed key to the CPU 110. By means of key input in the input unit 150, an input means for inputting an input character, selecting a dictionary mode, instructing search execution, starting a jump function, and the like is realized. The input unit 150 corresponds to the key group 105 in FIG. 1, but is not limited to the key group 105 and may be a touch panel or the like.

  The display unit 160 displays various screens based on display signals input from the CPU 110, and includes an LCD or the like. The display unit 160 corresponds to the display 103 shown in FIG.

[3.2 Operation]
[3.2.1 Dictionary expansion processing]
[3.2.1 (a) Process flow]
FIG. 15 is a flowchart for explaining the operation of the electronic dictionary device 100 according to the dictionary expansion processing. This dictionary expansion process is a process realized by the CPU 110 executing the dictionary expansion program 1200 stored in the ROM 120.

  First, when a character is input (step F10), the CPU 110 searches for a headword corresponding to the input character (step F12). Specifically, the CPU 110 performs a process of selecting any start position from the start positions stored in the headword table 1416 and expanding the compressed English-Japanese dictionary data 1400. Since the headword unit data start position is stored in the headword table 1416 in the storage order of the compressed English-Japanese dictionary data 1400, the start position is determined by, for example, a known search method using a binary tree. The search for the headword is executed by repeating the selection, the expansion of the headword, and the determination of whether or not the headword is suitable.

  Next, the CPU 110 extracts a code string of headword unit data of the headword corresponding to the input character (step F14). Then, the CPU 110 first extracts a code string to be decoded (hereinafter referred to as “decoding target code string” as appropriate) from the extracted code strings. Next, the CPU 10 determines whether or not the first 1 bit of the decoding target code string is “1” (step F16). Here, when the first 1 bit is “0” (step F16; No), the CPU 110 reads the character string corresponding to the Huffman code from the Huffman code table 1414 with the subsequent code string as the Huffman code, The decoding target code string is decoded (step F18).

  If the first bit of the decoding target code string is “1” (step F16; Yes), the CPU 110 determines whether or not the subsequent second bit is “1” (step F20). ). If the second bit is “1”, the sub-table number and word length are extracted from the subsequent code string (step F22). Then, the CPU 110 reads an address value corresponding to the number of the subtable from the subtable 1412 and decodes the decoding target code string based on the read address value and word length.

  On the other hand, if the second bit is “0”, the CPU 110 extracts the address value and word length of the character string to be referenced from the subsequent code string (step F24). Then, the CPU 110 decodes the decoding target code string based on the extracted address value and word length.

  Then, the CPU 110 determines whether or not all of the read code strings of headword unit data have been decoded, and when all of the code strings of headword unit data have not been decoded (step F26). No), the CPU 110 determines the subsequent decoding target code string (step F28), and executes the same processing from step F16.

  If all of the code strings of the headword unit data have been decoded (step F26; Yes), the CPU 110 executes a character code expansion process for the decoded character code (step F30). ).

[3.2.1 (b) Specific example]
This will be specifically described with reference to FIG. Here, the encoded code string in FIG. 13A is a code string included in the post-encoding English-Japanese main data part 1404, and the post-encoding English-Japanese main data part 1404 is used as a decoding target code string. , “111000100” is extracted.

  First, the CPU 110 extracts the first bit of the decoding target code string. Then, since the first first bit is “1” (step F16; Yes), the CPU 110 next extracts the second bit. Since the second bit is also “1” (step F20; Yes), the CPU 110 reads the address from the sub-table 1412 and reads the word length from the subsequent code string. Specifically, the address corresponding to the subsequent 2 bits “10” is read from the subtable 1412 as “00000000000000010000”. Then, CPU 110 extracts the subsequent code string “00100” as the word length. CPU 110 uses the extracted address as a start position, and reads the word length “4” characters from processed English-Japanese reference section 1402. Then, it is decoded into the corresponding character string “[noun]”.

  13B, it is assumed that “100000000000000001111001000” is extracted from the encoded English-Japanese main data section 1404 as a decoding target code string.

