WO2019184103A1

WO2019184103A1 - Person ip-based human-computer interaction method and system, medium and device

Info

Publication number: WO2019184103A1
Application number: PCT/CN2018/091636
Authority: WO
Inventors: 宋亚楠; 邱楠; 梁剑华; 邓婧文; 邹创华
Original assignee: 深圳狗尾草智能科技有限公司
Priority date: 2018-03-30
Filing date: 2018-06-15
Publication date: 2019-10-03
Also published as: CN108804698A

Abstract

The present invention provides a person IP-based human-computer interaction method and system, a medium and a device. The method of the present invention comprises: acquiring input information of a user; extracting an input entity included in the input information; identifying a user intention according to the input information; according to the input entity and the user intention and on the basis of a pre-established knowledge graph, generating reply information in the persona of a preset default person IP; and outputting, in the persona of a default person IP, the reply information to the user. The present invention can give the robot a humanoid personality, making the robot simulate a specific person IP and interact with the user with the identity of the specific person IP, thereby responding to the user in a more customized manner; furthermore, the personal habits of the person IP are taken into consideration, enabling the robot to be smarter and humanoid and to better interact with the user, thereby improving the user experience.

Description

Human-computer interaction method, system, medium and device based on character IP

Technical field

The present invention relates to the field of artificial intelligence technologies, and in particular, to a human-computer interaction method, system, medium and device based on a person IP.

Background technique

In the prior art, when the robot responds to the user input, it searches for a question similar to the user input in the existing dialog library, and then uses the answer corresponding to the question as a reply to the user input; or Using artificial intelligence algorithms such as machine learning, deep learning, and RNN, the existing interactive data is labeled as an input of the artificial intelligence algorithm, and the artificial intelligence algorithm is trained to obtain a training model, which can obtain a new input from the user. The user's reply is automatically generated, and the personality of the robot is not considered. The user cannot be replied in an anthropomorphic manner, and the user experience is poor.

Summary of the invention

In view of the deficiencies in the prior art, the present invention provides a human-computer interaction method, system, medium and device based on a person IP, which can give a robot anthropomorphic personality, and enable the robot to simulate a specific person IP, with the identity and user of the specific person IP. Interacting, more targeted response to users; and taking into account the character habits of the character IP, making the robot more intelligent, anthropomorphic, better interact with the user, can improve the user experience.

In a first aspect, the present invention provides a human-computer interaction method based on a person IP, including:

Obtain the user's input information;

Extracting an input entity included in the input information;

Identifying user intent based on the input information;

And generating, according to the input entity and the user intention, reply information with a preset default person IP as an image based on the pre-established knowledge map;

The reply message is output to the user with the default person IP as an image.

Preferably, the knowledge map comprises: a knowledge map and/or a general knowledge map with the person name of the default person IP as a core node.

Preferably, according to the input entity and the user intention, based on the pre-established knowledge map, generating reply information with a preset default person IP as an image, including:

Associating the input entity with an entity in the knowledge map;

Extracting, from the knowledge map, a knowledge map subgraph related to the input entity and user intent;

According to the knowledge map sub-graph and the grammar rule, the reply information with the preset default person IP as the image is generated.

Preferably, according to the knowledge map sub-graph and the grammar rule, generating reply information with a preset default person IP as an image, including:

Generating original reply information corresponding to the input information according to the knowledge map sub-graph and the grammar rule;

Calculating a confidence level of the default character IP in the current scene according to the weight of the stored historical data and the user intention;

Determining a reasonable character habit of the current scene according to the confidence level;

According to the reasonable person habits and grammar rules, the original reply information is processed to generate reply information with a preset default person IP as an image.

Preferably, the method further comprises:

Obtaining the character IP change instruction input by the user;

According to the change instruction, the person IP is set from the default person IP to the new person IP;

Retrieving a new knowledge map corresponding to the IP of the new character from the database;

Performing human-computer interaction with the user according to the new knowledge map.

Preferably, the knowledge map of at least one person IP is stored in the database.

Preferably, the reply information includes: voice response information and/or action response information.

In a second aspect, the present invention provides a human-computer interaction system based on a person IP, including:

An information acquisition module, configured to acquire input information of a user;

An entity extraction module, configured to extract an input entity included in the input information;

An intention identification module, configured to identify a user intention according to the input information;

a reply generation module, configured to generate, according to the input entity and the user intention, reply information with a preset default person IP as an image based on the pre-established knowledge map;

The information output module is configured to output the reply information to the user by using the default character IP as an image.

