# US 2021/0174241 A1 · Trait-modeled chatbots

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

- Number: US 2021/0174241 A1
- Document ID: US20210174241A1
- Status: Published application
- Kind: A1
- Application number: 16/705,849
- Filed: 2019-12-06
- Published: 2021-06-10
- Earliest claimed: 2019-12-06
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Abhijit Mishra; Enara C. Vijil; Seema Nagar; Kuntal Dey
- Web page: https://patentgrep.com/patent/US20210174241A1

## CPC

- G06N 20/00: Machine learning
- G06F 16/313: {Selection or weighting of terms for indexing}
- H04L 51/02: using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages
- G06F 16/36: Creation of semantic tools, e.g. ontology or thesauri
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

One embodiment of the invention provides a method of training a chatbot. The method comprises identifying one or more chat logs that exhibit a trait. The method further comprises identifying one or more labels associated with the trait based on the one or more chat logs. The method further comprises training the chatbot to generate a response that models the trait based on the one or more chat logs. The method further comprises labeling the chatbot with the one or more labels.

## Claims

### Claim 1

A method of training a chatbot, comprising:
identifying one or more chat logs that exhibit a trait;
identifying one or more labels associated with the trait based on the one or more chat logs;
training the chatbot to generate a response that models the trait based on the one or more chat logs; and
labeling the chatbot with the one or more labels.

### Claim 2 (depends on 1)

The method of claim 1, wherein the chatbot is invoked to join a virtual chatroom in response to a chat session involving a user in the virtual chatroom who requires one or more responses that models the trait.

### Claim 3 (depends on 1)

The method of claim 1, wherein identifying one or more chat logs that exhibit a trait comprises:
receiving a database of chat logs including the one or more chat logs;
for each text unit of each chat log of the database, determining a degree of similarity between the text unit and the trait, and generating a mapping between the text unit and the trait, wherein the mapping has a weight that is based on the degree of similarity; and
identifying the one or more chat logs based on one or more mappings between one or more text units of the one or more chat logs and the trait, wherein each of the one or more mappings has a weight that satisfies a pre-determined weight threshold.

### Claim 4 (depends on 3)

The method of claim 3, further comprising:
receiving a set of traits including the trait;
identifying a set of chat logs of the database that exhibit at least one trait of the set of traits;
identifying a set of labels associated with at least one trait of the set of traits based on the set of chat logs;
training the chatbot to generate a response that jointly models the set of traits based on the set of chat logs; and
labeling the chatbot with the set of labels.

### Claim 5 (depends on 4)

The method of claim 4, wherein each label comprises a keyword or a concept associated with a trait of the set of traits.

### Claim 6 (depends on 4)

The method of claim 4, wherein each trait of the set of traits comprises a personality trait or an expertise trait that a domain expert has.

### Claim 7

A system for training a chatbot, comprising:
at least one processor; and
a non-transitory processor-readable memory device storing instructions that when executed by the at least one processor causes the at least one processor to perform operations including:
identifying one or more chat logs that exhibit a trait;
identifying one or more labels associated with the trait based on the one or more chat logs;
training the chatbot to generate a response that models the trait based on the one or more chat logs; and
labeling the chatbot with the one or more labels.

### Claim 8 (depends on 7)

The system of claim 7, wherein the chatbot is invoked to join the virtual chatroom in response to a chat session involving a user in the virtual chatroom who requires one or more responses that models the trait.

### Claim 9 (depends on 7)

The system of claim 7, wherein identifying one or more chat logs that exhibit a trait comprises:
receiving a database of chat logs including the one or more chat logs;
for each text unit of each chat log of the database, determining a degree of similarity between the text unit and the trait, and generating a mapping between the text unit and the trait, wherein the mapping has a weight that is based on the degree of similarity; and
identifying the one or more chat logs based on one or more mappings between one or more text units of the one or more chat logs and the trait, wherein each of the one or more mappings has a weight that satisfies a pre-determined weight threshold.

### Claim 10 (depends on 9)

The system of claim 9, wherein the operations further include:
receiving a set of traits including the trait;
identifying a set of chat logs of the database that exhibit at least one trait of the set of traits;
identifying a set of labels associated with at least one trait of the set of traits based on the set of chat logs;
training the chatbot to generate a response that jointly models the set of traits based on the set of chat logs; and
labeling the chatbot with the set of labels.

### Claim 11

A method of controlling invocation of at least one of a plurality of available chatbots in a virtual chatroom, comprising:
initiating a chat session with a user in the virtual chatroom;
determining an intent of the user based on a chat log of the chat session;
extracting a set of traits from the chat session based on the intent of the user, wherein the set of traits extracted is indicative of at least one trait requirement that a response to the user must satisfy;
selecting a chatbot from the plurality of available chatbots based on the set of traits extracted, wherein the chatbot selected is configured to generate a response that satisfies the at least one trait requirement; and
invoking the chatbot selected in the virtual chatroom to interact with the user.

### Claim 12 (depends on 11)

The method of claim 11, wherein extracting a set of traits from the chat session based on the intent of the user comprises:
receiving one or more mappings, wherein each mapping is between a label and a trait; and
identifying one or more text units of the chat log that map to the set of traits extracted based on the one or more mappings.

### Claim 13 (depends on 11)

The method of claim 11, wherein selecting a chatbot from the plurality of available chatbots based on the set of traits extracted comprises:
determining whether the plurality of available chatbots includes a first chatbot configured to generate a response that models the set of traits extracted.

### Claim 14 (depends on 13)

The method of claim 13, wherein the first chatbot is labeled with one or more labels that are associated with the set of traits extracted.

### Claim 15 (depends on 14)

The system of claim 14, wherein each label of the one or more labels comprises a keyword or a concept associated with a trait of the set of traits extracted.

### Claim 16 (depends on 13)

The method of claim 13, wherein the chatbot selected is the first chatbot in response to determining the plurality of available chatbots includes the first chatbot.

### Claim 17 (depends on 13)

The method of claim 13, wherein selecting a chatbot from the plurality of available chatbots based on the set of traits extracted further comprises:
in response to determining the plurality of available chatbots does not include the first chatbot:
identifying a second chatbot included in the plurality of available chatbots based on distance mapping and a pre-determined similarity threshold, wherein the second chatbot is configured to generate a response that models a set of different traits similar to the set of traits extracted, and the chatbot selected is the second chatbot.

### Claim 18 (depends on 11)

The method of claim 11, further comprising:
detecting a change in the intent of the user based on the chat log;
extracting an additional set of traits from the chat session based on the change, wherein the additional set of traits extracted is indicative of at least one different trait requirement that a different response to the user must satisfy;
selecting a different chatbot from the plurality of available chatbots based on the additional set of traits extracted, wherein the different chatbot selected is configured to generate a response that satisfies the at least one different trait requirement; and
invoking the different chatbot selected in the virtual chatroom to interact with the user.

### Claim 19 (depends on 11)

The method of claim 11, wherein each trait of the set of traits extracted comprises a personality trait or an expertise trait that a domain expert has.

### Claim 20 (depends on 11)

The method of claim 11, wherein the set of traits extracted comprises one of a single trait or multiple traits.