  First, the CPU 110 extracts the first bit of the decoding target code string. Then, since the first first bit is “1” (step F16; Yes), the CPU 110 next extracts the second bit. Since the extracted second bit is “0” (step F20; No), the CPU 110 extracts the code string “00000000000000011110” for the subsequent “20” bits as an address. Then, CPU 110 extracts the subsequent code string “01000” as the word length. The CPU 110 uses the extracted address as a start position, and reads the word length “8” characters from the processed English-Japanese reference unit 1402. Then, it is decoded into the corresponding character string “CDR”.

[3.2.2 Character code expansion processing]
[3.2.2 (a) Process flow]
FIG. 16 is a flowchart for explaining the operation of the electronic dictionary device 100 according to the character code decompression process. This character code expansion process is a process realized by the CPU 110 executing the character code expansion program 1202 stored in the ROM 120.

  First, the CPU 110 selects the first character from the character string decrypted by the dictionary expansion process (hereinafter referred to as “decoded character string” as appropriate), and extracts the first byte of the character code of the selected character ( Step G10). Next, CPU 110 extracts whether the extracted character code is a 1-byte character or a 2-byte character based on 1-byte character conversion table 1406 (step G12). Specifically, for example, in the table of the 1-byte character conversion table 1406, the type of the character is extracted by a flag indicating whether it is a 1-byte character or a 2-byte character, or the character code is a 1-byte character conversion table. If it is within the predetermined range of 308, it is extracted that it is a 2-byte character.

  Next, the CPU 10 determines whether the extracted character type is a 1-byte character or a 2-byte character (step G14). If the extracted character type is a 2-byte character (step G14; 2-byte character), the CPU 110 extracts the next 1 byte (step G28) and displays it on the display unit 160 as a 2-byte character. (Step G30).

  On the other hand, when the type of the extracted character is a 1-byte character, the CPU 110 determines whether it is a 1-byte converted character (step G16). If the extracted character is not a 1-byte converted character, the CPU 110 displays it on the display unit 160 as a normal 1-byte character (step G31).

  If the extracted 1-byte character is a 1-byte converted character (step G16; Yes), the CPU 110 extracts the first 1 byte from the corresponding character code from the character code conversion table 1408 (step G18). ). Then, the CPU 110 extracts whether the character type extracted in the 1-byte character conversion table is a 1-byte character or a 2-byte character (step G20), and determines (step G22). When the character type is a 1-byte character (step G22; 1-byte character), the CPU 110 displays it as a 1-byte character on the display unit 160 (step G31). If the character type is a 2-byte character (step G22; 2-byte character), the CPU 110 extracts the next 1 byte of the corresponding character code (step G24), and displays the display unit 160 as a 2-byte character. (Step G26).

  If all characters have not been expanded (step G32; No), the next character is extracted and the same processing is executed from step G10.

[3.2.2 (b) Specific example]
This will be specifically described with reference to FIG. CPU110 extracts "21" from the character code of FIG.8 (d) (step G10). Next, when the character type is determined using the 1-byte character conversion table 1406, “21” is a 1-byte converted character. Therefore, the first byte of the corresponding character code is extracted from the character code conversion table 310 (step G18). Then, since the first byte is “89”, the CPU 110 determines that the character type is a 2-byte character (step G22; 2 bytes). Therefore, the CPU 110 extracts the subsequent 1-byte “BD” and displays the corresponding character “what” with the 2-byte character code “89BD” as the general-purpose character code (step G26).

  Further, the last character code “1300” of the character string in FIG. First, the CPU 110 extracts the first byte “13” of the character code (step G10). Next, when determining the type of the character code “13”, the CPU 110 determines that the character is a 2-byte character associated with the character saving table 312 (step G14). Therefore, the next 1 byte “00” is read. Then, the CPU 110 extracts the character corresponding to “00” from the character saving table 312 to obtain the character “!” (Step G30).