Preferably, the reply generation module includes: an association unit, configured to associate the input entity with an entity in the knowledge map; and an extraction unit, configured to extract, from the knowledge map, the input entity and the user An intent-related knowledge map sub-graph; an information generating unit configured to generate reply information with a preset default person IP as an image according to the knowledge map sub-picture and the grammar rule.

Preferably, the information generating unit includes: an original information generating subunit, configured to generate original reply information corresponding to the input information according to the knowledge map subgraph and grammar rules; and a confidence calculation subunit for storing according to the The weight of the historical data and the user intent, calculating a confidence level of the default character IP in the current scene, each character habit; the habit determining subunit is configured to determine a reasonable character habit of the current scene according to the confidence level; The information generating subunit is configured to process the original reply information according to the reasonable character habit and grammar rules, and generate reply information with a preset default person IP as an image.

Preferably, the method further includes: an instruction obtaining module, configured to acquire a character IP change instruction input by the user; and a change module, configured to set the character IP from the default character IP to the new person IP according to the change instruction; the knowledge map retrieval module And the new knowledge map corresponding to the new person IP is retrieved from the database; the interaction module is configured to perform human-computer interaction with the user according to the new knowledge map.

In a third aspect, the present invention provides a computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements a human IP interaction method based on a person IP in the first aspect.

In a fourth aspect, the present invention provides a human IP interaction device based on a person IP, comprising: a memory, a processor, and a computer program stored on the memory and operable on the processor, when the processor executes the program A human-computer interaction method based on the character IP in the first aspect is implemented.

The human-computer interaction method based on the person IP provided by the embodiment of the present invention extracts the input entity included in the input information by acquiring the input information of the user, identifies the user intention according to the input information, and then based on the input entity and the user intention, based on the pre-established The knowledge map generates a reply message with a preset default person IP as an image, and finally outputs a reply message to the user with the default character IP as an image. It can give the robot a personification personality, enable the robot to simulate the specific person IP, interact with the user as the specific person IP, and respond to the user more specifically; and consider the character habits of the person IP to make the robot more intelligent and anthropomorphic Better interaction with users can improve the user experience.

The human IP interaction system based on the person IP, a computer readable storage medium and a human-computer interaction device based on the character IP provided by the invention are the same as the above-mentioned human-computer interaction method based on the character IP The inventive concept has the same advantageous effects.

DRAWINGS

In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings to be used in the specific embodiments or the description of the prior art will be briefly described below. In all the figures, like elements or parts are generally identified by like reference numerals. In the figures, elements or parts are not necessarily drawn to scale.

1 is a flowchart of a human-computer interaction method based on a character IP according to the present invention;

FIG. 2 is a schematic diagram of a human-computer interaction system based on a person IP according to the present invention.

detailed description

The embodiments of the technical solution of the present invention will be described in detail below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and thus are merely exemplary and are not intended to limit the scope of the present invention.

It should be noted that the technical terms or scientific terms used herein should be used in the ordinary meaning as understood by those skilled in the art to which the invention belongs, unless otherwise stated.

The invention provides a human-computer interaction method, system, medium and device based on a person IP. Embodiments of the present invention will be described below with reference to the accompanying drawings.

The execution body of the present invention is a chat robot. The robot can be set with a default character IP when it leaves the factory. When the user is using the robot, the robot can interact with the user with the default character IP as the image. The invention can adapt to the fan group and meet the user's needs.

Please refer to FIG. 1. FIG. 1 is a flowchart of a human-computer interaction method based on a person IP according to a specific embodiment of the present invention. The human-computer interaction method based on a person IP provided by this embodiment includes:

Step S101: Acquire input information of the user.

The input information of the user may be one or more of voice input information, text input information, key operation input information, and the like, which are all within the protection scope of the present invention.

Step S102: Extract an input entity included in the input information.

After the input information is obtained, the input information needs to be converted into text input information; if the input information obtained is text input information, no conversion is required.

When extracting the input entities contained in the input information, the following steps are included:

The first step is to convert the input information into text input information;

The second step is to use a natural language understanding method to perform word segmentation on text input information;

In the third step, the physical alignment is used to align each word after the word segmentation with the entity;

The fourth step is to use disambiguation and generalization techniques to identify the entity in the word and extract the entity as the input entity.

