Electronic spelling machine having ordered candidate words

• US 5,218,536 A
• Filed: 01/02/1990
• Issued: 06/08/1993
• Est. Priority Date: 05/25/1988
• Status: Expired due to Fees

• Current Assignee: Franklin Electronic Publishers Incorporated (Saunders Acquisition Corp.)
• Original Assignee: Franklin Electronic Publishers Incorporated (Saunders Acquisition Corp.)
• Inventors: McWherter, David
• Primary Examiner(s): Weinhardt, Robert A.
• Assistant Examiner(s): Tran, Khai
• Application Number:
  US07/459,942
• Time in Patent Office: 1,253 Days
• Field of Search: 364/419, 364/900, 364/200, 395/600, 400/61, 400/62, 400/63, 400/66
• US Class Current: 715/267
• CPC Class Codes: G06F 16/90348 by searching ordered data, e.g. alpha-numerically ordered data
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G09B 7/02 of the type wherein the student is expected to construct an answer to the question which is presented or wherein the machine gives an answer to the question presented by a student

Abstract

A spelling corrector develops a phonetic correction candidate list and a typographic correction candidate list. These lists are merged using similarity values that measure the degree of similarity between the query (input) word and each word on the candidate lists. The similarity values include one based on phonetic comparison one based on typographic (ASCII, for example) comparison, one based on a vowel and consonant typographic pattern comparison and one based on a consonant phonetic comparison.

Claims

1. In an electronic machine spelling corrector to provide an output list of candidate words in response to an input query word, the method of merging a list of phonetic correction candidate words with a list of typographic correction candidate words to provide a single merged output list for presentation to a user, comprising:

for each candidate word on the phonetic list, providing a first set of similarity values including a first value P1 vased on a phonetic comparison of the candidate word with the query word and a second value A1 based on a typographic comaprison of the candidate word with the query word,

for each candidate word on the typographic list, providing a second set of similarity values including a third value P2 based on a phonetic comparison of the candidate word with the query word and a fourth value A2 based on a typographic comparison of the candidate word with the query word,

merging candidate words on the phonetic list and typographic list into the output list, and

ordering said candidate words on said merged list as a partial function of said four similarity values.

2. The method of claim 1 including the steps of:

calculating an average of said similarity values for each of said candidate words from said phonetic and typographic lists, and wherein;

said step of ordering includes using said averages as a partial basis for determining the position of the candidate word on the merged list.

3. The method of claim 1 further comprising the steps of:

for each candidate word on the phonetic list, including in said first set of similarity values, a fifth similarity value E1 based on an essence comparison of the consonants of the candidate word with the consonants of the query word,

assigning to a first priority class, (a) candidate words from said phonetic list wherein said first similarity value P1 and said second similarity value A1 indicate a perfect match with the query word and (b) candidate words from said typographic list wherein said third similarity value P2 and said fourth similarity value A2 indicate a perfect match with the query word,

assigning to a second priority class, (a) candidate words from said phonetic list wherein said first similarity value P1 and said fifth similarity value E1 indicate a perfect match with the query word and (b) candidate words from said typographic list which have a vowel transposition compared to the query word,

assigning to a third priority class, (a) candidate words from said phonetic list wherein (i) said first value P1 exceeds a first predetermined threshold and (ii) said fifth value E1 exceed a second predetermined threshold and (iii) the first and last phonetic codes of the candidate words and the query words are the same, and (b) candidate words from said typographic list, have either (i) a vowel substitution compared to the query word or (ii) have a vowel addition adjacent to another vowel in the query word or (iii) a vowel deletion adjacent to another vowel in the query word, and

assigning to one or more lesser priority classes, those words wherein (a) said first similarity value P1 exceeds a predetermined threshold or (b) said fourth similarity value A2 exceeds a predetermined threshold.

4. The method of claim 3 including the steps of:

calculating an average of said similarity values for each of said candidate words from said phonetic and typographic lists, and

ordering said candidate words on said merged list within each of said priority classes as a function of said averages.

