# US 2017/0364505 A1 · Systems and methods for building state specific multi-turn contextual language understanding systems

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

- Number: US 2017/0364505 A1
- Document ID: US20170364505A1
- Status: Published application
- Kind: A1
- Application number: 15/185,488
- Filed: 2016-06-17
- Published: 2017-12-21
- Earliest claimed: 2016-06-17
- Assignees: Microsoft Technology Licensing, LLC
- Applicants: Microsoft Technology Licensing, LLC
- Inventors: Ruhi Sarikaya; Young-Bum Kim; Alexandre Rochette
- Web page: https://patentgrep.com/patent/US20170364505A1

## CPC

- G06F 40/35: Discourse or dialogue representation
- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06F 40/40: Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30)
- G06F 40/56: Natural language generation
- G10L 15/063: {Training}
- G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning}
- G10L 15/1822: {Parsing for meaning understanding}
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

Systems and methods for building a dialog-state specific multi-turn contextual language understanding system are provided. More specifically, the systems and methods infer or are configured to infer a state-specific schema and/or state-specific rules from a formed single-shot language understanding model and/or a single-shot rule set. As such, the systems and methods only require the information necessary to form a single-shot language understanding model and/or a single-shot rule set from a builder to form or build the dialog-state specific multi-turn contextual language understanding system. Accordingly, the systems and methods for building a dialog-state specific multi-turn contextual language understanding system reduce the expertise, time, and resources necessary to build a dialog-state specific multi-turn contextual language understanding system for an application when compared to systems and methods that require further input from the builder than necessary to build a single-shot language understanding system.

## Claims

### Claim 1

A system with a platform for building a dialog-state specific multi-turn contextual language understanding system for an application, the system comprising:
at least one processor; and
a memory for storing and encoding computer executable instructions that, when executed by the at least one processor is operative to:
receive information from a builder for building a single-shot language understanding model, wherein the information includes any slots and entities necessary for performing a task in one turn of a dialog;
configure the single-shot language understanding model based on the information;
constrain a decoding of the single-shot language understanding model to output only dialog-state specific slots and dialog-state specific entities for each determined dialog-state based on a dialog-state dependent semantic schema; and
implement a constrained single-shot language understanding model to form the dialog-state specific multi-turn contextual language understanding system,
wherein the multi-turn contextual language understanding system is capable of analyzing an utterance from a user.

### Claim 2 (depends on 1)

The system of claim 1, wherein the at least one processor is further operative to infer the dialog-state dependent semantic schema for different dialog-states based on the single-shot language understanding model and a semantic schema for the single-shot language understanding model.

### Claim 3 (depends on 1)

The system of claim 1, wherein the at least one processor is further operative to provide a capability to infer the dialog-state dependent semantic schema for different dialog-states based on the single-shot language understanding model and based on user input received by the dialog-state specific multi-turn contextual language understanding system from a dialog with a user and a system prompt given for the dialog or a task.

### Claim 4 (depends on 1)

The system of claim 1, wherein the at least one processor is further operative to receive a task definition for the application from the builder, wherein the single-shot language understanding model is built for a task.

### Claim 5 (depends on 1)

The system of claim 1, wherein the information includes parameters, slots, and entities necessary to define a task.

### Claim 6 (depends on 1)

The system of claim 1, wherein the single-shot language understanding model is a machine learned model.

### Claim 7 (depends on 1)

The system of claim 1, wherein the system is a server.

### Claim 8 (depends on 1)

The system of claim 1, wherein the at least one processor is further operative to implement the constrained single-shot language understanding model to form the dialog-state specific multi-turn contextual language understanding system in response to receiving an implementation request from the builder.

### Claim 9 (depends on 1)

The system of claim 1, wherein the application is at least one of:
a digital assistant application;
a voice recognition application;
an email application;
a social networking application;
a collaboration application;
an enterprise management application;
a messaging application;
a word processing application;
a spreadsheet application;
a database application;
a presentation application;
a contacts application;
a gaming application;
an e-commerce application;
an e-business application;
a transactional application;
a device control application;
a web interface application;
an exchange application; or
a calendaring application.