Through this method, the input entity can be accurately extracted, thereby generating more reasonable reply information and better interacting with the user.

Step S103: Identify the user's intention according to the input information.

The process of identifying user intent based on input information is:

Extracting the input entity in the input information is the same as the method of step S102;

Identify the sentence pattern of the input information; when identifying the sentence pattern, it can be identified by using the regular rules;

The user's intent is identified based on the extracted input entity and the sentence pattern of the input information.

When the user's intent is identified based on the input entity and sentence pattern, methods such as machine learning methods, rules, and slot filling may be employed to identify the user's intent, and any applicable conventional techniques may be used to identify the user's intent. The specific user intent identification method is not limited herein.

Step S104: Generate reply information with a preset default person IP as an image based on the pre-established knowledge map according to the input entity and the user intention.

The knowledge map includes: a knowledge map and/or a general knowledge map with the person name of the default person IP as a core node.

Before generating the reply information based on the pre-established knowledge map according to the input entity and the user's intention, the method may further include:

The knowledge map is established, including: establishing a knowledge map with a character name of the character IP as a core node and a general knowledge map. Wherein, the knowledge map of the established person IP may include a knowledge map of a plurality of characters IP. For each character IP, multiple knowledge maps can be created, or a knowledge map can be created. There may also be one or more general knowledge maps, which are all within the scope of the present invention. In the present invention, the execution subject that establishes the knowledge map may be the server.

When establishing the knowledge map with the character name of the character IP as the core node, the method includes: collecting the character information of the character IP; and, according to the character information, establishing the knowledge map with the corresponding character name as the core node.

The process of establishing the knowledge map of the specific person IP is:

Set the character knowledge map structure; determine the character IP; collect the character information of the person IP, and establish a knowledge map with the corresponding person name as the core node according to the character information and the character knowledge map structure.

Among them, the character IP refers to a specific person image, which can be a star character. For example, user A, user B, and the like IP.

The character knowledge map architecture refers to the knowledge map structure with the character name as the core node.

The character information may include: objective information of the character, habit information, event information, subjective attitude information, and the like.

Among them, objective information, including: alias/nickname, gender, date of birth, occupation, specialty, belief, preference, appearance, Weibo account, number of fans, works (and number of times of play, number of people, time to market, specific role, Other participants), social relations (friends, relationships, previous marriages).

Customary information, including: character dialogue habits and character action habits.

Event information refers to some recorded things that people have done over a period of time.

Subjective attitude information refers to the attitudes and opinions held by a person on something.

Character dialogue habits, including: word habits, sentence habits, pronunciation habits, tone habits, etc.

The character action habit information refers to the usual habitual actions of the character, for example, grinning, touching, picking up hands, and the like.

The above-mentioned person information includes all the information of the person collected from various channels, for example, information can be obtained by directly providing an existing knowledge map, extracting from a web page, extracting from an interview, and the like.

After obtaining the character information, the knowledge map with the character name as the core node can be constructed according to the character information and the character knowledge map structure. At the time of construction, the character information is classified and stored according to the character knowledge map architecture. For example, a character dialogue library, an action library, and the like can be established. In this way, it is convenient to respond to the user's input information according to the knowledge map.

Among them, the character dialogue library contains the dialogue content of the character, such as the dialogue of the characters, the audio and video of the question and answer obtained from the webpage, the film and television works, the interview, etc., and then the content and the text of the dialogue are extracted from the audio and video to form a dialogue library. . The character's action library is used to help the virtual chat bot mimic the action of the character. The action library includes: the entire video of the character dance, the action segments of the characters participating in the film and television drama, and the typical habitual action segments of the characters.

The dialog library and action library can also be stored in the knowledge map in the form of a triple. For example, the character IP is the character A, and the triplet format of the dialog library corresponding to the character IP in the knowledge map is <person A, dialog library, “DialogDatabase123055601”>, wherein DialogDatabase123055601 represents the dialog library ID corresponding to the character A, which can be used as The index of the character A dialog library, through which the character A's dialog library can be found.

After the knowledge map of the character IP is constructed, the update information of the character image may be collected in real time; and the knowledge map of the corresponding person IP is updated in time according to the update information. In this way, more accurate reply information can be generated, thereby improving the user experience.

The update information refers to information that changes the person information. The knowledge map of the person's IP can be updated in time according to information, news, system push, etc. on the network.