5. The method of claim 3 further comprising the steps of:

for each candidate word on said phonetic list, including in said first set of similarity values, a sixth similarity value V1 based on a vowel and consonant pattern comparison of the candidate word with the query word,

for each candidate word on said typographic list, includign in said second set of similarity values, a seventh similarity value V2 based on a vowel and consonant pattern comparison of the candidate word with the query word,

calculating an average of said similarity values for each of said candidate words from each of said lists, and

ordering said candidate words within each of said priority classes as a function of said averages.

6. In an electronic machine spelling corrector which provides an output list of candidate words in response to an input query word, the output list being the merger of a list of phonetic correction candidate words and a list of typographic correction candidate words comprising:

a first class of words comprising (a) candidate words from said phonetic list wherein a phonetic comparison of the candidate word with the query word and a typographic comaprison of the candidate word with the query word indicates a perfect match, and (b) candidate words from said typographic list wherein a phonetic comparison of the candidate word with the query word and a typographic comparison of the candidate word with the query word indicates a perfect match,

a second class of words excluding words in said first class comprisign (a) candidate words from said phonetic list whereni a phonetic comparison of the candidate word with the query word and an essence comparison of the candidate word with the query word indicates a perfect match, and (b) candidate words from said typographic list which have a vowel transposition compared to the query word,

a third class of words excluding words in said first and second classes and comprising (a) words from said phonetic list wherein (i) a phonetic comparison of the candidate word with the query word and an essence comparison of the candidate word with the query word exceed a pre-determined threshold and (ii) the first and last phonetic codes of the candidate word and the query word are the same, and (b) candidate words from said typographic list which have a vowel substitution compared to the query word or which have a vowel addition or a vowel deletion adjacent to another vowel in the query word, and

at least a fourth class of words excluding words from said first, second and third classes and comprising those candidate words from said phonetic list whose phonetic comparison with the query word exceeds a predetermined threshold and those candidate words from said typographic list whose typographic comparison with the query word exceeds a predetermined threshold.

7. The output list of claim 6 wherein:

position within each of said classes is determined by an average of comparison values calculated from each of said phonetic, typographic and essence comparisons.

8. The otuput list of claim 7 wherein:

said average of comparison values includes a comparison value based on a vowel and consonant pattern comparison of the candidate word with the query word.

9. In an electronic machine spelling corrector, the output list of candidate words in response to an input query word, the output list being the merger of a list of phonetic correction candidate words and a list of typographic correction candidate words, comprising:

a first set of candidate words comprisign candidate words from said phonetic list and said typographic list where the phonetic code and typographic code for the candidate word are substantially identical to the phonetic code and typographic code for the query word,

a second set of candidate words comprising;

(a) candidate words from the phonetic list where the phonetic code and the essence code for the candidate words are substantially identical to the phonetic code and the essence code for the query word, and

(b) words from said typographical list where the only difference between the candidate word and the query word is a vowel transposition,

said second set excluding all words in said first set,

a third set of candidate words comprising;

(a) candidate words from the phonetic list which meet the three criteria that (i) a comparison of the phonetic codes of the candidate word and query word exceed a first predetermined value, (ii) a comparison of the essence codes of the candidate word and the query word exceed a second predetermined value, and (iii) the first and last phonetic codes of the candidate word and query word are identical, and

(b) candidate words from said typographic list which have either (i) a vowel substitution as comapred with the query word or (ii) a vowel addition or deletion adjacent to a vowel in the query word,

said third set excluding all words in aid first and second set, and

a fourth set of candidate words comprising;

(a) words from said phonetic list whose phonetic code comparison with the phonetic code of the query word provides a comparison value that exceeds a predetermined value, and

(b) words from said typographic code whose typographic code comparison with the typographic code of the query word exceeds a second predetermined value,

said fourth set excluding all words in said first, second or third sets.

10. The output list of claim 9 wherein:

position within each of said sets is determined by an average of comparison values calculated from each of said phonetic, typographic and essence comparisons.

11. The output list of claim 10 wherein:

said average of comparison values includes a comparison value based on a vowel and consonant pattern comparison of the candidate word with the query word.