# US 2013/0086059 A1 · Method for Discovering Key Entities and Concepts in Data

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2013/0086059 A1
- Document ID: US20130086059A1
- Status: Published application
- Kind: A1
- Application number: 13/251,322
- Filed: 2011-10-03
- Published: 2013-04-04
- Earliest claimed: 2011-10-03
- Assignees: Nuance Communications, Inc.
- Inventors: Rajesh Balchandran; Leonid Rachevsky; Bhuvana Ramabhadran
- Web page: https://patentgrep.com/patent/US20130086059A1

## CPC

- G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning}
- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06F 40/30: Semantic analysis
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method of automatically processing text data is described. An initial set of data tags is developed that characterize text data in a text database. Higher order entities are determined which are characteristic of patterns in the data tags. Then the text data is automatically tagged based on the higher order entities.

## Claims

### Claim 1

A method of automatically processing text data comprising:
developing an initial set of data tags characterizing text data in a text database;
automatically determining higher order entities characteristic of patterns in the data tags; and
automatically tagging the text data based on the higher order entities.

### Claim 2 (depends on 1)

A method according to claim 1, further comprising:
iteratively repeating the determining and tagging steps a plurality of times.

### Claim 3 (depends on 1)

A method according to claim 1, further comprising:
training a text classifier statistical model based on the tags and text data.

### Claim 4 (depends on 1)

A method according to claim 1, wherein the automatically determining higher order entities includes using n-gram models.

### Claim 5 (depends on 4)

A method according to claim 4, wherein the n-gram models are limited to the data tags.

### Claim 6 (depends on 4)

A method according to claim 4, wherein using n-gram models includes accumulating and grouping the data tags.

### Claim 7 (depends on 1)

A method according to claim 1, wherein the higher order entities include semantic qualities.

### Claim 8 (depends on 1)

A method according to claim 1, wherein the higher order entities include user intentions.

### Claim 9 (depends on 1)

A method according to claim 1, wherein the text database is for a natural language understanding (NLU) application.

### Claim 10 (depends on 1)

A method according to claim 1, wherein the text database is for a user dialog application.

### Claim 11 (depends on 1, 2, 3, 4, 5, 6, 7, 8, 9, 10)

A developer interface, executing on a computer system, for tagging text data using the method according to any of claims 1-10.

### Claim 12 (depends on 1, 2, 3, 4, 5, 6, 7, 8, 9, 10)

A computer program product in a non-transitory computer readable storage medium for execution on at least one processor of a method of automatically processing text data, the computer program product having instructions for execution on the at least one processor comprising program code for performing the method according to any of claims 1-10.
