System and method for processing morphological and syntactical analyses of inputted Chinese language phrases

• US 6,014,615 A
• Filed: 08/29/1997
• Issued: 01/11/2000
• Est. Priority Date: 08/16/1994
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chen, Chengjun Julian
• Primary Examiner(s): Thomas, Joseph
• Application Number:
  US08/920,805
• Time in Patent Office: 865 Days
• Field of Search: 704/2, 704/3, 704/4, 704/7, 704/8, 704/9, 704/10, 707/535, 707/536, 400/110, 345/171
• US Class Current: 704/3
• CPC Class Codes: G06F 3/018 Input/output arrangements for oriental characters
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/279 Recognition of textual entities

Abstract

Phonetic Chinese (Pinyin and BPMF) is entered into a computer system and accurately converted into the Hanzi form. The system has a novel keyboard with diacritic keys (and corresponding ASCII coding) that permit the user to annotate each entered phonetic text syllable with a diacritic that indicates the tone of the syllable. A process executing on the system determines that a syllable has been entered when a diacritic (or delimiter) key is struck. An entered phonetic syllable is then compared to a list of acceptable phonetic syllables and abbreviations. If the entered syllable is on the list, the correctly spelled and accented syllable is stored in memory and displayed on a phonetic portion of a graphical display. The process continues for succeeding syllables until a delimiter is entered. Upon encountering a delimiter, the word string (defined as the string of characters between two delimiters) is analyzed using morphological and syntactical processes and/or a statistical language model to unambiguously determine the proper Hanzi characters that represent the word(s) in the word string. The unique Hanzi translation is stored in memory and displayed on a Hanzi portion of the graphical interface.

Claims

1. A method where a word string is processed by a morphology process comprising the steps of:

removing one or more affixes from the word string to create a root, the removed affix being one of the affixes on an affix list;

comparing the root to one or more of the words in a vocabulary to find a match, the vocabulary having a plurality of words, each with one or more Hanzi translations, the word in the vocabulary matching the root being the root match; and

storing in the computer memory the Hanzi translation of the root match.

2. A method, as in claim 1, where one or more roots are unmatched roots, the unmatched roots having no match in the vocabulary, and the unmatched roots are decomposed into constituent words.

3. A computer system for processing Chinese language text comprising:

an input apparatus for entering a phonetic Chinese language phrase, the phrase having one or more words, each word having one or more syllables, each syllable having one or more characters, the phrase being a string of the characters between a first and second phrase delimiter;

an affix list having a plurality of entries being phonetic Chinese affixes;

a vocabulary of Chinese words, the vocabulary being a list of a plurality of phonetic Chinese words with a Hanzi translation; and

a morphology unit that removes one or more affixes from the phrase to create a root, the removed affix being one of the affixes on the affix list, the morphology unit comparing the root to one or more of the words in the vocabulary to find a match, and storing in a computer memory the Hanzi translation of the word in the vocabulary that matches the root.

4. A system, as in claim 3, further comprising a syntactical unit, the syntactical unit parsing the root into accented words and unaccented words, the accented words being marked with diacritics and unaccented words being not marked with diacritics.

5. A system, as in claim 4, where the entries include affixes, function words, and particles, each entry having a Hanzi translation, the syntactical unit matching the unaccented words with one or more of the entries, the entry matching he unaccented word being a matched entry, and the syntactical unit using the Hanzi translation of the matched entry as a translation of unaccented word in the phrase.

6. A system, as in claim 5, where the syntactical unit further translates one or more of the accented words as an ambiguous notional word, the ambiguous notional word having an ambiguity because of there is more than one Hanzi translation of the ambiguous notional word in the vocabulary, and the syntactical unit removing the ambiguity by using a relationship of the ambiguous notional word with one or more of the unaccented words words in the phrase.

7. A system, as in claim 4, where the first and second phrase delimiters are any one of the following:

a carriage return, a space, a phrase mark, a Chinese period, a left quotation mark, a right quotation mark, a left title mark, a right title mark, an exclamation mark, a comma, a slash, a left parenthesis, a right parenthesis, a colon, a semicolon, a bullet, a question mark, and two or more strokes of a tone key.

8. A system, as in claim 4, where the morphology unit decomposes the word into constituent words if the word does not match one of the words in the vocabulary and the syntactical unit parses the constituent words.

9. A method of syntactically analyzing a Chinese phrase of phonetic syllables comprising the steps of:

parsing the Chinese phrase into accented words with one or more syllables marked with a diacritic indicating a tone of the syllable and unaccented words having no syllables marked with a diacritic;

matching the the unaccented words with one or more of the entries, the entries being zero or more affixes, function words, and particles on an affix list, each entry having a Hanzi translation;

using the respective Hanzi translation to translate the unaccented word into Hanzi.

10. A method, as in claim 9, comprising the further step of:

translating one or more of the accented words as a notional word, zero or more of the notional words being an ambiguous notional word, the ambiguous notional word having more than one translation into Hanzi in a vocabulary.

11. A method, as in claim 10, where one or more of the translations is deleted because the deleted translation does not have a relationship with one of the unaccented words.