The server can establish a knowledge map with the name of the person as the core node, and provide the chat robot with a knowledge map corresponding to the IP of the character, thereby enabling the chat robot to interact with the user with the specific person IP, and can provide the user with more targeted. Accurate reply information can reduce the number of questions and answers, and thus improve the user experience.

In the present invention, a general knowledge map is used to store some of the well-known knowledge. When establishing a general knowledge map, first of all, it is necessary to obtain common knowledge; then, based on common knowledge, establish a general knowledge map.

In the present invention, the structures of different knowledge maps are consistent and conform to the general definition and structure of the knowledge map.

In the knowledge map, this ternary relationship is generally represented formally by RDF. RDF (Resource Description Framework), the resource description framework, is a standard data model developed by the W3C to describe entities/resources. The knowledge map consists of a number of interconnected entities and their attributes. It is simple to understand that the knowledge map consists of a set of knowledge, each of which is represented as an SPO triple (Subject-Predicate-Object).

After the knowledge map is established, the knowledge map can be stored in the storage device that is provided by the robot, or the knowledge map can be stored in the database of the remote server or the cloud server, if the required knowledge map is not stored in the storage device. In the process of interacting with the user, the corresponding knowledge map can be temporarily retrieved from the database.

After the knowledge map is established, based on the input entity and the user's intention, based on the pre-established knowledge map, generating reply information with a preset default person IP as an image, including: inputting the input entity and the knowledge map The entity in the association is associated; extracting, from the knowledge map, a knowledge map sub-graph related to the input entity and the user intent; and generating, according to the knowledge map sub-graph and the grammar rule, a preset default person IP Image reply message.

Reply message generation process:

The process of associating an input entity with an entity in a knowledge map is: associating an input entity with an entity in a knowledge map to associate the same entity or a similar entity.

After the association, the corresponding plurality of triples may be directly extracted, or the new triplet may be regenerated according to the associated triplet and stored, thereby obtaining a plurality of triples related to the input entity.

Then, from the triples related to the input entity in the knowledge map, the triples related to the user's intention are extracted to form a knowledge map subgraph.

Finally, according to the knowledge map subgraph and grammar rules, the triples in the knowledge map subgraph are split and combined according to the grammar rules, and the reply information with the preset default person IP as the image is generated.

According to the reply information generated by the process, the user's input information can be well matched, and the response is targeted.

In a specific embodiment provided by the present invention, according to the knowledge map sub-graph and the grammar rule, generating reply information with a preset default person IP as an image, including: according to the knowledge map sub-graph and grammar rules, Generating original reply information corresponding to the input information; calculating, according to the weight of the stored historical data and the user's intention, a confidence level of the default character IP in the current scene, and each character's habit appears; determining according to the confidence level Reasonable character habits of the current scene; processing the original reply information according to the reasonable character habits and grammar rules, and generating reply information with a preset default person IP as an image.

In the generation of reply information according to the knowledge map sub-graph and grammar rules, character habits can also be added, so that the reply information is more targeted and more realistic, thereby improving the user experience.

The specific steps are:

In the first step, the original reply information corresponding to the input information is generated according to the knowledge map subgraph and the grammar rule; the specific process is the same as the above process of generating the reply information according to the knowledge map subgraph and the grammar rule.

In the second step, according to the weight of the stored historical data and the user's intention, the confidence of the default character IP in the current scene and the habit of each character is calculated;

Before calculating the confidence level of each character's habit, it is necessary to find the character habits, the character habits can be stored in the knowledge map, or can be stored in the habit subgraphs taken out in advance, which are all within the scope of the present invention. By extracting the habit subgraphs, the efficiency of reply information generation can be improved.

The historical data refers to the spoken data of the stored person IP, and may include: the spoken content and the data of the speaking scene.

Preferably, the weight of the historical data is related to the generation time of the speech data, and the closer the data is, the more the data can reflect the most recent habit of the character image. Therefore, the data weight near the time point can be set high.

Based on the weight of the historical data and the user's intention, the confidence level of each character's habit appears in the current scene.

When calculating the confidence, all the character habits in the current scene in the historical data that match the user's intention are included, including language habits and action habits, and then according to the weight of each character's habits, that is, the weight of the corresponding historical data, calculate each The confidence of a person's habits.

The third step is to determine the reasonable character habits of the current scene based on the calculated confidence.

If it is found that there are multiple character habits used by the user in the current scene, and the confidence of each character's habit is different, then the person with the highest confidence is selected as a reasonable character habit.

