Computer-implemented automatic classification of product description information

1. A computer-implemented system for automatically classifying product description information, comprising:

a server component operable to:

automatically determine whether a predetermined strip word is found in the product description information, before the first word is selected and if so, eliminate the strip word and each word found subsequent to the strip word and before a next delineator in the product description information;

automatically select a first word from the product description information;

automatically determine whether the first word is defined as a prefix word within one or more keywords, each keyword being associated with one of a plurality of classes, each class being associated with one or more keywords;

if the first word is defined as a prefix word within one or more keywords, then for each keyword for which the first word is defined as a prefix word, automatically determine whether all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence;

for each keyword for which all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence, automatically generate a new result for each class associated with the keyword;

perform the following:

automatically select a first new result;

automatically compare the first new result with one or more previous results each corresponding to a class;

if a suffix count for the first new result is greater than a suffix count for a previous result, then automatically mark the new result as unambiguous, the new result being thereafter considered a previous result; and

if one or more new results remain unselected, then automatically select a next new result and repeat the above-described steps of comparing, marking, and selecting a next new result until no new results generated for the first word remain unselected;

if one or more words in the product description information remain unselected after processing the first word, then automatically select a next word and repeat the above-described steps of determining, determining, generating, performing, and selecting a next word until no words remain unselected; and

if a single previous result is marked as unambiguous after all the words in the product description information have been selected and processed, then automatically classify the product description information in the class corresponding to the previous result marked as unambiguous.

2. The system of claim 1 , wherein the suffix words are consecutive and corresponding words found in sequence in the product description information are not consecutive, the product description information comprising an intervening word not corresponding to a suffix word located between two words corresponding to suffix words.

3. The system of claim 1 , wherein the server component is further operable to:

maintain each result as a separate result data structure until deleted; and

maintain a list data structure comprising a list of all results for which result exist, each result having an associated indicator in the list data structure that may be used to mark the result as unambiguous, ambiguous, or unclassified as appropriate.

4. The system of claim 1 , wherein the new result is either:

compared with a single previous result marked as unambiguous; or

compared with a plurality of previous results sharing a highest suffix count.

5. The system of claim 4 , wherein each previous result that shares the highest suffix count has been marked as ambiguous to indicate that classification of the product description information in the class corresponding to the previous result should be manually validated.

6. The system of claim 1 , wherein the server component is further operable to, if the suffix count for the first new result is less than the suffix count for the previous result, then automatically select a next new result and compare the next new result with the one or more previous results until:

the suffix count for a selected next new result is greater than or equal to the suffix count for a previous result; or

there are no more new results from which to select.

7. The system of claim 1 , wherein the server component is further operable to, if the suffix count for the first new result is equal to the suffix count for the previous result, then automatically determine whether the class for the first new result is the same as the class for the previous result and:

if not, mark the previous result as ambiguous to indicate that classification of the product description information in the class corresponding to the previous result should be manually validated; and

if so, add the one or more keywords associated with the class to a sublist of keywords maintained for the previous result.

8. The system of claim 1 , wherein the server component is further operable to, if a single previous result is not marked as unambiguous after all words in the product description information have been selected and processed, then:

for each previous result with a highest suffix count, automatically determine whether a predetermined exclude word for the class corresponding to the previous result is found among all the words of the product description information and, if so, eliminate the previous result; and

if a single previous result with a highest suffix count remains, automatically mark this previous result as unambiguous and classify the product description information in the class corresponding to this previous result.

9. The system of claim 8 , wherein the server component is further operable to, if more than one previous result with a highest suffix count remains, automatically mark this previous result as ambiguous to indicate that classification of the product description information in the class corresponding to this previous result should be manually validated.

