System amd method of analyzing text using dynamic centering resonance analysis

• US 20040243388A1
• Filed: 10/21/2003
• Published: 12/02/2004
• Est. Priority Date: 06/03/2002
• Status: Active Grant

• Current Assignee: Arizona Board of Regents (University of Arizona)
• Original Assignee: Arizona Board of Regents (University of Arizona)
• Inventors: Dooley, Kevin J., Corman, Steven R.
• Granted Patent: US 7,295,967 B2
• Time in Patent Office: Days
• US Class Current: 704/1
• CPC Class Codes: G06F 40/194 Calculation of difference between files
• CPC Class Codes: G06F 40/205 Parsing
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates

Abstract

A document analysis tool provides a 3-dimensional network representation of temporal evolution of text in a document. First and second portions of text are extracted from one or more documents. The texts are parsed to identify noun phrases. The noun phrases are stemmed to remove word variations. Words from the first portion of the text are linked together to form a first layer in the network. Words from the second portion of the text are linked together to form a second layer in the network. The second layer exists in a different temporal plane from the first layer. Words that are common between first and second layers are linked together between the first and second layers to provide a visual and analytical representation of relationship between the first and second portions of the text.

Claims

1. A method of analyzing temporal changes in text, comprising:

extracting first and second portions of the text;

linking words of the first portion of the text;

linking words of the second portion of the text;

forming a first layer of a network from the linked words of the first portion of the text;

forming a second layer of the network from the linked words of the second portion of the text, wherein the second layer exists in a different temporal plane from the first layer; and

linking words common between first and second layers to provide a representation of relationship between the first and second portions of the text.

2. The method of claim.1, wherein the words of the first and second portions of the text comprise noun phrases.

3. The method of claim 1, further including parsing the words of the first portion of the text.

4. The method of claim 3, wherein the step of parsing the words of the first portion of the text include:

performing lexical analysis on the words of the first portion of the text;

stemming the words of the first portion of the text;

removing stop words from the first portion of the text; and

indexing the words of the first portion of the text by assigning a weighting function.

5. The method of claim 4, wherein the step of indexing the words of the first portion of the text include computing a centrality for the words of the first portion of the text.

6. The method of claim 1, wherein the step of linking words common between first and second layers include connecting a first word from the first layer to the same first word found in the second layer to analyze a point of commonality between the first and second portions of the text.

7. The method of claim 1, further including the step of forming a discourse chain between words that occur in the first and second layers with words that occur in a third layer.

8. The method of claim 1 wherein the text comprises recorded speech.

9. A method of representing evolving text, comprising:

extracting first noun phrases from a first text and second noun phrases from a second text;

linking the first noun phrases to form a first layer of a network;

linking the second noun phrases to form a second layer of the network; and

connecting common noun phrases between first and second layers to provide a representation of relationship between the first and second text.

10. The method of claim 9, wherein the second layer of the network exists in a different temporal plane from the first layer of the network.

11. The method of claim 9, further including parsing the noun phrases in the first text.

12. The method of claim 11, wherein the step of parsing the noun phrases in the first text include:

performing lexical analysis on the noun phrases in the first text;

stemming the noun phrases in the first text;

removing stop words from the first text; and

indexing the noun phrases in the first text by assigning a weighting function.

13. The method of claim 12, wherein the step of indexing the noun phrases in the first text includes computing a centrality for the noun phrases in the first text.

14. The method of claim 9, wherein the step of connecting common noun phrases include assign a link between a first noun phrases from the first layer to the same noun phrases found in the second layer to analyze a point of commonality between the first and second text.

15. The method of claim 9, further including the step of forming a discourse chain between words that occur in the first and second layers with words that occur in a third layer.

16. A method of providing a 3-dimensional network representation of temporal evolution of text in a document, comprising:

extracting first and second portions of the text;

linking words of the first portion of the text to form a first layer in the network;

linking words of the second portion of the text to form a second layer in the network, wherein the second layer exists in a different temporal plane from the first layer; and

linking words common between first and second layers to provide a representation of relationship between the first and second portions of the text.

17. The method of claim 16, wherein the words of the first portion of the text comprise noun phrases.

18. The method of claim 16, further including parsing the words of the first portion of the text.

19. The method of claim 18, wherein the step of parsing the words of the first portion of the text include:

performing lexical analysis on the words of the first portion of the text;

stemming the words of the first portion of the text;

removing stop words from the first portion of the text; and

indexing the words of the first portion of the text by assigning a weighting function.

20. The method of claim 19, wherein the step of indexing the words of the first portion of the text include computing a centrality for the words of the first portion of the text.

21. The method of claim 16, wherein the step of linking words common words between first and second layers include connecting a first word from the first layer to the same first word found in the second layer to analyze a point of commonality between the first and second portions of the text.

22. The method of claim 16, further including the step of forming a discourse chain between words that occur in the first and second layers with words that occur in a third layer.

23. The method of claim 16, wherein the first portion of the text exists in a first document and the second portion of the text exists in a second document.

24. A computer system for providing a 3-dimensional network representation of temporal evolution of text in a document, comprising:

means for extracting first and second portions of the text;

means for linking words of the first portion of the text to form a first layer in the network;

means for linking words of the second portion of the text to form a second layer in the network, wherein the second layer exists in a different temporal plane from the first layer; and

means for linking words common between first and second layers to provide a representation of relationship between the first and second portions of the text.

25. The computer system of claim 24, wherein the words of the first portion of the text comprise noun phrases.

26. The computer system of claim 24, further including means for parsing the words of the first portion of the text.

27. The computer system of claim 26, wherein the means for parsing include:

means for performing lexical analysis on the words of the first portion of the text;

means for stemming the words of the first portion of the text;

means for removing stop words from the first portion of the text; and

means for indexing the words of the first portion of the text by assigning a weighting function.

28. The computer system of claim 27, wherein the means for indexing include means for computing a centrality for the words of the first portion of the text.

29. The computer system of claim 24, wherein the means for linking words common between first and second layers include means for connecting a first word from the first layer to the same first word found in the second layer to analyze a point of commonality between the first and second portions of the text.