# US 2020/0395000 A1 · Human-machine dialogue method and electronic device

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

- Number: US 2020/0395000 A1
- Document ID: US20200395000A1
- Status: Published application
- Kind: A1
- Application number: 16/769,272
- Filed: 2019-01-22
- Published: 2020-12-17
- Earliest claimed: 2018-01-23
- Assignees: Ai Speech Co., Ltd.
- Applicants: Ai Speech Co., Ltd
- Inventors: Min CHU; Taotao GUO; Zhongyuan DAI; Chao YANG
- Web page: https://patentgrep.com/patent/US20200395000A1

## CPC

- G10L 15/02: Feature extraction for speech recognition; Selection of recognition unit
- G06F 16/3329: Natural language query formulation
- G06F 16/337: {Profile generation, learning or modification}
- G06N 5/04: Inference or reasoning models
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- 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)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Disclosed is a human-computer dialogue method including determining a set number of jump topics about a target topic, and generating a topic jump map converging to the target topic based on the correlation intensions among the set number of jump topics; after an initial response to a user's dialogue request, selecting from the topic jump map a jump topic to which the user's dialogue request relates as an initial topic for a first round of recommendation; after completing a human-machine dialogue of the initial topic, determining a jump topic to jump according to the jump probability of jumping out of the initial topic to the k jump topics at the downstream level for a next round of recommendation; and gradually guiding from the initial topic to the target topic by step-by-step recommendation. A more fluent and efficient human-machine dialogue based on a clear communication goal can be realized.

## Claims

### Claim 1

A human-machine dialogue method applicable to an electronic device, the method comprising:
determining a set number of jump topics around a target topic, and generating a topic jump map converging to the target topic based on the correlation intensions among the set number of jump topics, wherein k path segments from each jump topic to k jump topics at a downstream level are configured respectively with an initial jump probability;
selecting from the topic jump map a jump topic to which the user's dialogue request belongs as an initial topic for a first round of recommendation to the user after an initial response to a user's dialogue request;
determining a jump topic for jumping to according to the jump probability of jumping out of the initial topic to the k jump topics at the downstream level for a next round of recommendation to the user after completing a human-machine dialogue of the initial topic; and
gradually guiding the dialogue from the initial topic to the target topic in a step-by-step recommendation manner.

### Claim 2 (depends on 1)

The method according to claim 1, further comprising:
after jumping to the target topic through the step-by-step recommendation and completing the human-machine dialogue, determining a jump path from the initial topic to the target topic; and
increasing the jump probability of each path segment on the jump path, and updating the topic jump map for a next human-machine dialogue.

### Claim 3 (depends on 2)

The method according to claim 2, wherein the sum of the initial jump probabilities of the k path segments from each jump topic to the k jump topics at the downstream level is 1; and
while increasing the jump probability of the path segment P (i, i+1) from the i-th jump topic on the jump path to the (i+1)-th jump topic at the downstream level, decreasing accordingly the jump probabilities of k−1 path segments of k−1 jump topics at the downstream level other than the (i+1)-th jump topic from the i-th jump topic to maintain the sum of the jump probabilities of the k path segments at the downstream level from the i-th jump topic to be 1.

### Claim 4

A human-machine dialogue method applicable to electronic devices, the method comprising:
determining a set number of jump topics related to a specified field, and constructing a topic jump graph between the topics based on the correlation intensions among the set number of jump topics, wherein k path segments from each jump topic to k jump topics at a downstream level are configured respectively with an initial jump probability;
determining a jump topic to which the user's dialogue request belongs as an initial topic for a first round of recommendation to the user after an initial response to a user's dialogue request;
determining a jump topic for jumping to according to the jump probability of jumping out of the initial topic to the k jump topics at the downstream level for a next round of recommendation to the user after completing a human-machine dialogue of the initial topic; and
continuing recommending the jump topics of the topic jump map for human-machine dialogue in a round-by-round recommendation manner.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the sum of the initial jump probabilities of the k path segments from each jump topic to the k jump topics at the downstream level is 1; and
the process of recommending jump topics further comprises:
when one of second jump topics at a downstream level of a first jump topic is selected for the next round of recommendation,
if the user accepts the second jump topic, increasing the jump probability of the path segment from the first jump topic to the second jump topic, and decreasing the jump probabilities of the path segments from the first jump topic to other second jump topics at the downstream level, to maintain the sum of the jump probabilities of the path segments from the first jump topic to all the second jump topics at the downstream level to be 1; and
if the user does not accept the second jump topic, decreasing the jump probability of the path segment from the first jump topic to the second jump topic, and increasing the jump probabilities of the path segments from the first jump topic to the other second jump topics at the downstream level, to maintain the sum of the jump probabilities of the path segments from the first jump topic to all the second jump topics at the downstream level to be 1, and select another second jump topic at the downstream level of the first jump topic for the next round of recommendation.

### Claim 6 (depends on 5)

The method according to claim 5, further comprising:
when the times of changing selections exceeds a predetermined number, reselecting a topic independent of the first jump topic.

