DATA CLASSIFICATION USING MACHINE LEARNING TECHNIQUES

• US 20110145178A1
• Filed: 02/23/2011
• Published: 06/16/2011
• Est. Priority Date: 07/12/2006
• Status: Active Grant

• Current Assignee: Kofax Incorporated
• Original Assignee: Kofax Incorporated
• Inventors: Schmidtler, Mauritius A.R., Borrey, Roland
• Granted Patent: US 8,239,335 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]

Abstract

A system and article of manufacture enabling adapting to a shift in document content according to one embodiment of the present invention includes instructions for: receiving at least one labeled seed document; receiving unlabeled documents; receiving at least one predetermined cost factor; training a transductive classifier using the at least one predetermined cost factor, the at least one seed document, and the unlabeled documents; classifying the unlabeled documents having a confidence level above a predefined threshold into a plurality of categories using the classifier; reclassifying at least some of the categorized documents into the categories using the classifier; and outputting identifiers of the categorized documents to at least one of a user, another system, and another process. Systems and articles of manufacture for separating documents are also presented. Systems and articles of manufacture for document searching are also presented.

Claims

1. An article of manufacture comprising:

a program storage medium readable by a computer, where the medium tangibly embodies one or more programs of instructions executable by a computer to perform a method of data classification, the one or more programs of instructions comprising;

instructions for receiving at least one labeled seed document;

instructions for receiving unlabeled documents;

instructions for receiving at least one predetermined cost factor;

instructions for training a transductive classifier using the at least one predetermined cost factor, the at least one seed document, and the unlabeled documents;

instructions for classifying the unlabeled documents having a confidence level above a predefined threshold into a plurality of categories using the classifier;

instructions for reclassifying at least some of the categorized documents previously categorized by a different classifier into the categories using the classifier; and

instructions for outputting identifiers of the categorized documents to at least one of a user, another system, and another process.

2. The article of manufacture of claim 1, wherein the method further comprises moving an unlabeled document having a confidence level below the predefined threshold into one or more new categories.

3. The article of manufacture of claim 1, further comprising instructions for training the transductive classifier through iterative calculation using at least one predetermined cost factor, the at least one seed document, and the unlabeled documents, wherein for each iteration of the calculations the cost factor is adjusted as a function of an expected label value, and using the trained classifier to classify the unlabeled documents.

4. The article of manufacture of claim 3, further comprising instructions for receiving a data point label prior probability for the seed document and unlabeled documents, wherein for each iteration of the calculations the data point label prior probability is adjusted according to an estimate of a data point class membership probability.

5. The article of manufacture of claim I, wherein the unlabeled documents are customer complaints, and further comprising linking product changes with customer complaints.

6. The article of manufacture of claim I, wherein the unlabeled documents are invoices.

7. An article of manufacture comprising:

a program storage medium readable by a computer, where the medium tangibly embodies one or more programs of instructions executable by a computer to perform a method for separating documents, the One or more programs of instructions comprising;

instructions for receiving labeled data;

instructions for receiving a sequence of unlabeled documents;

instructions for adapting probabilistic classification rules using transduction based on the labeled data and the unlabeled documents;

instructions for updating weights used for document separation according to the probabilistic classification rules;

instructions for determining locations of separations in the sequence of documents;

instructions for outputting indicators of the determined locations of the separations in the sequence to at least one of a user, another system, and another process; and

instructions for flagging the documents with codes, the codes correlating to the indicators.

8. An article of manufacture comprising:

a program storage medium readable by a computer, where the medium tangibly embodies one or more programs of instructions executable by a computer to perform a method of document searching, the one or more programs of instructions comprising;

instructions for receiving a search query;

instructions for retrieving documents based on the search query;

instructions for outputting the documents;

instructions for receiving user-entered labels for at least some of the documents, the labels being indicative of a relevance of the document to the search query;

instructions for training a classifier based on the search query and the user-entered labels;

instructions for performing a document classification technique on the documents using the classifier for reclassifying the documents; and

instructions for outputting identifiers of at least some of the documents based on the classification thereof.

9. The article of manufacture of claim 8, wherein the document classification technique includes a transductive process.

10. The article of manufacture of claim 9, wherein the classifier is a transductive classifier, and further comprising training the transductive classifier through iterative calculation using at least one predetermined cost factor, the search query, and the documents, wherein for each iteration of the calculations the cost factor is adjusted as a function of an expected label value, and using the trained classifier to classify the documents.

