Method, system, and program for filtering content using neural networks

• US 6,633,855 B1
• Filed: 01/06/2000
• Issued: 10/14/2003
• Est. Priority Date: 01/06/2000
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Auvenshine, John Jason
• Primary Examiner(s): Patel, Ramesh
• Assistant Examiner(s): Hirl, Joseph P.
• Application Number:
  US09/478,925
• Time in Patent Office: 1,377 Days
• Field of Search: 706/15, 706/25, 706/16
• US Class Current: 706/15
• CPC Class Codes: G06F 16/00 Information retrieval; Database structures therefor; File system structures therefor

Abstract

Disclosed is a method, system, and program for filtering a data object for content deemed unacceptable by a user. A data object requested by a viewer program is received. The data object is processed to determine predefined language statements. Information on the determined language statements is inputted into a neural network to produce an output value. A determination is then made as to whether the output value indicates that the data object is unacceptable. Viewer program access to the data object is inhibited upon determining that the data object is unacceptable.

Claims

1. A method for filtering a data object for content deemed unacceptable by an administrator for a protected user, comprising:

receiving a data object requested by a viewer program;

processing the data object to determine predefined language statements;

inputting information on the determined language statements into a neural network to produce an output value,

determining whether the output value indicates that the data object is unacceptable; and

inhibiting the viewer program access to the data object upon determining that the data object is unacceptable.

2. The method of claim 1, wherein the data object is acceptable if the output value is within a first predefined range of output values and the data object is unacceptable if the output value is within a second predefined range of output values, wherein determining whether the output value indicates that the data object is unacceptable comprises determining whether the output value falls within the second predefined range, further comprising forwarding the data object to the viewer program to display upon determining that the output value falls within the first predefined range.

3. The method of claim 1, further comprising:

determining whether the output value indicates that the data object requires further consideration;

logging information on the data object;

receiving user input on a rating indicating the acceptability of the data object; and

training the neural network to process the logged data object as input to produce the received rating as output.

4. The method of claim 1, wherein processing the data object comprises determining a weighted number of occurrences of predefined language statements, wherein the neural network includes an input node for each predefined language statement and wherein inputting the information on the determined language statements comprises inputting the weighted number of occurrences of each predefined language statement on the input node corresponding to the predefined language statement.

5. The method of claim 4, further comprising:

changing the number of predefined language statements by adding and removing predefined language statements;

altering the neural network to change the number of input nodes such that there is one input node for each predefined language statement; and

training the neural network to produce output values based on input data objects.

6. The method of claim 4, wherein the predefined language statements include unacceptable language statements and content language statements, further comprising:

determining the proximity of predefined content and unacceptable language statements; and

inputting information on the proximity of the predefined language statements at input nodes, whereby the output value is based on the number of occurrences of unacceptable and content language statements and the proximity of content and unacceptable language statements.

7. The method of claim 1, wherein each language statement is a member of a class of language statements, wherein processing the data object comprises determining a weighted number of occurrences of the predefined language statements for each class, wherein the neural network includes an input node for each class of predefined language statements, and wherein inputting the information on the determined language statements comprises inputting the weighted number of occurrences of predefined language statements for each class on the input node corresponding to the class.

8. The method of claim 1, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the entire data object.

9. The method of claim 1, further comprising determining language statements in the data object that are unacceptable, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the unacceptable language statements and allowing access to language statements not determined to be unacceptable.

10. The method of claim 1, wherein the data object is one of multiple packets that combined form a document requested by the viewer.

11. The method of claim 10, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the entire document of which the packet is a part.

12. The method of claim 10, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to only those packets determined to be unacceptable and allowing the viewer access to packets not determined to be unacceptable.

13. The method of claim 1, wherein the data object comprises a document, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the entire document.

14. The method of claim 1, wherein the data object comprises a document, further comprising determining language statements in the document that are unacceptable, wherein blocking the viewer from accessing the document upon determining that the document is unacceptable comprises blocking access to only those language statements in the document determined to be unacceptable and allowing access to language statements in the document not determined to be unacceptable.

15. A system for filtering a data object for content deemed unacceptable by an administrator for a protected user, comprising:

means for receiving a data object requested by a viewer program;

means for processing the data object to determine predefined language statements;

means for inputting information on the determined language statements into a neural network to produce an output value,

means for determining whether the output value indicates that the data object is unacceptable; and

means for inhibiting the viewer program access to the data object upon determining that the data object is unacceptable.

16. The system of claim 15, wherein the data object is acceptable if the output value is within a first predefined range of output values and the data object is unacceptable if the output value is within a second predefined range of output values, wherein the means for determining whether the output value indicates that the data object is unacceptable comprises means for determining whether the output value falls within the second predefined range, and further comprising means for forwarding the data object to the viewer program to display upon determining that the output value falls within the first predefined range.

17. The system of claim 15, further comprising:

means for determining whether the output value indicates that the data object requires further consideration;

means for logging information on the data object;

means for receiving user input on a rating indicating the acceptability of the data object; and

means for training the neural network to process the logged data object as input to produce the received rating as output.

18. The system of claim 15, wherein the means for processing the data object comprises means for determining a weighted number of occurrences of predefined language statements, wherein the neural network includes an input node for each predefined language statement and wherein inputting the information on the determined language statements comprises means for inputting the weighted number of occurrences of each predefined language statement on the input node corresponding to the predefined language statement.