10. A computer-implemented method for automatically classifying product description information, comprising:

automatically determining whether a predetermined strip word is found in the product description information, before the first word is selected, and if so, eliminating the strip word and each word found subsequent to the strip word and before a next delineator in the product description information;

automatically selecting a first word from the product description information;

automatically determining whether the first word is defined as a prefix word within one or more keywords, each keyword being associated with one of a plurality of classes, each class being associated with one or more keywords;

if the first word is defined as a prefix word within one or more keywords, then for each keyword for which the first word is defined as a prefix word, automatically determining whether all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence;

for each keyword for which all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence, automatically generating a new result for each class associated with the keyword;

performing the following:

automatically selecting a first new result;

automatically comparing the first new result with one or more previous results each corresponding to a class;

if a suffix count for the first new result is greater than a suffix count for a previous result, then automatically marking the new result as unambiguous, the new result being thereafter considered a previous result; and

if one or more new results remain unselected, then automatically selecting a next new result and repeating the above-described steps of comparing, marking, and selecting a next new result until no new results generated for the first word remain unselected;

if one or more words in the product description information remain unselected after processing the first word, then automatically selecting a next word and repeating the above-described steps of determining, generating, performing, and selecting a next word until no words remain unselected; and

if a single previous result is marked as unambiguous after all words in the product description information have been selected and processed, then automatically classifying the product description information in the class corresponding to the previous result marked as unambiguous.

11. The method of claim 10 , wherein the suffix words are consecutive and corresponding words found in sequence in the product description information are not consecutive, the product description information comprising an intervening word not corresponding to a suffix word located between two words corresponding to suffix words.

12. The method of claim 10 , wherein:

once generated, each result is maintained as a separate result data structure until deleted; and

a list of all results for which result data structures exist is maintained using a list data structure, each result having an associated indicator in the list data structure that may be used to mark the result as unambiguous, ambiguous, or unclassified as appropriate.

13. The method of claim 10 , wherein the new result is either:

compared with a single previous result marked as unambiguous; or

compared with a plurality of previous results sharing a highest suffix count.

14. The method of claim 13 , wherein each previous result that shares the highest suffix count has been marked as ambiguous to indicate that classification of the product description information in the class corresponding to the previous result should be manually validated.

15. The method of claim 10 , further comprising if the suffix count for the first new result is less than the suffix count for the previous result, then automatically selecting a next new result and comparing the next new result with the one or more previous results until:

the suffix count for a selected next new result is greater than or equal to the suffix count for a previous result; or

there are no more new results from which to select.

16. The method of claim 10 , further comprising if the suffix count for the first new result is equal to the suffix count for the previous result, then automatically determining whether the class for the first new result is the same as the class for the previous result and:

if not, marking the previous result as ambiguous to indicate that classification of the product description information in the class corresponding to the previous result should be manually validated; and

if so, adding the one or more keywords associated with the class to a sublist of keywords maintained for the previous result.

17. The method of claim 10 , further comprising if a single previous result is not marked as unambiguous after all words in the product description information have been selected and processed, then:

for each previous result with a highest suffix count, automatically determining whether a predetermined exclude word for the class corresponding to the previous result is found among all the words of the product description information and, if so, eliminating the previous result; and

if a single previous result with a highest suffix count remains, automatically marking this previous result as unambiguous and classifying the product description information in the class corresponding to this previous result.

18. The method of claim 17 , further comprising if more than one previous result with a highest suffix count remains, automatically marking this previous result as ambiguous to indicate that classification of the product description information in the class corresponding to this previous result should be manually validated.

19. Software for automatically classifying product description information, the software being embodied in computer-readable media and when executed operable to:

automatically determine whether a predetermined strip word is found in the product description information, before the first word is selected, and if so, eliminate the strip word and each word found subsequent to the strip word and before a next delineator in the product description information;

automatically select a first word from the product description information;

automatically determine whether the first word is defined as a prefix word within one or more keywords, each keyword being associated with one of a plurality of classes, each class being associated with one or more keywords;

if the first word is defined as a prefix word within one or more keywords, then for each keyword for which the first word is defined as a prefix word, automatically determine whether all suffix words of the prefix word within the keyword are found among all remaining words in the product description information sequence;

for each keyword for which all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence, automatically generate a new result for each class associated with the keyword;

perform the following:

automatically select a first new result;

automatically compare the first new result with one or more previous results each corresponding to a class;

if a suffix count for the first new result is greater than a suffix count for a previous result, then automatically mark the new result as unambiguous, the new result being thereafter considered a previous result; and

if one or more new results remain unselected, then automatically select a next new result and repeat the above-described operations of comparing, marking, and selecting a next new result until no new results generated for the first word remain unselected;

if one or more words in the product description information remain unselected after processing the first word, then automatically select a next word and repeating the above-described operations of determining, generating, performing, and selecting a next word until no words remain unselected; and

if a single previous result is marked as unambiguous after all words in the product description information have been selected and processed, then automatically classify the product description information in the class corresponding to the previous result marked as unambiguous.