11. The article of manufacture of claim 10, further comprising receiving a data point label prior probability for the search query and documents, wherein for each iteration of the calculations the data point label prior probability is adjusted according to an estimate of a data point class membership probability.

12. The article of manufacture of claim 8, wherein the document classification technique includes a support vector machine process.

13. The article of manufacture of claim 8, wherein the document classification technique includes a maximum entropy discrimination process.

14. The article of manufacture of claim 8, wherein the reclassified documents are output, those documents having a highest confidence being output first.

15. A system comprising:

a processor; and

a program storage medium and/or memory storing one or more programs of instructions executable by the processor to perform a method of data classification, the one or more programs of instructions comprising;

instructions for receiving at least one labeled seed document;

instructions for receiving unlabeled documents;

instructions for receiving at least one predetermined cost factor;

instructions for training a transductive classifier using the at least one predetermined cost factor, the at least one seed document, and the unlabeled documents;

instructions for classifying the unlabeled documents having a confidence level above a predefined threshold into a plurality of categories using the classifier;

instructions for reclassifying at least some of the categorized documents previously categorized by a different classifier into the categories using the classifier; and

instructions for outputting identifiers of the categorized documents to at least one of a user, another system, and another process.

16. The system of claim 15, further comprising moving an unlabeled document having a confidence level below the predefined threshold into one or more new categories.

17. The system of claim 15, and further comprising training the transductive classifier through iterative calculation using at least one predetermined cost factor, the at least one seed document, and the unlabeled documents, wherein for each iteration of the calculations the cost factor is adjusted as a function of an expected label value, and using the trained classifier to classify the unlabeled documents.

18. The system of claim 17, further comprising receiving a data point label prior probability for the seed document and unlabeled documents, wherein for each iteration of the calculations the data point label prior probability is adjusted according to an estimate of a data point class membership probability.

19. The system of claim 15, wherein the unlabeled documents are customer complaints, and further comprising linking product changes with customer complaints.

20. The system of claim 15, wherein the unlabeled documents are invoices.

21. A system comprising:

a processor; and

a program storage medium and/or memory storing one or more programs of instructions executable by the processor to perform a method for separating documents, the one or more programs of instructions comprising;

instructions for receiving labeled data;

instructions for receiving a sequence of unlabeled documents;

instructions for adapting probabilistic classification rules using transduction based on the labeled data and the unlabeled documents;

instructions for updating weights used for document separation according to the probabilistic classification rules;

instructions for determining locations of separations in the sequence of documents;

instructions for outputting indicators of the determined locations of the separations in the sequence to at least one of a user, another system, and another process; and

instructions for flagging the documents with codes, the codes correlating to the indicators.

22. A system comprising:

a processor; and

a program storage medium and/or memory storing one or more programs of instructions executable by the processor to perform a method of document searching, the one or more programs of instructions comprising;

instructions for receiving a search query;

instructions for retrieving documents based on the search query;

instructions for outputting the documents;

instructions for receiving user-entered labels for at least some of the documents, the labels being indicative of a relevance of the document to the search query;

instructions for training a classifier based on the search query and the user-entered labels;

instructions for performing a document classification technique on at least some of the documents using the classifier for reclassifying the at least some of the documents; and

instructions for outputting identifiers of the at least some of the documents based on the classification thereof.

23. The system of claim 22, wherein the document classification technique includes a transductive process.

24. The system of claim 23, wherein the classifier is a transductive classifier, and further comprising training the transductive classifier through iterative calculation using at least one predetermined cost factor, the search query, and the documents, wherein for each iteration of the calculations the cost factor is adjusted as a function of an expected label value, and using the trained classifier to classify the documents.

25. The system of claim 24, further comprising receiving a data point label prior probability for the search query and documents, wherein for each iteration of the calculations the data point label prior probability is adjusted according to an estimate of a data point class membership probability.

26. The system of claim 22, wherein the document classification technique includes a support vector machine process.

27. The system of claim 22, wherein the document classification technique includes a maximum entropy discrimination process.

28. The system of claim 22, wherein the reclassified documents are output, those documents having a highest confidence being output first.