User portrait construction, human-computer interaction method and device, equipment and medium

Abstract

The application provides a user portrait construction, human-computer interaction method, device, equipment and medium, wherein the method comprises: performing speech transcription on historical interaction voice to obtain historical interaction text; extracting attribute data of a first dimension from the historical interaction text based on semantic information of the historical interaction text; counting attribute data of a second dimension based on an interaction topic associated with the historical interaction text; and constructing a user portrait of a user corresponding to the historical interaction voice based on the attribute data of the first dimension and the attribute data of the second dimension. The method, device, electronic equipment and storage medium provided by the application combine the attribute data of the first dimension and the attribute data of the second dimension, so that the content information referenced by the user portrait of the user corresponding to the historical interaction voice is more and more comprehensive, the accuracy and reliability of the user portrait construction are improved, and the construction of the user portrait does not require an interaction interface, thereby reducing the cost of the user portrait construction.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a user profile construction, human-computer interaction method, device, equipment and medium. Background Technology

[0002] With the development of intelligent dialogue technology, intelligent robots have been widely used in many scenarios, such as question-answering robots and chatbots in open scenarios.

[0003] In the process of interaction between users and intelligent robots, the information available for analyzing how to conduct human-computer interaction based solely on conversation content and the user's emotional state is very limited. This results in the content of conversations between artificial intelligence devices sometimes being abrupt, not meeting the user's psychological needs or not closely related to the content the user wants to communicate. The degree of anthropomorphism in the interaction is very limited, making it difficult to meet the user's need for a more human-like interaction. Moreover, the existing human-computer interaction between devices and users lacks significant changes over time. The devices always maintain a certain mode of interaction with users, lacking any "human"-like variation. Summary of the Invention

[0004] This invention provides a user profile construction, human-computer interaction method, device, equipment, and medium to address the shortcomings of existing technologies that rely solely on conversation content and user emotional states for effective human-computer interaction.

[0005] This invention provides a user profile construction method, comprising:

[0006] Transcribe the historical interactive speech to obtain the historical interactive text.

[0007] Based on the semantic information of the historical interaction text, attribute data of the first dimension is extracted from the historical interaction text;

[0008] Based on the interaction topics associated with the historical interaction text, the attribute data of the second dimension is statistically analyzed;

[0009] Based on the attribute data of the first dimension and the attribute data of the second dimension, a user profile of the user corresponding to the historical interactive voice is constructed.

[0010] According to a user profile construction method provided by the present invention, the step of extracting attribute data of the first dimension from the historical interaction text based on the semantic information of the historical interaction text includes:

[0011] Based on the semantic information of the historical interaction text, the content of the historical interaction text is classified to obtain the content type of the historical interaction text;

[0012] Extract entities related to the content type from the historical interactive text, and determine the attribute data of the first dimension based on the entities.

[0013] According to a user profile construction method provided by the present invention, the step of extracting entities related to the content type from the historical interaction text and determining attribute data of the first dimension based on the entities includes:

[0014] Identify candidate entity types related to the content type;

[0015] Entity extraction is performed on the historical interactive text, and target entities whose entity type belongs to the candidate entity type are selected from the extracted entities;

[0016] Based on the target entity and the entity type of the target entity, the attribute data of the first dimension is determined.

[0017] According to a user profile construction method provided by the present invention, the step of statistically analyzing attribute data of the second dimension based on the interaction topics associated with the historical interaction text includes:

[0018] Based on the interaction topics associated with the historical interaction texts, the interaction frequency of each interaction topic is counted.

[0019] The attribute data for the second dimension is determined based on the interaction frequency of each interactive topic.

[0020] According to a user profile construction method provided by the present invention, determining the attribute data of the second dimension based on the interaction frequency of each interactive topic includes:

[0021] The interaction records based on the historical interaction text associated with each interaction topic are statistically analyzed, and the interaction behavior data of each interaction topic are statistically analyzed.

[0022] Based on the interaction frequency of each interactive topic and the interaction behavior data of each interactive topic, the attribute data of the second dimension is determined.

[0023] According to a user profile construction method provided by the present invention, the method further includes, after constructing a user profile of the user corresponding to the historical interactive voice based on the attribute data of the first dimension and the attribute data of the second dimension:

[0024] Identify the missing attributes in the user profile;

[0025] Human-computer interaction is based on the missing attributes.

[0026] The present invention also provides a human-computer interaction method, comprising:

[0027] Get the current interactive voice;

[0028] Determine attribute data related to the current interactive voice from the user profile;

[0029] Human-computer interaction is performed based on the attribute data and the current interactive voice.