  As described above, according to the character code decompression process, even if the character of the post-compression English-Japanese dictionary data 1400 is converted, the first byte of the character is extracted and the original character is extracted according to the extracted character. It becomes possible to convert to.

[4. Modified example]
In the above-described embodiment, the electronic dictionary and the electronic dictionary have been described as an electronic dictionary device as a dedicated machine. However, the present invention is not limited to such a product. For example, the electronic dictionary and the electronic dictionary can be used for mobile phones, PDAs (Personal Digital Assistants), personal computers, and the like. It may be built in (including software built-in) as an electronic dictionary device.

1 is an overview diagram of a computer and an electronic dictionary device. The block diagram of a computer. The figure which showed an example of the data structure of original English-Japanese dictionary data. The figure which showed an example of the data structure of (a) encoding efficiency prediction value table, (b) sub-table, (c) Huffman code table, (d) headword table. The figure which showed the operation | movement flow of dictionary compression processing. The figure explaining operation | movement of the post-process English-Japanese reference part creation process. The figure which showed the operation | movement flow of the character code compression process. (A) An example of a data structure of a 1-byte character conversion table, (b) An example of a data structure of a character code conversion table, (c) An example of a data structure of a character saving table, (d) An operation of a character code compression process will be described. Figure. The figure which showed the operation | movement flow of the global search process. The figure explaining operation | movement of a global search process and a local search process. The figure which showed the operation | movement flow of the local search process. The figure which showed the operation | movement flow of an encoding process. The figure explaining operation | movement of the encoding process. The block diagram of an electronic dictionary apparatus. The figure which showed the operation | movement flow of dictionary expansion | extension processing. The figure which showed the operation | movement flow of the character code expansion | extension process.

Explanation of symbols

1 computer 10 CPU
20 Hard disk 200 Original English-Japanese dictionary data 206 Encoding efficiency prediction value table 210 Dictionary compression program 212 Character code compression program 214 Global search program 216 Local search program 218 Encoding program 30 RAM
300 English-Japanese dictionary data after compression 302 English-English reference part after processing 304 English-Japanese main data part after encoding 306 Main data part intermediate data 308 1-byte character conversion table 310 Character code conversion table 312 Character escape table 314 Sub-table 316 Huffman code table 318 Reserve Sequence storage area 320 Headword table 40 ROM
50 input unit 60 display unit 100 electronic dictionary device 110 CPU
120 ROM
1200 Dictionary expansion program 1202 Character code expansion program 130 RAM
140 EEPROM
1400 Compressed English-Japanese Dictionary Data 1402 Processed English-Japanese Reference Section 1404 Encoded English-Japanese Main Data Section 1406 1-byte Character Conversion Table 1408 Character Code Conversion Table 1410 Character Escape Table 1412 Sub-table 1414 Huffman Code Table 1416 Headword Table 150 Input Section 160 Display

Claims (14)