When the confidence of the two characters is the same, the two characters can be used as a reasonable character habit, or one of the characters can be randomly selected as a reasonable character habit.

In the fourth step, the original reply information is processed according to reasonable character habits and grammar rules, and a reply message with a preset default person IP as an image is generated.

After selecting a reasonable character habit, the reasonable character habits are added to the original reply information, and the original reply information is processed. In the process of processing, a reply message conforming to the grammatical rules is generated according to the grammar rules.

For example, the IP image of the robot character is the character A, and the input message is: “What is your age?” The original reply message is: “I was born in 1982, 35 years old this year.” Reasonable characters are used to: “Hey, good. "The processed reply message with the character IP is: "Hey, I was born in 1982, 35 years old this year, not bad."

In the present invention, if the character habit is not stored in the knowledge map, it is not necessary to generate reply information in combination with the character habit, and the reply information is generated directly according to the knowledge map subgraph and the grammar rule.

By considering different people's habits in different scenes, it is possible to combine different character habits according to different input information, and output more realistic reply information to the user, thereby improving the user experience.

In a specific embodiment provided by the present invention, the method may further include: acquiring a character IP change instruction input by the user; and setting the person IP from the default person IP to the new person IP according to the change instruction; Retrieving a new knowledge map corresponding to the new person IP; and performing human-computer interaction with the user according to the new knowledge map.

In the process of using the robot, if the user is dissatisfied with the current character image, the IP of the character can also be changed. The current character IP can be the default character IP, or the IP of the person that the user has changed, and can be replaced multiple times.

After the robot receives the character IP change command input by the user, according to the change instruction, the person IP is set from the default character IP to the new person IP, and the new knowledge map corresponding to the new person IP is retrieved from the database, and then according to the new knowledge map. Human-computer interaction with the user, the interaction process is the same as the interaction process of the default character IP.

When the user sends a character IP change command to the robot, the command can be sent by voice, or the command can be sent by other means such as a button or an action. This is all within the scope of the invention.

By switching the character IP of the chat robot in this way, there is no need to change the underlying code, and the IP switching of the character is more convenient, which is advantageous for the promotion and maintenance of the chat robot product.

Step S105: output the reply information to the user by using the default character IP as an image.

Finally, the default person IP is used as the image to output a reply message to the user.

The reply information may include: voice reply information and/or action reply information.

The user chats with the chat bot. When talking, asks the age of the chat bot. The chat bot will answer the age of the currently set person IP. The chat bot can answer his or her age in the form of voice, action (sign language, mouth type), etc. . When the user requests the chat robot to display a certain action, the chat robot can perform multi-modal reply to the user based on the user's input information, based on the custom action of the person's IP stored in the knowledge map, and present the corresponding action to the user.

For example, if the user asks the robot to display a certain action of the character A, such as a performance MV, the robot retrieves the action library of the character A from the knowledge map, and finds the action of the corresponding MV from the action library, and then presents the action to the user.

For example, if the user is asking the robot to meet the companion needs, such as reading, the robot needs to imitate the tone, tone, tone, etc. of the character A's speech. When selecting a book to read, the robot can retrieve the knowledge map, find books that the character A often reads, and/or books that the user likes, and display the book in a voice mode.

For example, if the user asks the robot to display skills, such as writing poetry, first search the character A knowledge map to determine whether the character A has the corresponding skill, and if so, select the work that the character A has written to display to the user (read aloud), if No, the robot responds to the user's "not done / not / not done" according to the response mode and character of the character A's habits.

Example 1: The current character IP of the robot is character A, and the user input information obtained is “What is your age this year?”

The user intent of the robot recognition is "chat", the subject is "machine", and the object is "age", in which both the object and the subject belong to the input entity. Among them, when identifying the subject, the generalization technique is adopted, and "you" is recognized as the subject "machine", that is, "person A".

Based on the input entity and user intent, the robot searches for the age-related triples from the established knowledge map of the character A and the general knowledge map. The search result may be: ("character A", "age", "35" ), ("People A", "Birth Time", "1982"), these two triples can form a knowledge map subgraph, according to the knowledge map subgraph and grammar rules, can generate the original reply message: "I was born In 1982, this year is 35 years old."