### Claim 7 (depends on 3)

The method according to claim 3, wherein after multiple human-machine dialogues, if the jump probability of at least one path segment is decreased below a predetermined threshold, the at least one path segment in the topic jump map is disconnected, and the topic jump map is updated for use in the next human-machine dialogue.

### Claim 8 (depends on 7)

The method according to claim 7, wherein when the at least one path segment in the topic jump map is disconnected, a new jump topic is configured for an endpoint of the at least one path segment.

### Claim 9 (depends on 1)

The method according to claim 1, further comprising classifying users according to user attributes to generate corresponding topic jump maps according to user categories.

### Claim 10

An electronic device, comprising at least one processor and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, wherein execution of the instructions by the at least one processor causes the at least one processor to perform the steps of:
determining a set number of jump topics around a target topic, and generating a topic jump map converging to the target topic based on the correlation intensions among the set number of jump topics, wherein k path segments from each jump topic to k jump topics at a downstream level are configured respectively with an initial jump probability;
selecting from the topic jump map a jump topic to which the user's dialogue request belongs as an initial topic for a first round of recommendation to the user after an initial response to a user's dialogue request;
determining a jump topic for jumping to according to the jump probability of jumping out of the initial topic to the k jump topics at the downstream level for a next round of recommendation to the user after completing a human-machine dialogue of the initial topic; and
gradually guiding the dialogue from the initial topic to the target topic in a step-by-step recommendation manner.

### Claim 11 (depends on 4)

The method according to claim 4, wherein the sum of the initial jump probabilities of the k path segments from each jump topic to the k jump topics at the downstream level is 1; and
the process of recommending jump topics further comprises:
when one of second jump topics at a downstream level of a first jump topic is selected for the next round of recommendation,
if the user accepts the second jump topic, increasing the jump probability of the path segment from the first jump topic to the second jump topic, and decreasing the jump probabilities of the path segments from the first jump topic to other second jump topics at the downstream level, to maintain the sum of the jump probabilities of the path segments from the first jump topic to all the second jump topics at the downstream level to be 1; and
if the user does not accept the second jump topic, decreasing the jump probability of the path segment from the first jump topic to the second jump topic, and increasing the jump probabilities of the path segments from the first jump topic to the other second jump topics at the downstream level, to maintain the sum of the jump probabilities of the path segments from the first jump topic to all the second jump topics at the downstream level to be 1, and select another second jump topic at the downstream level of the first jump topic for the next round of recommendation.

### Claim 12 (depends on 11)

The method according to claim 11, further comprising:
when the times of changing selections exceeds a predetermined number, reselecting a topic independent of the first jump topic.

### Claim 13 (depends on 5)

The method according to claim 5, wherein after multiple human-machine dialogues, if the jump probability of at least one path segment is decreased below a predetermined threshold, the at least one path segment in the topic jump map is disconnected, and the topic jump map is updated for use in the next human-machine dialogue.

### Claim 14 (depends on 13)

The method according to claim 13, wherein when the at least one path segment in the topic jump map is disconnected, a new jump topic is configured for an endpoint of the at least one path segment.

### Claim 15 (depends on 4)

The method according to claim 4, further comprising classifying users according to user attributes to generate corresponding topic jump maps according to user categories.

### Claim 16 (depends on 10)

The electronic device according to claim 10, wherein execution of the instructions by the at least one processor causes the at least one processor to perform the steps of:
after jumping to the target topic through the step-by-step recommendation and completing the human-machine dialogue, determining a jump path from the initial topic to the target topic; and
increasing the jump probability of each path segment on the jump path, and updating the topic jump map for a next human-machine dialogue.

### Claim 17 (depends on 16)

The electronic device according to claim 16, wherein the sum of the initial jump probabilities of the k path segments from each jump topic to the k jump topics at the downstream level is 1; and
while increasing the jump probability of the path segment P (i, i+1) from the i-th jump topic on the jump path to the (i+1)-th jump topic at the downstream level, decreasing accordingly the jump probabilities of k−1 path segments of k−1 jump topics at the downstream level other than the (i+1)-th jump topic from the i-th jump topic to maintain the sum of the jump probabilities of the k path segments at the downstream level from the i-th jump topic to be 1.

### Claim 18 (depends on 17)

The electronic device according to claim 17, wherein after multiple human-machine dialogues, if the jump probability of at least one path segment is decreased below a predetermined threshold, the at least one path segment in the topic jump map is disconnected, and the topic jump map is updated for use in the next human-machine dialogue.

### Claim 19 (depends on 18)

The electronic device according to claim 18, wherein when the at least one path segment in the topic jump map is disconnected, a new jump topic is configured for an endpoint of the at least one path segment.

### Claim 20 (depends on 10)

The electronic device according to claim 10, wherein execution of the instructions by the at least one processor causes the at least one processor to perform the steps of: classifying users according to user attributes to generate corresponding topic jump maps according to user categories.