Classifying means for classifying dictionary data in which character strings are described in a series of headword units into a reference part and a main data part,
A position memory for selecting a plurality of positions of characters appearing at a predetermined frequency as the frequent reference positions from the reference portion based on the appearance frequency of the characters, and storing a plurality of positions in association with codes for identifying each frequent reference position Means,
When the reference position is the frequent reference position, the code of the reference position and the word length from the reference position stored in the position storage means are used. When the reference position is not the frequent reference position, the reference position and A main data portion encoding means for encoding the main data portion so as to be decodable in units of headwords by a dictionary type encoding method using a word length from the reference position;
A dictionary data compression apparatus comprising: Classifying means for discriminating a plurality of data in terms of headwords from the entire dictionary data in which character strings are described in series in terms of headwords as reference parts and the rest as main data parts When,
A main data portion encoding means for encoding the main data portion so that it can be decoded in units of headwords by a dictionary-type encoding method using the reference portion as a reference source;
A dictionary data compression apparatus comprising: Dictionary data in which character strings are described in a series of headwords is divided into a reference part and a main data part, and a matching character string that matches the character string in the main data part exists in the reference part In this case, in a dictionary data compression apparatus that compresses dictionary data by a dictionary-type encoding method that encodes a character string of the main data portion with a code representing a copy of the matching character string,
A compression means for compressing a predetermined word length or more by compressing the character string in the main data portion by using the dictionary-type encoding method for a character string of a predetermined word length or more;
Storage means for storing a plurality of character strings and predicted encoding efficiency values of the character strings in association with each other;
Among the character strings included in the main data portion after being compressed by the compression means over the predetermined word length, the dictionary type encoding is performed in order from the character string having the highest encoding efficiency prediction value stored in the storage means. Evaluation value order compression means for compressing by the method;
A dictionary data compression apparatus comprising: Storage means for storing dictionary data composed of character string data;
Selecting means for selecting any one-byte character code from the standardized one-byte character codes;
A two-byte character code conversion table is created in which the first byte is a blank one-byte character code for which no corresponding character is specified from the standardized one-byte character code. Conversion table creating means for setting the selected 1-byte character code to an unset 2-byte character code;
Detecting means for detecting a double-byte character code having a predetermined appearance frequency from the character string data;
Corresponding character changing means for changing the corresponding character represented by the selected one-byte character code to the corresponding character represented by the detected two-byte character code;
First replacement means for replacing the selected one-byte character code in the character string data with a corresponding two-byte character code set in the two-byte character code conversion table;
Second replacement means for replacing the detected 2-byte character code in the character string data with the selected 1-byte character code;
A dictionary data compression apparatus comprising: Storage means for storing dictionary data composed of character string data;
Character code storage means for storing character codes and standardized character codes in association with each other;
Extracting means for extracting a first byte of a character code representing a character included in the character string data;
Character determination means for determining whether the character code is a 1-byte character code or a 2-byte character code based on the first byte extracted by the extraction means;
When the character determining unit determines that the character code is a one-byte character code, a standardized character code corresponding to the determined one-byte character code is read from the character code storage unit, and the standardized character is read. Replacement means for replacing with code,
An electronic dictionary device comprising: On the computer,
A classification step of classifying dictionary data in which a character string is described in a series of headword units into a reference part and a main data part;
A position memory for selecting a plurality of positions of characters appearing at a predetermined frequency as the frequent reference positions from the reference section based on the appearance frequency of the characters, and storing the plurality of positions in association with codes for identifying each frequent reference position Process,
When the reference position is the frequent reference position, the code of the reference position stored in the position storage step and the word length from the reference position are used. When the reference position is not the frequent reference position, the reference position and the reference position A main data portion output step for encoding and outputting the main data portion so that it can be decoded in units of headwords by a dictionary-type encoding method using a word length from a reference position;
A compressed dictionary data manufacturing method comprising: On the computer,
A classification process in which multiple pieces of data in terms of terms are discretely extracted from the entire dictionary data in which character strings are described in series in terms of terms, and used as a reference part, and the remaining part is classified as a main data part When,
A main data part output step for encoding and outputting the main data part so that it can be decoded in units of headwords by a dictionary-type encoding method using the reference part as a reference source;
A compressed dictionary data manufacturing method comprising: On the computer,
Dictionary data in which character strings are described in a series of headwords is divided into a reference part and a main data part, and a matching character string that matches the character string in the main data part exists in the reference part In this case, in a compressed dictionary data manufacturing method for compressing dictionary data by a dictionary-type encoding method for encoding a character string of the main data portion by a code representing a copy of the matched character string,
In the computer,