Then look up the character A's habits of the character A and the reply habits of the related questions from the knowledge map of the character A, and calculate the confidence of each character's habits, and select the character with higher confidence as a reasonable character habit, for example, only find The idiom habit of character A is “哎呦, not bad”, but the habit of using other related characters is not used, so the terminology is a reasonable character habit. Finally, according to reasonable character habits and grammatical rules, the original reply information is processed, and the resulting reply message is: “Hey, I was born in 1982, 35 years old this year, not bad.”

The robot then responded to the user with the identity of the character A: "Hey, I was born in 1982, 35 years old this year, not bad."

Example 2: The current character IP of the robot is the character B, and the obtained user input information is: "What is the age of the character A this year?"

The user's intention of the robot recognition is "chat", the subject is "person A", and the object is "age", in which both the object and the subject belong to the input entity.

According to the input entity and the user's intention, the robot searches for the triples related to the character A and the age from the knowledge map and the general knowledge map of the established character B. However, the character A and the character B have no social relationship, and finally the general knowledge map. The age information of the character A was found.

When the robot finds that the custom user of the character B is empty, the response information is generated according to the age information of the search and the grammar rule. The final reply message can be: "The character B just asked you, the character A was born in 1982. 35 years old this year."

The robot then responded to the user with the identity of the character B: "Character B just asked you, character A was born in 1982, 35 years old this year."

Example 3: The current character IP of the robot is the character A, and the obtained user input information is “put a song to listen”.

The extracted input entity is: "song", the user's intention is "relaxed listening to the song", and according to the input entity and the user's intention, the knowledge map sub-picture related to the song and the character A is searched from the knowledge map of the character A and the general knowledge map. When the character A is released with a new song, the robot outputs to the user: "Well, listen to my new song XXX."

Through the invention, the human character of the robot can be given, the robot simulates the IP of the specific character, interacts with the user with the identity of the specific person IP, and responds to the user more specifically; and considers the character habit of the character IP to make the robot smarter. Anthropomorphic and better interaction with users can improve the user experience.

The embodiment of the present invention provides a human-computer interaction system based on a person IP, as shown in FIG. 2, which includes an information acquisition module 101, which is used for the human-computer interaction system based on the above-mentioned human IP-based human-computer interaction method. Obtaining input information of the user; the entity extracting module 102 is configured to extract an input entity included in the input information; the intent identifying module 103 is configured to identify a user intent according to the input information; and the reply generating module 104 is configured to: Entering entity and user intent, based on the pre-established knowledge map, generating reply information with a preset default person IP as an image; the information output module 105 is configured to output the reply to the user by using the default person IP as an image information.

The reply generation module 104 includes: an association unit, configured to associate the input entity with an entity in the knowledge map; and an extraction unit, configured to extract, from the knowledge map, the input entity and the user An intent-related knowledge map sub-graph; an information generating unit configured to generate reply information with a preset default person IP as an image according to the knowledge map sub-picture and the grammar rule.

The information generating unit includes: an original information generating subunit, configured to generate original reply information corresponding to the input information according to the knowledge map subgraph and grammar rules; and a confidence calculation subunit for storing according to the The weight of the historical data and the user intent, calculating a confidence that the default character IP is in the current scene, the habit of each character; the habit determining subunit is configured to determine a reasonable character habit of the current scene according to the confidence level; And generating a subunit, configured to process the original reply information according to the reasonable character habit and grammar rules, and generate reply information with a preset default person IP as an image.

In a specific embodiment provided by the present invention, the method may further include: an instruction obtaining module, configured to acquire a character IP change instruction input by the user; and a change module, configured to set the character IP from the default person IP according to the change instruction a new character IP; a knowledge map retrieval module, configured to retrieve a new knowledge map corresponding to the new person IP from the database; and an interaction module, configured to perform human-computer interaction with the user according to the new knowledge map.

The knowledge map of at least one person IP is stored in the database.

The reply information includes: voice response information and/or action response information.

The human-computer interaction system based on the character IP provided by the embodiment of the present invention adds a knowledge map to the robot, and the knowledge map includes: a knowledge map with a character name of the person IP as a core node and a general knowledge map, which can give the robot anthropomorphic Personality, the robot simulates the specific person IP, interacts with the user with the identity of the specific person IP, and responds to the user more specifically; and considers the character habits of the character IP, making the robot more intelligent and anthropomorphic, better with Users interact to improve the user experience.

The embodiment of the present invention provides a computer readable storage medium having a computer program stored thereon, and the program is implemented by the processor to implement the above-mentioned one, based on the same inventive concept as the above-described human IP-based human-computer interaction method. Human-computer interaction method based on character IP.

