# US 2022/0277149 A1 · Statistical language models for simulating communication sessions

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

- Number: US 2022/0277149 A1
- Document ID: US20220277149A1
- Status: Published application
- Kind: A1
- Application number: 17/189,081
- Filed: 2021-03-01
- Published: 2022-09-01
- Earliest claimed: 2021-03-01
- Assignees: Asapp, Inc.
- Applicants: Asapp, Inc.
- Inventors: Samuel Andrew Altschul; Ramya Ramakrishnan; Hashan Buddhika Narangodage; Kilian Quirin Weinberger; Tianyi Zhang
- Web page: https://patentgrep.com/patent/US20220277149A1

## CPC

- G06F 40/56: Natural language generation
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/35: Discourse or dialogue representation
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0499: Feedforward networks
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A statistical language model may be used to simulate one or more users of a conversation. The statistical language model may be used to train a user to participate in a particular types of conversation by simulating communications by another type of user in the conversation. The communications may be simulated by selecting a simulation context from available simulation contexts and the simulation context may correspond to a difficulty level. Upon receiving a communication from a user, a responsive simulated communication may be generated by processing the received communication and the simulation context with the statistical language model. Upon completion of the simulation, another simulation context may be selected for the next simulation.

## Claims

### Claim 1

A computer-implemented method, comprising:
starting a simulated conversation between a user of a first user-type and a simulated user of a second user-type;
obtaining a statistical language model, wherein the statistical language model represents natural language communications of the second user-type and the statistical language model was trained on natural language communications between users of the first user-type and users of the second user-type;
selecting a first simulation context from a plurality of available simulation contexts, wherein the first simulation context corresponds to a first difficulty level;
presenting the first simulation context to the user;
receiving a first natural language communication from the user;
generating a second natural language communication by processing the first simulation context and the first natural language communication with the statistical language model;
presenting the second natural language communication to the user as a simulated communication from the simulated user;
processing natural language communications of the simulated conversation with a conversation scorer model to determine a performance score for the user in the simulated conversation;
selecting a second simulation context from the plurality of available simulation contexts, wherein the second simulation context corresponds to a second difficulty level and wherein the second simulation context is selected using the performance score; and
implementing a second simulated conversation using the statistical language model and the second simulation context.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, wherein the statistical language model is a neural network language model.

### Claim 3 (depends on 1)

The computer-implemented method of claim 1, wherein generating the second natural language communication comprises:
tokenizing the first natural language communication to obtain a sequence of tokens;
obtaining token embeddings for the sequence of tokens; and
obtaining positional encodings for the sequence of tokens.

### Claim 4 (depends on 1)

The computer-implemented method of claim 1, wherein generating the second natural language communication comprises processing the first natural language communication with one or more transformer decoder blocks.

### Claim 5 (depends on 1)

The computer-implemented method of claim 1, wherein generating the second natural language communication comprises generating tokens of the second natural language communication sequentially using top-p or top-k sampling.

### Claim 6 (depends on 1)

The computer-implemented method of claim 1, wherein the first natural language communication comprises text.

### Claim 7 (depends on 1)

The computer-implemented method of claim 1, wherein the first simulation context comprises a sequence of natural language communications from an archived conversation between two users.

### Claim 8 (depends on 1)

The computer-implemented method of claim 1, wherein the first simulation context comprises a representation of natural language from an archived conversation between two users.

### Claim 9

A system, comprising:
at least one server computer comprising at least one processor and at least one memory, the at least one server computer configured to:
start a simulated conversation between a user of a first user-type and a simulated user of a second user-type;
obtain a statistical language model, wherein the statistical language model represents natural language communications of the second user-type and the statistical language model was trained on natural language communications between users of the first user-type and users of the second user-type;
select a first simulation context from a plurality of available simulation contexts, wherein the first simulation context corresponds to a first difficulty level;
receive a first natural language communication from the user;
generate a second natural language communication by processing the first simulation context and the first natural language communication with the statistical language model;
present the second natural language communication to the user as a simulated communication from the simulated user; and
process natural language communications of the simulated conversation with a conversation scorer model to determine a performance score for the user in the simulated conversation.

### Claim 10 (depends on 9)

The system of claim 9, wherein the at least one server computer is configured to generate the second natural language communication by processing, with the statistical language model all natural language communications in the simulated conversation before the first natural language communication.

### Claim 11 (depends on 9)

The system of claim 9, wherein the first difficulty level was determined by:
obtaining a second statistical language model, wherein the second statistical language model represents communications of the second user-type;
executing an automated simulated conversation using the statistical language model and the second statistical language model, wherein in the automated simulated conversation is between a simulated user of the first user-type and a simulated user of the second user-type;
processing communications of the automated simulated conversation with the conversation scorer model to determine a simulation score for the automated simulated conversation; and
determining the first difficulty level using the simulation score.

### Claim 12 (depends on 9)

The system of claim 9, wherein the first user-type is an agent and the second user-type is a customer.

### Claim 13 (depends on 9)

The system of claim 9, comprising:
starting a second simulated conversation between the user and the simulated user; and
selecting a second simulation context from the plurality of available simulation contexts, wherein the second simulation context corresponds to a second difficulty level and the second simulation context is selected using the performance score.

### Claim 14 (depends on 9)

The system of claim 9, comprising:
starting a second simulated conversation between the user and the simulated user; and
selecting a second simulation context from the plurality of available simulation contexts, wherein the second simulation context corresponds to a second topic that is different from a first topic of the first simulation context.

### Claim 15 (depends on 9)

The system of claim 9, wherein the first simulation context comprises a sequence of natural language communications from an archived conversation between two users.

### Claim 16 (depends on 9)

The system of claim 9, wherein the first simulation context comprises a representation of natural language from an archived conversation between two users.

### Claim 17

One or more non-transitory, computer-readable media comprising computer-executable instructions that, when executed, cause at least one processor to perform actions comprising:
starting a simulated conversation between a user of a first user-type and a simulated user of a second user-type;
obtaining a statistical language model, wherein the statistical language model represents natural language communications of the second user-type and the statistical language model was trained on natural language communications between users of the first user-type and users of the second user-type;
selecting a first simulation context from a plurality of available simulation contexts, wherein the first simulation context corresponds to a first difficulty level;
receiving a first natural language communication from the user;
generating a second natural language communication by processing the first simulation context and the first natural language communication with the statistical language model;
presenting the second natural language communication to the user as a simulated communication from the simulated user; and
processing natural language communications of the simulated conversation with a conversation scorer model to determine a performance score for the user in the simulated conversation.

### Claim 18 (depends on 17)

The one or more non-transitory, computer-readable media of claim 17, wherein the actions comprise:
starting a second simulated conversation between the user and the simulated user; and
selecting a second simulation context from the plurality of available simulation contexts, wherein the second simulation context corresponds to a second difficulty level and the second simulation context is selected using the performance score.

### Claim 19 (depends on 17)

The one or more non-transitory, computer-readable media of claim 17, wherein the first simulation context comprises a sequence of natural language communications from an archived communication session between two users.

### Claim 20 (depends on 17)

The one or more non-transitory, computer-readable media of claim 17, wherein generating the second natural language communication comprises processing the first natural language communication with one or more transformer decoder blocks.
