Module for creating a language neutral syntax representation using a language particular syntax tree

• US 7,596,485 B2
• Filed: 06/30/2004
• Issued: 09/29/2009
• Est. Priority Date: 06/30/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Ringger, Eric K., Campbell, Richard G.
• Primary Examiner(s): Dorvil; Richemond
• Assistant Examiner(s): Adesanya; Olujimi A
• Application Number:
  US10/882,697
• Publication Number:
  US 20060004563A1
• Time in Patent Office: 1,917 Days
• Field of Search: 704/2, 704/4, 704/9
• US Class Current: 704/9
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A method or module for creating an Language Neutral Syntax (LNS) representation of a sentence from a language particular syntax representation such as found in the Penn Treebank for use by different applications. The method or module includes a node generator configured to create hierarchical and dependent nodes using phrasal and constituent nodes of the language particular syntax. A node dependency generator is configured to create an unordered hierarchical dependency structure for the hierarchical and dependent nodes using a semantic relation between the hierarchical and dependent nodes derived from the language particular syntax.

Claims

1. A system including computer executable instructions on a computer storage media and the computer executable instructions being executed by a processing unit to implement one or more components comprising:

a node generator configured to receive a parsed natural language sentence or phrase and recover phrasal and constituent nodes and grammatical tags for the phrasal and constituent nodes of the parsed natural language sentence or phrase and the node generator uses a head analysis component to analyze the phrasal and constituent nodes and grammatical tags of the parsed natural language sentence or phrase to generate hierarchical and dependent nodes of a language neutral representation of the parsed natural language sentence or phrase; and

a node dependency generator configured to receive the hierarchical and dependent nodes and the grammatical tags of the parsed natural language sentence and create an iterative dependency structure including a preliminary dependency structure including one or more unlabeled dependencies to one or more semantic heads and a secondary dependency structure including semantic or grammatical labels replacing the one or more unlabeled dependencies to generate an unordered hierarchical dependency structure for the hierarchical and dependent nodes and the semantic or grammatical labels representing a language neutral relation between the hierarchical and dependent nodes different from the grammatical tags of the parsed natural language sentence or phrase using a semantic relation between the hierarchical and dependent nodes derived from the grammatical tags of the parsed natural language sentence or phrase.

2. The system of claim 1 wherein the node generator uses a morphological analysis component to analyze the phrasal and constituent nodes and the grammatical tags of the parsed natural language sentence or phrase to generate the hierarchical and dependent nodes.

3. The system of claim 1 wherein the node generator creates a head node for each of the phrasal nodes of the parsed natural language sentence or phrase.

4. The system of claim 1 wherein the node generator derives a base form of a word or element for one or more constituent nodes of the parsed natural language sentence or phrase and one or more corresponding grammatical features or attributes of the base form are represented in one or more hierarchical or head nodes of the unordered hierarchical dependency structure corresponding to one or more dependent nodes for the one or more constituent nodes.

5. The system of claim 1 wherein the node generator provides a node derivation for an auxiliary or complex verb construction including a main verb and aspectual and voice features.

6. The system of claim 1 wherein the node generator provides a derivation to represent scope and logical relation of constituents of a compound and coordinating noun phrase in the unordered hierarchical dependency structure.

7. The system of claim 1 wherein the node dependency generator is configured to create the unordered hierarchical dependency structure to represent logical scope and relation of constituents of a noun phrase or verb phrase using the grammatical tags and syntax of the parsed natural language sentence or phase and scope assignment criteria.

8. The system of claim 1 wherein the node dependency generator creates abstract nodes to represent tense, comparison, superlative, negation, and unspecified arguments not represented by individual words or phrases in the parsed natural language sentence or phrase.

9. The system of claim 1 wherein the parsed natural language sentence or phrase is a Penn Treebank annotation.

10. A method for creating an unordered hierarchical dependency structure to represent a natural language sentence or phrase comprising computer executable instructions on a computer storage media and executable by a processing unit to implement the steps comprising:

receiving a parsed natural language sentence or phrase;

recovering phrasal and constituent nodes and grammatical tags of the parsed natural language sentence or phrase;

deriving hierarchical and dependent nodes using the phrasal and constituent nodes of the parsed natural language sentence or phrase;

creating a preliminary dependency structure including one or more temporary dependencies to one or more hierarchical nodes using the grammatical tags of the parsed natural language sentence or phrase; and

replacing the one or more temporary dependencies with semantic or grammatical labels representing logical relation between one or more dependent nodes and the one or more hierarchical nodes and creating an unordered hierarchical dependency structure for the hierarchical and dependent nodes where the semantic or grammatical labels represent a language neutral relation between the hierarchical and dependent nodes different from the grammatical tags of the parsed natural language sentence or phrase.

11. The method of claim 10 wherein the step of creating the unordered hierarchical dependency structure comprises:

lemmatizing constituents of terminal nodes of the unordered hierarchical dependency structure and assigning attributes to a nonterminal node corresponding to the terminal node for the lemmatized constituent to provide a language neutral representation.

12. The method of claim 10 and further comprising the step of:

performing a head analysis of the parsed natural language sentence or phrase; and

creating the unordered hierarchical dependency structure based upon a semantic relation between constituents of the parsed natural language sentence or phrase derived using the head analysis.

13. The method of claim 12 wherein hierarchical head nodes are created for one or more of the constituent nodes of the parsed natural language sentence or phrase based upon the head analysis.

14. The method of claim 10 and further comprising the step of:

creating abstract nodes for the unordered hierarchical dependency structure to represent negation, tense or modal operator.

15. The method of claim 10 and further comprising the step of:

performing a scope analysis to create the unordered hierarchical dependency structure to represent logical and grammatical relation of constituents of a noun or verb phrase.

16. The method of claim 10 wherein the parsed natural language sentence or phrase is a data string and comprising:

deserializing the data string to recover the phrasal and constituent nodes and the grammatical tags of parsed natural language sentence or phrase.

17. The method of claim 10 and comprising:

parsing the natural language sentence to provide the phrasal and constituent nodes for the natural language sentence or phrase and the grammatical tags for the phrasal and constituent nodes.