19. The system of claim 18, further comprising:

means for changing the number of predefined language statements by adding and removing predefined language statements;

means for altering the neural network to change the number of input nodes such that there is one input node for each predefined language statement; and

means for training the neural network to produce output values based on input data objects.

20. The system of claim 18, wherein the predefined language statements include unacceptable language statements and content language statements, further comprising:

means for determining the proximity of predefined content and unacceptable language statements; and

means for inputting information on the proximity of the predefined language statements at input nodes, whereby the output value is based on the number of occurrences of unacceptable and content language statements and the proximity of content and unacceptable language statements.

21. The system of claim 15, wherein each language statement is a member of a class of language statements, wherein the means for processing the data object comprises means for determining a weighted number of occurrences of the predefined language statements for each class, wherein the neural network includes an input node for each class of predefined language statements, and wherein the means for inputting the information on the determined language statements comprises means for inputting the weighted number of occurrences of predefined language statements for each class on the input node corresponding to the class.

22. The system of claim 15, wherein the means for inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises means for blocking access to the entire data object.

23. The system of claim 15, further comprising means for determining language statements in the data object that are unacceptable, wherein the means for inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises means for blocking access to the unacceptable language statements and allowing access to language statements not determined to be unacceptable.

24. The system of claim 15, wherein the data object is one of multiple packets that combined form a document requested by the viewer.

25. The system of claim 24, wherein the means for inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises means for blocking access to the entire document of which the packet is a part.

26. The system of claim 24, wherein the means for inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises means for blocking access to only those packets determined to be unacceptable and allowing the viewer access to packets not determined to be unacceptable.

27. The system of claim 15, wherein the data object comprises a document, wherein the means for inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises means for blocking access to the entire document.

28. The system of claim 15, wherein the data object comprises a document, further comprising means for determining language statements in the document that are unacceptable, wherein blocking the viewer from accessing the document upon determining that the document is unacceptable comprises means for blocking access to only those language statements in the document determined to be unacceptable and allowing access to language statements in the document not determined to be unacceptable.

29. An article of manufacture for filtering a data object for content deemed unacceptable by an administrator for a protected user, the article of manufacture comprising computer readable storage media including at least one computer program embedded therein that causes a computer to perform:

receiving a data object requested by a viewer program;

processing the data object to determine predefined language statements;

inputting information on the determined language statements into a neural network to produce an output value,

determining whether the output value indicates that the data object is unacceptable; and

inhibiting the viewer program access to the data object upon determining that the data object is unacceptable.

30. The article of manufacture of claim 29, wherein the data object is acceptable if the output value is within a first predefined range of output values and the data object is unacceptable if the output value is within a second predefined range of output values, wherein determining whether the output value indicates that the data object is unacceptable comprises determining whether the output value falls within the second predefined range, further comprising forwarding the data object to the viewer program to display upon determining that the output value falls within the first predefined range.

31. The article of manufacture of claim 29, further comprising:

determining whether the output value indicates that the data object requires further consideration;

logging information on the data object;

receiving user input on a rating indicating the acceptability of the data object; and

training the neural network to process the logged data object as input to produce the received rating as output.

32. The article of manufacture of claim 29, wherein processing the data object comprises determining a weighted number of occurrences of predefined language statements, wherein the neural network includes an input node for each predefined language statement and wherein inputting the information on the determined language statements comprises inputting the weighted number of occurrences of each predefined language statement on the input node corresponding to the predefined language statement.

33. The article of manufacture of claim 32, further comprising:

changing the number of predefined language statements by adding and removing predefined language statements;

altering the neural network to change the number of input nodes such that there is one input node for each predefined language statement; and

training the neural network to produce output values based on input data objects.

34. The article of manufacture of claim 32, wherein the predefined language statements include unacceptable language statements and content language statements, further comprising:

determining the proximity of predefined content and unacceptable language statements; and

inputting information on the proximity of the predefined language statements at input nodes, whereby the output value is based on the number of occurrences of unacceptable and content language statements and the proximity of content and unacceptable language statements.

35. The article of manufacture of claim 29, wherein each language statement is a member of a class of language statements, wherein processing the data object comprises determining a weighted number of occurrences of the predefined language statements for each class, wherein the neural network includes an input node for each class of predefined language statements, and wherein inputting the information on the determined language statements comprises inputting the weighted number of occurrences of predefined language statements for each class on the input node corresponding to the class.

36. The article of manufacture of claim 29, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the entire data object.

37. The article of manufacture of claim 29, further comprising determining language statements in the data object that are unacceptable, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the unacceptable language statements and allowing access to language statements not determined to be unacceptable.

38. The article of manufacture of claim 29, wherein the data object is one of multiple packets that combined form a document requested by the viewer.

39. The article of manufacture of claim 38, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the entire document of which the packet is a part.

40. The article of manufacture of claim 38, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to only those packets determined to be unacceptable and allowing the viewer access to packets not determined to be unacceptable.

41. The article of manufacture of claim 29, wherein the data object comprises a document, wherein inhibiting the viewer program access to the data object upon determining that the data object is unacceptable comprises blocking access to the entire document.

42. The article of manufacture of claim 29, wherein the data object comprises a document, further comprising determining language statements in the document that are unacceptable, wherein blocking the viewer from accessing the document upon determining that the document is unacceptable comprises blocking access to only those language statements in the document determined to be unacceptable and allowing access to language statements in the document not determined to be unacceptable.