Identifying semantic positions of portions of a text

• US 20080103760A1
• Filed: 10/27/2006
• Published: 05/01/2008
• Est. Priority Date: 10/27/2006
• Status: Active Grant

• Current Assignee: Micro Focus LLC (Open Text Corporation)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Kirshenbaum, Evan R.
• Granted Patent: US 8,359,190 B2
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A computer executable method includes determining semantic positions in a semantic space of portions of a text. Once the semantic positions of portions of the text have been determined, an action is performed with respect to the text based on the determined semantic positions.

Claims

1. A method executable in a computer, comprising:

identifying senses of words appearing in portions of a text;

determining, based on the identified senses of the words in the portions of the text, semantic positions in a semantic space of the portions of the text; and

performing an action with respect to the text based on the determined semantic positions.

3. The method of claim 1, further comprising accessing a position-based feature dictionary to identify semantic positions of the senses, wherein determining the semantic positions of the portions is based on the identified semantic positions of the senses.

4. The method of claim 3, wherein accessing the position-based feature dictionary comprises accessing the position-based feature dictionary maps word forms to corresponding senses, the position-based feature dictionary to also identify semantic positions for the respective senses in the position-based feature dictionary.

5. The method of claim 1, wherein performing the action comprises identifying one or more topics of the text.

6. The method of claim 5, further comprising providing a list of topics in which the topics are associated with corresponding decision procedures for assigning a degree of likelihood to a semantic position, wherein identifying the one or more topics comprises applying the decision procedures associated with the topics in the list to a corresponding semantic position of a portion of the text to determine a degree of likelihood that the corresponding topic should be identified.

7. The method of claim 6, wherein applying a decision procedure comprises one of identifying a semantic position of the corresponding topic, identifying a region of the semantic space associated with the corresponding topic, applying a Boolean function, applying a real-valued function, applying an integer-valued function, applying a discrete-valued function, and applying a computer program.

8. The method of claim 1, further comprising determining semantic positions of segments of the text, wherein at least one segment includes multiple ones of the portions, and wherein determining the semantic position of the at least one segment is based on the semantic positions of the portions in the at least one segment.

9. The method of claim 1, wherein determining the semantic positions of the portions comprises:

determining the semantic position of a first one of the portions; and

iteratively determining the semantic positions of remaining portions using semantic positions of previous portions.

10. The method of claim 9, further comprising:

for each portion i following the first portion,

computing a second semantic position of the portion i based on a current semantic position and semantic portions of words in the portion i, and

combining the current semantic position with the second semantic position to determine a new current semantic position to be used in determining a semantic position of a subsequent portion.

11. The method of claim 10, wherein combining the current semantic position and the second semantic position of each portion i comprises performing a weighted avenge of the current semantic position and the second semantic position.

12. The method of claim 1, wherein at least some words in the text are associated with plural senses, the method further comprising:

identifying, for at least one portion, a set of senses from among plural candidate sets of senses; and

identifying semantic positions of the senses in the set,

wherein determining the semantic position for the at least one portion is based on the semantic positions of the senses in the set.

13. The method of claim 12, further comprising performing syntactic analysis to remove senses as possibilities for the at least one portion.

14. The method of claim 12, wherein identifying the set of senses from among the plural candidate sets of senses comprises selecting the set of senses according to fitness measures of corresponding candidate sets.

15. The method of claim 14, wherein selecting the set of senses according to the fitness measures uses one of a greedy approach, a genetic algorithm, an analytic approach, and an exhaustive search.

21. The method of claim 1, wherein the text portions comprise one of sentences, paragraphs, and sections.

22. The method of claim 1, wherein identifying the senses of the words comprises identifying sets of senses of the words appearing in the corresponding text portions, wherein each of the sets of senses is a most compatible set of senses from among plural candidate sets of senses for the corresponding text portion,

wherein determining the semantic positions of the text portions is based on the corresponding sets of senses.

2. (canceled)

16. A method executable in a computer, comprising:

identifying senses of words appearing in corresponding portions of a text item;

determining semantic positions of the senses in a semantic space; and

identifying one or more topics associated with the text item based on the determined semantic positions of the senses.

17. The method of claim 16, further comprising determining semantic positions in the semantic space of the portions of the text item based on the semantic positions of the senses,

wherein identifying the one or more topics is based on the determined semantic positions of the portions.

18. The method of claim 16, wherein identifying the one or more topics is based on a vote to select from among topics of corresponding portions of the text item.

19. The method of claim 16, wherein identifying the one or more topics is based on identifying the one or more topics associated with an amount of the text item greater than a predefined threshold.

23. The method of claim 17, wherein identifying the senses of words appearing in the corresponding portions of the text item comprises identifying sets of senses of the words appearing in the corresponding portions, wherein each of the sets of senses is a most compatible set of senses from among plural candidate sets of senses for the corresponding portion of the text item,

wherein determining the semantic positions in the semantic space of the portions of the text item is based on the corresponding sets of senses.

24. The method of claim 17, wherein determining the semantic positions of the senses in the semantics space is based on accessing a position-based feature dictionary that maps word forms to corresponding senses, the feature-based position dictionary to also identify with each sense of each word form a corresponding semantic position in the semantics base.

20. Instructions on a computer-usable medium that when executed cause a computer to:

identify sets of senses of words for corresponding portions in a text;

determine, based on the identified sets of senses of the words, semantic positions in a semantic space of the corresponding portions of the text; and

perform an action with respect to the text based on the determined semantic positions.

25. The instructions of claim 20, wherein each of the sets of senses is a most compatible set of senses from among plural candidate sets of senses for the corresponding text portion,

wherein determining the semantic positions of the text portions is based on the corresponding sets of senses.

26. The instructions of claim 20, which when executed cause the computer to further:

access a position-based feature dictionary to identify semantic positions of the senses, wherein determining the semantic positions of the text portions is based on the identified semantic positions of the senses, wherein the position-based feature dictionary maps word forms to corresponding senses, the position-based feature dictionary to also identify semantic positions for respective senses in the position-based feature dictionary.