Of the character strings in the main data portion, a compression step of a predetermined word length or more that compresses the character string of a predetermined word length or more by the dictionary-type encoding method;
Among the character strings included in the main data portion after being compressed by the compression step with a predetermined word length or more, the evaluation value order is compressed by the dictionary encoding method in order from the character string having the highest encoding efficiency prediction value. A compression process;
An output step of outputting the character string data after compression in the evaluation value order compression step as compressed character string data;
A compressed dictionary data manufacturing method comprising: On the computer,
A compressed dictionary data production method for producing compressed dictionary data obtained by compressing dictionary data composed of character string data,
In the computer,
A selection step for selecting any one-byte character code from the standardized one-byte character codes;
A two-byte character code conversion table is created in which a blank one-byte character code for which no corresponding character is specified is specified as the first byte from the standardized one-byte character code. A conversion table creation step for setting the selected one-byte character code to an unset two-byte character code;
A detection step of detecting a 2-byte character code having a predetermined frequency of appearance from the character string data;
A corresponding character changing step for changing the corresponding character represented by the selected one-byte character code to the corresponding character represented by the detected two-byte character code;
A first replacement step of replacing the selected 1-byte character code of the character string data with a corresponding 2-byte character code set in the 2-byte character code conversion table;
A second replacement step of replacing the detected 2-byte character code with the selected 1-byte character code in the character string data after the replacement in the first replacement step;
An output step of outputting the character string data after replacement in the second replacement step as compressed character string data;
A compressed dictionary data manufacturing method comprising: On the computer,
A classification function for classifying dictionary data in which character strings are described in series in terms of headwords into a reference part and a main data part;
A position memory for selecting a plurality of positions of characters appearing at a predetermined frequency as the frequent reference positions from the reference portion based on the appearance frequency of the characters, and storing a plurality of positions in association with codes for identifying each frequent reference position Function and
When the reference position is the frequent reference position, the code of the reference position and the word length from the reference position stored in the position storage function are used. When the reference position is not the frequent reference position, the reference position and A main data part encoding function for encoding the main data part so that it can be decoded in units of headwords by a dictionary-type encoding method using a word length from the reference position;
A program to realize On the computer,
A classification function that separates the data of the headword unit from the entire dictionary data in which character strings are described in series in the unit of headword, and uses it as a reference part and the rest as the main data part When,
A main data part encoding function for encoding the main data part in a unit of headwords by a dictionary-type encoding method using the reference part as a reference source;
A program to realize Dictionary data in which character strings are described in a series of headwords is divided into a reference part and a main data part, and a matching character string that matches the character string in the main data part exists in the reference part A computer that compresses dictionary data by a dictionary-type encoding method that encodes a character string of the main data portion with a code representing a copy of the matching character string,
Among the character strings in the main data portion, a compression function of a predetermined word length or more that compresses the character string having a predetermined word length or more by the dictionary encoding method;
A storage function for storing a plurality of character strings and encoding efficiency prediction values of the character strings in association with each other;
Among the character strings included in the main data portion after being compressed by the compression function with a predetermined word length or longer, the dictionary type encoding is performed in order from the character string having the highest encoding efficiency prediction value stored in the storage function. Evaluation value order compression function to compress by the method,
A program to realize On the computer,
A storage function for storing dictionary data composed of character string data;
A selection function for selecting any one-byte character code from standardized one-byte character codes;
A two-byte character code conversion table is created in which the first byte is a blank one-byte character code for which no corresponding character is specified from the standardized one-byte character code. A conversion table creation function for setting the selected 1-byte character code to an unset 2-byte character code;
A detection function for detecting a 2-byte character code having an appearance frequency of a predetermined frequency from the character string data;
A corresponding character changing function for changing the corresponding character represented by the selected one-byte character code to the corresponding character represented by the detected two-byte character code;
A first replacement function for replacing the selected one-byte character code in the character string data with a corresponding two-byte character code set in the two-byte character code conversion table;
A second replacement function for replacing the detected 2-byte character code in the character string data with the selected 1-byte character code;
A program to realize On the computer,
A storage function for storing dictionary data composed of character string data;
A character code storage function for storing a character code and a standardized character code in association with each other;
An extraction function for extracting the first byte of a character code representing a character included in the character string data;
A character determination function for determining whether the character code is a 1-byte character code or a 2-byte character code based on the first byte extracted by the extraction function;
When the character determination function determines that the character code is a 1-byte character code, the standardized character code corresponding to the determined 1-byte character code is read from the character code storage function, Replace function to replace with code,
A program to realize

Images