The computer readable storage medium provided by the embodiment of the present invention can realize the human-computer interaction method based on the character IP under the control of the processor, and can give the robot a personification personality, and the robot simulates the specific person IP to the specific person IP. The identity interacts with the user and responds to the user more specifically; and considers the character habits of the character IP, making the robot more intelligent and anthropomorphic, and better interacting with the user, which can improve the user experience.

Based on the same inventive concept as the above-described human IP-based human-computer interaction method, the embodiment of the present invention further provides a human-computer interaction device based on a person IP, including: a memory, a processor, and a memory and can be stored in the memory A computer program running on a processor, which implements the above-described human IP-based human-computer interaction method when the processor executes the program.

The human-computer interaction device based on the character IP provided by the embodiment of the present invention can realize the above-mentioned human-computer interaction method based on the character IP when the processor works, can give the robot anthropomorphic personality, and enable the robot to simulate the specific person IP to the specific person. The identity of the IP interacts with the user, and responds to the user more specifically; and considers the character habits of the character IP, making the robot more intelligent and anthropomorphic, and better interacting with the user, thereby improving the user experience.

In the description of the invention, numerous specific details are illustrated. However, it is understood that the embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of the description.

In the description of the present specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" and the like means a specific feature described in connection with the embodiment or example. A structure, material or feature is included in at least one embodiment or example of the invention. In the present specification, the schematic representation of the above terms is not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, various embodiments or examples described in the specification, as well as features of various embodiments or examples, may be combined and combined.

Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will understand that The technical solutions described in the foregoing embodiments may be modified, or some or all of the technical features may be equivalently replaced; and the modifications or substitutions do not deviate from the technical solutions of the embodiments of the present invention. The scope is intended to be included within the scope of the claims and the description of the invention.

Claims

A human-computer interaction method based on character IP, characterized in that it comprises:

Obtain the user's input information;

Extracting an input entity included in the input information;

Identifying user intent based on the input information;

And generating, according to the input entity and the user intention, reply information with a preset default person IP as an image based on the pre-established knowledge map;

The reply message is output to the user with the default person IP as an image.
The method according to claim 1, wherein the knowledge map comprises: a knowledge map and/or a general knowledge map with a person name of a default person IP as a core node.
The method according to claim 2, wherein, based on the input entity and the user's intention, based on the pre-established knowledge map, generating reply information with a preset default person IP as an image, including:

Associating the input entity with an entity in the knowledge map;

Extracting, from the knowledge map, a knowledge map subgraph related to the input entity and user intent;

According to the knowledge map sub-graph and the grammar rule, the reply information with the preset default person IP as the image is generated.
The method according to claim 3, wherein the reply information based on the preset default person IP is generated according to the knowledge map sub-graph and the grammar rule, and includes:

Generating original reply information corresponding to the input information according to the knowledge map sub-graph and the grammar rule;

Calculating a confidence level of the default character IP in the current scene according to the weight of the stored historical data and the user intention;

Determining a reasonable character habit of the current scene according to the confidence level;

According to the reasonable person habits and grammar rules, the original reply information is processed to generate reply information with a preset default person IP as an image.
The method of claim 1 further comprising:

Obtaining the character IP change instruction input by the user;

According to the change instruction, the person IP is set from the default person IP to the new person IP;

Retrieving a new knowledge map corresponding to the IP of the new character from the database;

Performing human-computer interaction with the user according to the new knowledge map.
The method according to claim 5, characterized in that the knowledge map of at least one person IP is stored in the database.
The method according to claim 1, wherein the reply information comprises: voice response information and/or action response information.
A human-computer interaction system based on a character IP, characterized in that it comprises:

An information acquisition module, configured to acquire input information of a user;

An entity extraction module, configured to extract an input entity included in the input information;

An intention identification module, configured to identify a user intention according to the input information;

a reply generation module, configured to generate, according to the input entity and the user intention, reply information with a preset default person IP as an image based on the pre-established knowledge map;

The information output module is configured to output the reply information to the user by using the default character IP as an image.
A computer readable storage medium having stored thereon a computer program, characterized in that the program, when executed by a processor, implements the method of any one of claims 1-7.
A human-computer interaction device based on a person IP, comprising: a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the processor executes the program when implementing the claim 1 -7 one of the methods described.