Method and apparatus for mapping multiple-byte characters to unique strings of ASCII characters for use in text retrieval

• US 5,337,233 A
• Filed: 04/13/1992
• Issued: 08/09/1994
• Est. Priority Date: 04/13/1992
• Status: Expired due to Term

• Current Assignee: Sun Microsystems Incorporated (Oracle Corporation)
• Original Assignee: Sun Microsystems Incorporated (Oracle Corporation)
• Inventors: Hofert, David K., Yoshida, Yutaka
• Primary Examiner(s): Weinhardt, Robert A.
• Application Number:
  US07/867,852
• Time in Patent Office: 848 Days
• Field of Search: 364/419, 364/514, 364/419.15, 364/419.09, 364/419.14, 364/419.16, 364/419.01, 364/419.03, 364/419.1, 364/419.19
• US Class Current: 715/234
• CPC Class Codes: G06F 40/126 Character encoding
• CPC Class Codes: G06F 40/151 Transformation

Abstract

An apparatus and method for converting a non-English language document text or search and retrieval argument into a form which can be processed by an existing ASCII based automated text processing system, even though the non-English language may have thousands of characters in it, thereby allowing the use of existing text processing systems and existing text data bases without the need to convert these text processing systems to handle multi-byte character languages.

Claims

1. A method for preparing language text to be used by a text processing system, where said language comprises more than 256 characters, said method comprising the computer implemented steps of:

a) capturing an input stream of characters which represent said language;

b) separating said input stream of characters into strings of characters which represent words;

said separation into words being accomplished by using rules of grammar of said language to determine where to insert word separators into said stream of characters to delimit groups of said characters which constitute words in said language, and whereby no special characters in said input stream of characters is required to distinguish one word or character set from another;

c) mapping said character strings into unique sets of single byte ASCII characters; and

d) transferring said unique sets of single byte ASCII characters which represent words to said text processing system for further processing.

2. The method described in claim 1 comprising the additional steps of:

a) detecting whether said words in said language are compound words, a compound word being comprised of character strings of more than two bytes of character code;

b) comparing a detected compound word to words stored in a look-up table of compound words, said look-up table comprising compound words as arguments of said look-up table and related component words of said arguments as function elements of each argument;

c) designating said component word as a single word if said compound word is not found in said look-up table; and

d) if said compound word is found in said look-up table, substituting component words contained in said look-up table which are related to said compound word in place of said compound word;

whereby a look-up table of compound words of said language is used to facilitate the separation of certain ones of said compound words into their respective simple word components in said character strings.

3. The method described in claim 2 wherein said language is Japanese.

4. The method described in claim 3 wherein said step of capturing said Japanese language into said input stream of characters is accomplished by the use of extended UNIX code ("EUC") which comprises a two byte hexadecimal code for each of said Japanese characters.

5. The method described in claim 4 wherein said step of inserting said word separators into said hexadecimal character strings comprises the following additional steps:

a) determining whether each of said two byte hexadecimal codes belongs to a kanji, katakana or hiragana character-set;

b) inserting one of said word separators between any two adjacent ones of said two byte hexadecimal codes which belong to different ones of said kanji, katakana or hiragana character-set codes;

c) identifying a string of two or more of said two byte hexadecimal codes which are of the same one of said kanji, katakana or hiragana character-sets as a compound word; and

d) comparing each of said compound words to entries in a compound word look-up table and if a match occurs, separating said compound word which matches into its constituent word parts as indicated by said look-up table and inserting one of said word separators between each of said word parts.

6. An apparatus for preparing language text to be used by a text processing system, where said language comprises more than 256 characters, said apparatus comprising:

a) a filter device for capturing an input stream of characters which represent said language comprising;

a word falter for separating said input stream of characters into strings of characters which represent words, wherein said word filter comprises a grammar analyzer to facilitate the separation of said characters into strings of said characters which represent words, wherein no special characters in said input stream of characters is required to distinguish one word or character set from another, and wherein character strings comprising more than two bytes of data art called compound words;

b) a mapping device coupled to said word filter, for mapping said strings of characters which represent words into unique strings of single-byte ASCII characters; and

c) an output device coupled to said mapping device, for passing said unique strings of single-byte ASCII characters which represent words to said text processor.

7. The apparatus in claim 6 wherein said language is Japanese.

8. The apparatus in claim 7 wherein said input stream of characters which represent said Japanese language used by said filter device employs extended UNIX code ("EUC") to represent said characters.

9. The apparatus in claim 8 wherein said word filter further comprises a separating mechanism for comparing each of said compound words with entries in a first look-up table, said first look-up table comprising compound words as arguments of said look-up table and related component words of said arguments as function elements of each argument;

said separating mechanism designating said compound word as a single word if said compound word is not found in said first look-up table; and

if said compound word is found in said first look-up table, said separating mechanism substituting component words contained in said first look-up table which are related to said compound word in place of said compound word.

10. The apparatus in claim 9 wherein said word filter further comprises an insertion device for inserting English words into said word groups of characters by using a second table look-up device for comparing each of said word groups of characters with entries in said second look-up table, said second look-up table comprising word groups of characters as arguments of said second look-up table and related English words as function elements of each argument:

and if a word group of characters is found to match an argument in said second look-up table, said insertion device inserting any related English words found in said second look-up table into said word groups of characters representing said Japanese words.

11. The apparatus in claim 7 wherein said word filter comprises:

a) a character-set analyzer for determining whether a code representing each of said Japanese characters belongs to a kanji, katakana, or hiragana character-set;

b) a character insertion device coupled to said character-set analyzer, for inserting a word separator mark in said character code string whenever said character-set analyzer determines that said character-set has changed, said word separator mark comprising a unique character which is inserted into said character string between adjacent character code strings which represent character-sets which are different from each other;

c) a word size analyzer device coupled to said character-set analyzer, for selecting a series of said Japanese characters as a compound word when said character-set analyzer determines that a character code string between two consecutive word separators comprises more than two bytes of data; and

d) a word separating device coupled to said word size analyzer for separating said compound word into its constituent word parts and marking each of said constituent word parts as a separate word by attaching a word separator mark to each of said constituent word parts.

12. In a computer system comprising an ASCII based text processing system, a method for preparing language text to be used by said ASCII based text processing system, where said language comprise more than 256 characters, said method comprising the step of:

a) translating said language text into hexadecimal character strings;

b) mapping said hexadecimal character strings into unique sets of single byte ASCII characters;

c) inserting word separators in said hexadecimal character strings to delimit groups of said hexadecimal characters which consitute words in said language wherein the rules of grammar or said language are used to determine where to insert said word separators into said hexadecimal character strings, and wherein no special character in said input stream of characters is required to distinguish one word or character set from another; and

d) transferring said unique sets of single byte ASCII characters to said ASCII based text processing system for further processing.

13. The method described in claim 12 wherein a look-up table of compound words of said language is used to facilitate the separation of certain ones of said compound words into their respective simple word components in said hexadecimal character strings.