# US 2021/0264112 A1 · Bot dialog manager

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

- Number: US 2021/0264112 A1
- Document ID: US20210264112A1
- Status: Published application
- Kind: A1
- Application number: 17/182,217
- Filed: 2021-02-22
- Published: 2021-08-26
- Earliest claimed: 2020-02-25
- Assignees: Prosper Funding LLC
- Applicants: Prosper Funding LLC
- Inventors: Paul Golding
- Web page: https://patentgrep.com/patent/US20210264112A1

## CPC

- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/35: Discourse or dialogue representation
- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06F 40/30: Semantic analysis
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning
- G06N 3/09: Supervised learning
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/045: Combinations of networks
- G06N 3/08: Learning methods
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A method manages bot dialogue. A user input is converted to a phrase vector. A set of identified tokens are identified by a token identification engine from the phrase vector. An unsupervised token is selected from the set of identified tokens. A supervised token is selected from the set of identified tokens. A voted token selected from the unsupervised token and the supervised token. A next token is identified based on a set of recent tokens that includes the voted token. The next token is presented as one of a voice communication and an email communication.

## Claims

### Claim 1

A method comprising:
converting a user input to a phrase vector;
identifying, by a token identification engine, a set of identified tokens from the phrase vector;
selecting an unsupervised token from the set of identified tokens;
selecting a supervised token from the set of identified tokens;
selecting a voted token from the unsupervised token and the supervised token;
identifying a next token based on a set of recent tokens that includes the voted token; and
presenting the next token as one of a voice communication and an email communication.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
converting, by an input formatting engine, the user input to a text string;
converting, by a text to word vector engine, the text string to a set of word vectors; and
converting, by a word vector to phrase vector engine, the set of word vectors to the phrase vector.

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
training the word vector to phrase vector engine with a plurality of sets of word vectors, a plurality of hashtags, and a plurality of tokens to generate a phrase vector space of latent features based on the plurality of sets of word vectors, the plurality of hashtags, and the plurality of tokens.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
scoring, with an unsupervised scoring engine, each identified token of the set of identified tokens using the set of recent tokens prior to the voted token being included in the set of recent tokens,
wherein the unsupervised token has a highest unsupervised score from the set of identified tokens.

### Claim 5 (depends on 1)

The method of claim 1, further comprising:
scoring, with a supervised scoring engine, each identified token of the set of identified tokens using an explicit path state that identifies a current path state and a recent path list,
wherein the supervised token has a highest supervised score from the set of identified tokens.

### Claim 6 (depends on 1)

The method of claim 1, further comprising:
scoring, with a voting engine, the unsupervised token and the supervised token using a set of saliency weights,
wherein the voted token has a highest voted score from the set of identified tokens.

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
updating the set of recent tokens to include the voted token;
setting an explicit path state based on the voted token;
updating a recent path list based on the voted token; and
identifying, by a next token engine, the next token using the set of recent tokens, the explicit path state, and the recent path list.

### Claim 8

A system comprising:
a processor;
a memory coupled to the processor;
the memory comprising an application, wherein the application executes on the processor and is configured for:
converting a user input to a phrase vector;
identifying, by a token identification engine, a set of identified tokens from the phrase vector;
selecting an unsupervised token from the set of identified tokens;
selecting a supervised token from the set of identified tokens;
selecting a voted token from the unsupervised token and the supervised token;
identifying a next token based on a set of recent tokens that includes the voted token; and
presenting the next token as one of a voice communication and an email communication.

### Claim 9 (depends on 8)

The system of claim 8, wherein the application is further configured for:
converting, by an input formatting engine, the user input to a text string;
converting, by a text to word vector engine, the text string to a set of word vectors; and
converting, by a word vector to phrase vector engine, the set of word vectors to the phrase vector.

### Claim 10 (depends on 9)

The system of claim 9, wherein the application is further configured for:
training the word vector to phrase vector engine with a plurality of sets of word vectors, a plurality of hashtags, and a plurality of tokens to generate a phrase vector space of latent features based on the plurality of sets of word vectors, the plurality of hashtags, and the plurality of tokens.

### Claim 11 (depends on 8)

The system of claim 8, wherein the application is further configured for:
scoring, with an unsupervised scoring engine, each identified token of the set of identified tokens using the set of recent tokens prior to the voted token being included in the set of recent tokens,
wherein the unsupervised token has a highest unsupervised score from the set of identified tokens.

### Claim 12 (depends on 8)

The system of claim 8, wherein the application is further configured for:
scoring, with a supervised scoring engine, each identified token of the set of identified tokens using an explicit path state that identifies a current path state and a recent path list,
wherein the supervised token has a highest supervised score from the set of identified tokens.

### Claim 13 (depends on 8)

The system of claim 8, wherein the application is further configured for:
scoring, with a voting engine, the unsupervised token and the supervised token using a set of saliency weights,
wherein the voted token has a highest voted score from the set of identified tokens.

### Claim 14 (depends on 8)

The system of claim 8, wherein the application is further configured for:
updating the set of recent tokens to include the voted token;
setting an explicit path state based on the voted token;
updating a recent path list based on the voted token; and
identifying, by a next token engine, the next token using the set of recent tokens, the explicit path state, and the recent path list.

### Claim 15

One or more non-transitory computer readable mediums comprising computer readable program code for:
converting a user input to a phrase vector;
identifying, by a token identification engine, a set of identified tokens from the phrase vector;
selecting an unsupervised token from the set of identified tokens;
selecting a supervised token from the set of identified tokens;
selecting a voted token from the unsupervised token and the supervised token;
identifying a next token based on a set of recent tokens that includes the voted token; and
presenting the next token as one of a voice communication and an email communication.

### Claim 16 (depends on 15)

The one or more non-transitory computer readable mediums of claim 15, further comprising computer readable program code for:
converting, by an input formatting engine, the user input to a text string;
converting, by a text to word vector engine, the text string to a set of word vectors; and
converting, by a word vector to phrase vector engine, the set of word vectors to the phrase vector.

### Claim 17 (depends on 16)

The one or more non-transitory computer readable mediums of claim 16, further comprising computer readable program code for:
training the word vector to phrase vector engine with a plurality of sets of word vectors, a plurality of hashtags, and a plurality of tokens to generate a phrase vector space of latent features based on the plurality of sets of word vectors, the plurality of hashtags, and the plurality of tokens.

### Claim 18 (depends on 15)

The one or more non-transitory computer readable mediums of claim 15, further comprising computer readable program code for:
scoring, with an unsupervised scoring engine, each identified token of the set of identified tokens using the set of recent tokens prior to the voted token being included in the set of recent tokens,
wherein the unsupervised token has a highest unsupervised score from the set of identified tokens.

### Claim 19 (depends on 15)

The one or more non-transitory computer readable mediums of claim 15, further comprising computer readable program code for:
scoring, with a supervised scoring engine, each identified token of the set of identified tokens using an explicit path state that identifies a current path state and a recent path list,
wherein the supervised token has a highest supervised score from the set of identified tokens.

### Claim 20 (depends on 15)

The one or more non-transitory computer readable mediums of claim 15, further comprising computer readable program code for:
scoring, with a voting engine, the unsupervised token and the supervised token using a set of saliency weights,
wherein the voted token has a highest voted score from the set of identified tokens.