20. The software of claim 19 , wherein the suffix words are consecutive and corresponding words found in sequence in the product description information are not consecutive, the product description information comprising an intervening word not corresponding to a suffix word located between two words corresponding to suffix words.

21. The software of claim 19 , operable to:

maintain each result as a separate result data structure until deleted; and

maintain a list data structure comprising a list of all results for which result exist, each result having an associated indicator in the list data structure that may be used to mark the result as unambiguous, ambiguous, or unclassified as appropriate.

22. The software of claim 19 , wherein the new result is either:

compared with a single previous result marked as unambiguous; or

compared with a plurality of previous results sharing a highest suffix count.

23. The software of claim 22 , wherein each previous result that shares the highest suffix count has been marked as ambiguous to indicate that classification of the product description information in the class corresponding to the previous result should be manually validated.

24. The software of claim 19 , operable to, if the suffix count for the first new result is less than the suffix count for the previous result, then automatically select a next new result and compare the next new result with the one or more previous results until:

the suffix count for a selected next new result is greater than or equal to the suffix count for a previous result; or

there are no more new results from which to select.

25. The software of claim 19 , operable to, if the suffix count for the first new result is equal to the suffix count for the previous result, then automatically determine whether the class for the first new result is the same as the class for the previous result and:

if not, mark the previous result as ambiguous to indicate that classification of the product description information in the class corresponding to the previous result should be manually validated; and

if so, add the one or more keywords associated with the class to a sublist of keywords maintained for the previous result.

26. The software of claim 19 , operable to, if a single previous result is not marked as unambiguous after all words in the product description information have been selected and processed, then:

for each previous result with a highest suffix count, automatically determine whether a predetermined exclude word for the class corresponding to the previous result is found among all the words of the product description information and, if so, eliminate the previous result; and

if a single previous result with a highest suffix count remains, automatically mark this previous result as unambiguous and classifying the product description information in the class corresponding to this previous result.

27. The software of claim 26 , operable to, if more than one previous result with a highest suffix count remains, then automatically mark this previous result as ambiguous to indicate that classification of the product description information in the class corresponding to this previous result should be manually validated.

28. A computer-implemented system for automatically classifying product description information comprising:

means for automatically determining whether a predetermined strip word is found in the product description information, before the first word is selected, and if so, eliminating the strip word and each word found subsequent to the strip word and before a next delineator in the product description information;

means for automatically selecting a first word from the product description information;

means for automatically determining whether the first word is defined as a prefix word within one or more keywords, each keyword being associated with one of a plurality of classes, each class being associated with one or more keywords;

means for, if the first word is defined as a prefix word within one or more keywords, then for each keyword for which the first word is defined as a prefix word, automatically determining whether all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence;

means for, for each keyword for which all suffix words of the prefix word within the keyword are found among all remaining words in the product description information in sequence, automatically generating a new result for each class associated with the keyword;

means for performing the following:

automatically selecting a first new result;

automatically comparing the first new result with one or more previous results each corresponding to a class;

if a suffix count for the first new result is greater than a suffix count for a previous result, then automatically marking the new result as unambiguous, the new result being thereafter considered a previous result; and

if one or more new results remain unselected, then automatically selecting a next new result and repeating the above-described steps of comparing, marking, and selecting a next new result until no new results generated for the first word remain unselected;

means for, if one or more words in the product description information remain unselected after processing the first word, then automatically selecting a next word and repeating the above-described steps of determining, generating, performing, and selecting a next word until no words remain unselected; and

means for, if a single previous result is marked as unambiguous after all words in the product description information have been selected and processed, then automatically classifying the product description information in the class corresponding to the previous result marked as unambiguous.