### Claim 10 (depends on 1)

The system of claim 1, wherein the at least one processor is further operative to run a simulation of the constrained single-shot language understanding model in response to receiving a simulation request from the builder.

### Claim 11

A system with a platform for building a dialog-state specific multi-turn contextual language understanding system for an application, the system comprising:
at least one processor; and
a memory for storing and encoding computer executable instructions that, when executed by the at least one processor is operative to:
receive information from a builder for authoring a combined single-shot language understanding system based on a combination of a machine learned single-shot model and single-shot rules, wherein the information includes any slots and entities necessary for performing a task in one turn of a dialog;
form the combined single-shot language understanding system based on the information, wherein the combined single-shot language understanding system includes a machine learned single-shot language understanding model and a the single-shot rules;
adapt the combined single-shot language understanding system for dialog-state dependent decoding to form an adapted combined single-shot language understanding model; and
implement the adapted combined single-shot language understanding model to form the dialog-state specific multi-turn contextual language understanding system,
wherein the multi-turn contextual language understanding system is capable of analyzing an utterance from a user.

### Claim 12 (depends on 11)

The system of claim 11, wherein adapting the combined single-shot language understanding system for dialog-state dependent decoding comprises:
constrain a decoding of the machine learned single-shot language understanding model to output only dialog-state specific slots and dialog-state specific entities for at least one determined dialog-state based on a dialog-state dependent semantic schema; and
process the single-shot rules and the dialog-state dependent semantic schema to derive or to provide a capability to derive dialog-state dependent rules to tag the dialog-state specific slots and the dialog-state specific entities for the at least one determined dialog-state or another determined dialog-state.

### Claim 13 (depends on 12)

The system of claim 12, wherein the at least one processor is operative to:
infer the dialog-state dependent semantic schema for different dialog-states based on the information; and
derive the dialog-state dependent rules.

### Claim 14 (depends on 12)

The system of claim 12, wherein the at least one processor is operative to:
provide a capability to infer the dialog-state dependent semantic schema for different dialog-states based on the single-shot rules or the machine learned single-shot language understanding model, and based on received user input by the dialog-state specific multi-turn contextual language understanding system from a dialog with a user during the decoding; and
provide the capability to derive the dialog-state dependent rules.

### Claim 15

A method for building a dialog-state specific multi-turn contextual language understanding system for an application, the method comprising:
receiving information from a builder for building a single-shot language understanding model, wherein the information includes any slots and entities necessary for performing a task in one turn of a dialog;
configuring the single-shot language understanding model based on the information;
constraining a decoding of the single-shot language understanding model to output only dialog-state specific slots and dialog-state specific entities for each determined dialog-state based on a dialog-state dependent semantic schema; and
implementing a constrained single-shot language understanding model to form the dialog-state specific multi-turn contextual language understanding system,
wherein the multi-turn contextual language understanding system is capable of analyzing an utterance from a user.

### Claim 16 (depends on 15)

The method of claim 15, further comprising:
inferring the dialog-state dependent semantic schema for different dialog-states based on the single-shot language understanding model and a semantic schema for the single-shot language understanding model.

### Claim 17 (depends on 15)

The method of claim 15, further comprising:
providing a capability to infer the dialog-state dependent semantic schema for different dialog-states based on the single-shot language understanding model and based on user input received by the dialog-state specific multi-turn contextual language understanding system from a dialog with a user and a system prompt given for the dialog or a task.

### Claim 18 (depends on 15)

The method of claim 15, wherein the information includes parameters, slots, and entities necessary to define a task.

### Claim 19 (depends on 15)

The method of claim 15, further comprising:
receiving an implementation request from the builder;
implementing the constrained single-shot language understanding model to form the dialog-state specific multi-turn contextual language understanding system in response to the implementation request.

### Claim 20 (depends on 15)

The method of claim 15, further comprising:
receiving a simulation request from the builder;
running a simulation of the constrained single-shot language understanding model in response to the simulation request.