[0030] The user profile is determined based on the user profile construction method described in any of the above.

[0031] The present invention also provides a user profile building device, comprising:

[0032] The speech-to-text unit is used to transcribe historical interactive speech to obtain historical interactive text.

[0033] The extraction unit is used to extract attribute data of the first dimension from the historical interaction text based on the semantic information of the historical interaction text;

[0034] The statistical unit is used to statistically analyze the attribute data of the second dimension based on the interaction topics associated with the historical interaction text.

[0035] The construction unit is used to construct a user profile of the user corresponding to the historical interactive voice based on the attribute data of the first dimension and the attribute data of the second dimension.

[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the user profile construction method as described above.

[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the user profile construction method as described above.

[0038] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the user profile construction method as described above.

[0039] The user profile construction, human-computer interaction method, device, equipment, and medium provided by this invention utilize first-dimensional attribute data extracted from historical interaction text based on semantic information. Second-dimensional attribute data is statistically derived from interaction topics associated with the historical interaction text. The combination of these two dimensions provides more comprehensive and abundant reference information for the user profile corresponding to historical interaction voice, improving the accuracy and reliability of user profile construction. Furthermore, since the user profile is constructed based on historical interaction voice, it eliminates the need for an interactive interface, reducing the cost of user profile construction. Subsequent human-computer interaction based on the constructed user profile can meet the user's psychological needs or closely approximate the content the user wants to communicate, increasing the degree of humanization in the interaction and satisfying the user's need for a more human-like experience. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0041] Figure 1 This is a flowchart illustrating the user profile construction method provided by the present invention;

[0042] Figure 2 This is a flowchart illustrating step 120 in the user profile construction method provided by the present invention;

[0043] Figure 3 This is a flowchart illustrating step 122 in the user profile construction method provided by the present invention;

[0044] Figure 4 This is a flowchart illustrating step 130 in the user profile construction method provided by the present invention;

[0045] Figure 5 This is a flowchart illustrating step 132 of the user profile construction method provided by the present invention;

[0046] Figure 6 This is a schematic diagram of the human-computer interaction process for missing attributes provided by the present invention;

[0047] Figure 7 This is a flowchart illustrating the human-computer interaction method provided by the present invention;

[0048] Figure 8 This is a schematic diagram of the user profile building device provided by the present invention;

[0049] Figure 9 This is a schematic diagram of the human-computer interaction device provided by the present invention;

[0050] Figure 10 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0052] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and are not used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and that the objects distinguished by "first," "second," etc., are generally of the same class.

[0053] In related technologies, human-computer interaction refers to the process of information exchange between humans and computers using a certain dialogue language and interactive methods to complete a specific task.

[0054] With the continuous development of artificial intelligence technology, the ways in which humans interact with computers are becoming more and more diverse. Some devices can interact with humans through various means such as text, voice, vision, action, and environment, fully simulating the way humans interact with each other, making the interaction between intelligent robots and humans more intelligent and diversified.

[0055] In the interaction between users and chatbots, the information available for analyzing human-computer interaction methods based solely on conversation content and the user's emotional state is very limited. This results in sometimes abrupt conversations from AI devices that do not meet the user's psychological needs or desired communication, and the degree of humanization in the interaction is very limited, making it difficult to satisfy the user's need for a more human-like experience. Furthermore, the existing human-computer interaction between devices and users lacks significant changes over time. For example, some of the user's attribute data may change during the interaction, but the device maintains a single mode of interaction, lacking any "human"-like variation.

[0056] To address the aforementioned problems, this invention provides a user profile construction method. The user profile constructed in this method can be applied to human-computer interaction to achieve a "human-like" interactive experience. It is understood that the user profile constructed in this way can also be used for friend recommendations, application recommendations, etc., and this invention does not specifically limit its use in this regard.

[0057] Figure 1 This is a flowchart illustrating the user profile construction method provided by the present invention, as shown below. Figure 1 As shown, the method includes:

[0058] Step 110: Transcribe the historical interactive speech to obtain the historical interactive text.

[0059] Specifically, in order to reduce the cost of user profile construction, the user profile in this embodiment of the invention is constructed based on historical interactive voice, so that the construction of user profile does not require an interactive interface, thereby reducing the cost of user profile construction. Furthermore, since it is based solely on voice, it can be applied to devices without interactive interfaces, such as smart cameras and smart speakers, thus expanding the application scenarios of profile construction.

