Browse Prior Art Database

Resolution of Difficult Pronouns Using the ROSS Method

IP.com Disclosure Number: IPCOM000239557D
Publication Date: 2014-Nov-15
Document File: 277 page(s) / 3M

Publishing Venue

The IP.com Prior Art Database

Related Documents

8,407,169: PATENT

Abstract

A new natural language understanding method for disambiguation of difficult pronouns is described. Difficult pronouns are those pronouns for which a level of world or domain knowledge is needed in order to perform anaphoral or other types of resolution. Resolution of difficult pronouns may in some cases require a prior step involving the application of inference to a situation that is represented by the natural language text. A general method is described: it performs entity resolution and pronoun resolution. An extension to the general pronoun resolution method performs inference as an embedded commonsense reasoning method. The general method and the embedded method utilize features of the ROSS representational scheme; in particular the methods use ROSS ontology classes and the ROSS situation model.

ROSS ontology classes include the object frame class and the behavior class. The ROSS behavior class defines associations among a set of objects that have attribute-based state descriptions and nested behaviors. In addition to the classes of the ontology, the methods use several working memory data structures, including a spanning information data structure and a pronoun feature set structure. The ROSS internal situation model (or “instance model”) is an instance of a meaning representation; it is a spatial/temporal representation of declarative information from the input natural language text.

A new representational formalism called “semantic normal form” (SNF) is also introduced. This is a specification at the abstract level for a set of data structures that are used to store the syntax and content of input natural language text that has been transformed and augmented with semantic role and other information. It is an intermediate form of the input information that is processable by a semantic NLU engine that implements the pronoun resolution method.

The overall method is a working solution that solves the following Winograd schemas: a) trophy and suitcase, b) person lifts person, c) person pays detective, and d) councilmen and demonstrators.

Many of the features described in this paper have been productized - the functionality is implemented in an NLU system that is available for use via a RESTful API server (currently English-only).

Supporting documentation is appended to this paper, including "ROSS User’s Guide and Reference Manual (Version 1.0)", and "Introduction to ROSS: A New Representational Scheme".

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 3% of the total text.

Page 01 of 277

 Resolution of Difficult Pronouns Using the ROSS Method

Glenn R. Hofford, Software Engineering Concepts, Inc.

Date of Publication: 11/14/2014

(Version 1.0)

Copyright © 2014 Glenn R. Hofford



Page 02 of 277

Abstract:

A new natural language understanding method for disambiguation of difficult pronouns is described. Difficult pronouns are those pronouns for which a level of world or domain knowledge is needed in order to perform anaphoral or other types of resolution. Resolution of difficult pronouns may in some cases require a prior step involving the application of inference to a situation that is represented by the natural language text. A general method is described: it performs entity resolution and pronoun resolution. An extension to the general pronoun resolution method performs inference as an embedded commonsense reasoning method. The general method and the embedded method utilize features of the ROSS representational scheme; in particular the methods use ROSS ontology classes and the ROSS situation model.

ROSS ontology classes include the object frame class and the behavior class. The ROSS behavior class defines associations among a set of objects that have attribute-based state descriptions and nested behaviors. In addition to the classes of the ontology, the methods use several working memory data structures, including a spanning information data structure and a pronoun feature set structure. The ROSS internal situation model (or "instance model") is an instance of a meaning representation; it is a spatial/temporal representation of declarative information from the input natural language text.

A new representational formalism called "semantic normal form" (SNF) is also introduced. This is a specification at the abstract level for a set of data structures that are used to store the syntax and content of input natural language text that has been transformed and augmented with semantic role and other information. It is an intermediate form of the input information that is processable by a semantic NLU engine that implements the pronoun resolution method.

The overall method is a working solution that solves the following Winograd schemas: a) trophy and suitcase, b) person lifts person, c) person pays detective, and d) councilmen and demonstrators.

Many of the features described in this paper have been productized - the functionality is implemented in an NLU system that is available for use via a RESTful API server (currently English-only).

Contact: glennhofford(at)gmail.com

2



Page 03 of 277

Table of Contents


1. Introduction and Background ............................................................................................... 6


1.1. The ROSS Representational Method ................................................................................ 7


1.2. Background: Winograd Schema Challenge ...................................................................... 7


2. Main Concepts ...........................................