[0060] The historical interactive voice can be transcribed to obtain historical interactive text. Here, historical interactive voice refers to interactive voice obtained in previous human-computer interaction processes. Historical interactive voice can be obtained through a sound pickup device, which can be a smart camera, a smart speaker, a smart air conditioner, etc. After the sound pickup device obtains the historical interactive voice through a microphone array, it can also amplify and reduce noise. This embodiment of the invention does not specifically limit this.

[0061] The historical interactive text here refers to the text obtained by transcribing the historical interactive speech. The historical interactive text can be "My favorite animal is a puppy", "My favorite song is Sunny Day", or "My favorite drink is a full cup of passion fruit", etc. This embodiment of the invention does not make specific limitations on this.

[0062] Step 120: Based on the semantic information of the historical interaction text, extract the attribute data of the first dimension from the historical interaction text.

[0063] Specifically, after obtaining the historical interaction text, the first-dimensional attribute data can be extracted from it based on its semantic information. The semantic information of the historical interaction text reflects the semantic level of the historical interaction text.

[0064] For example, based on the semantic information of historical interactive texts, the content of historical interactive texts can be classified to obtain the content type of historical interactive texts, and then the attribute data of the first dimension can be extracted from the historical interactive texts based on the content type of historical interactive texts.

[0065] The attribute data of the first dimension here may include basic attribute data and social attribute data, or interest preference attribute data and content preference attribute data, or basic attribute data, social attribute data, interest preference attribute data and content preference attribute data, etc. The embodiments of the present invention do not specifically limit this.

[0066] Step 130: Based on the interaction topics associated with the historical interaction text, statistically analyze the attribute data of the second dimension.

[0067] Specifically, attribute data for the second dimension can be statistically analyzed based on the interaction topics associated with the historical interaction text. The interaction topics associated with the historical interaction text can be topics about nursery rhyme preferences, game preferences, or professions, etc. This embodiment of the invention does not impose specific limitations on this.

[0068] For example, attribute data for the second dimension can be determined based on the frequency of interactions with topics associated with historical interactive text.

[0069] The attribute data of the second dimension here may include long-term preference attribute data and short-term preference attribute data, or long-term preference attribute data and interaction behavior data, or long-term preference attribute data, short-term preference attribute data and interaction behavior data, etc. The embodiments of the present invention do not make specific limitations in this regard.

[0070] Step 140: Based on the attribute data of the first dimension and the attribute data of the second dimension, construct a user profile of the user corresponding to the historical interactive voice.

[0071] Specifically, after obtaining the attribute data of the first dimension and the attribute data of the second dimension, a user profile of the user corresponding to the historical interactive voice can be constructed based on the attribute data of the first dimension and the attribute data of the second dimension.

[0072] The user profile here refers to a labeled user model abstracted from information such as user attributes, user preferences, lifestyle habits, and user behavior.

[0073] Understandably, the combination of the first-dimensional attribute data and the second-dimensional attribute data makes the user profile reference content more comprehensive and complete, thus improving the accuracy and reliability of user profile construction.

[0074] Understandably, the attribute data in both the first and second dimensions will be continuously updated over time. During subsequent interactions, some user attribute data, such as those related to friends, teachers, pets, long-term preferences, and short-term preferences, will change. Therefore, the user profiles corresponding to historical interactive voices constructed based on the first and second dimension attribute data will also be continuously updated as human-computer interaction progresses, further improving the accuracy and reliability of the constructed user profiles. Furthermore, subsequent human-computer interactions based on these user profiles can better meet users' psychological needs or closely approximate the content they want to communicate, increasing the degree of humanization in the interaction and satisfying users' needs for a more personalized experience.

[0075] The method provided in this invention uses first-dimensional attribute data extracted from historical interaction text based on semantic information. Second-dimensional attribute data is statistically derived from interaction topics associated with the historical interaction text. The combination of these two dimensions provides more comprehensive and detailed information for the user profile referenced by the historical interaction voice, improving the accuracy and reliability of user profile construction. Furthermore, since the user profile is built based on historical interaction voice, it eliminates the need for an interactive interface, reducing the cost of user profile construction. Subsequent human-computer interaction based on the constructed user profile can meet the user's psychological needs or closely approximate the content the user wants to communicate, increasing the degree of humanization in the interaction and satisfying the user's need for a more personalized experience.

[0076] Based on the above embodiments, Figure 2 This is a flowchart illustrating step 120 of the user profile construction method provided by the present invention, as shown below. Figure 2 As shown, step 120 includes:

[0077] Step 121: Based on the semantic information of the historical interaction text, classify the content of the historical interaction text to obtain the content type of the historical interaction text;

[0078] Step 122: Extract entities related to the content type from the historical interactive text, and determine the attribute data of the first dimension based on the entities.

[0079] Specifically, after obtaining the historical interaction text, the content of the historical interaction text can be classified based on its semantic information to obtain the content type of the historical interaction text.

[0080] The content type of the historical interactive text here reflects the category to which the historical interactive text belongs. The content type of the historical interactive text may include basic attributes, social attributes, interest preference attributes and content preference attributes, or basic attributes, social attributes, interest preference attributes and content preference attributes, etc. The embodiments of the present invention do not make specific limitations in this regard.

[0081] After obtaining the content type, entities related to the content type can be extracted from the historical interaction text. For example, if the content type is a basic attribute, entities related to the content type extracted from the historical interaction text may include the user's name, gender, and age, as well as the user's birthday, zodiac sign, and occupation. They may also include the user's height, weight, address, school, self-description of personality, etc. This embodiment of the invention does not impose specific limitations on these aspects.

[0082] For example, if the content type is social, entities related to the content type extracted from historical interactive text can include the user's family and friends, the user's classmates and teachers, as well as the user's family, friends, classmates, teachers, neighbors, pets, etc. This embodiment of the invention does not specifically limit this.

[0083] For example, if the content type is an interest preference attribute, entities related to the content type extracted from historical interactive text may include food and beverages, movies and animations, or food, beverages, songs, stories, toys and animals, etc. This embodiment of the invention does not specifically limit this.

[0084] For example, the content type is a content preference attribute. Entities related to the content type extracted from historical interactive text may include topic scene preferences and story preferences, as well as nursery rhyme preferences and game preferences. They may also include scene preferences, story preferences, nursery rhyme preferences, game preferences, active request for XX, saying "I like XX the most", "favorite XX", etc. The embodiments of the present invention do not make specific limitations in this regard.

[0085] After extracting entities related to content type from historical interactive text, attribute data for the first dimension can be determined based on the entities.

[0086] Accordingly, the attribute data of the first dimension may include basic attribute data and social attribute data, or interest preference attribute data and content preference attribute data, or basic attribute data, social attribute data, interest preference attribute data and content preference attribute data, etc., and the embodiments of the present invention do not specifically limit this.

[0087] The method provided in this invention classifies historical interaction texts based on their semantic information to obtain content types. It then extracts entities related to the content types from the historical interaction texts and determines attribute data for the first dimension based on these entities. This process only extracts entity information with specific meaning related to the content types from the historical interaction texts, thereby further improving the reliability of the determined attribute data for the first dimension and thus improving the accuracy and reliability of subsequent user profile construction.

[0088] Based on the above embodiments, Figure 3 This is a flowchart illustrating step 122 of the user profile construction method provided by the present invention, as shown below. Figure 3 As shown, step 122 includes:

[0089] Step 1221: Determine the candidate entity types related to the content type;

[0090] Step 1222: Extract entities from the historical interaction text and select target entities whose entity type belongs to the candidate entity type from the extracted entities;

[0091] Step 1223: Based on the target entity and the entity type of the target entity, determine the attribute data of the first dimension.

[0092] Specifically, candidate entity types related to the content type can be determined. Here, candidate entity type refers to the type to which the candidate entity related to the content type belongs. For example, if the content type is a basic attribute, the candidate entity type related to the content type may include the user's name, gender, and age, as well as the user's birthday, zodiac sign, and occupation. It may also include the user's height, weight, address, school, self-description of personality, etc. This embodiment of the invention does not make specific limitations in this regard.

[0093] For example, if the content type is a social attribute, the candidate entity types related to the content type can include the user's family and friends, the user's classmates and teachers, as well as the user's family, friends, classmates, teachers, neighbors, pets, etc. This embodiment of the invention does not specifically limit these.

[0094] Entity extraction can be performed on historical interactive text, and target entities whose entity type belongs to the candidate entity type can be selected from the extracted entities. Here, entity extraction on historical interactive text can be performed using LSTM-RNN (Long Short-Term Memory networks-Recurrent Neural Network) or BERT (Bidirectional Encoder Representation from Transformers) model, etc. The embodiments of the present invention do not specifically limit this.

[0095] For example, if the candidate entity type is the user's pet, and the entity is extracted from the historical interaction text, and the extracted entities are puppy, dad, and school, then the target entity whose entity type belongs to the candidate entity type is selected from the extracted entities as puppy.

[0096] For example, if the candidate entity type is the user's zodiac sign, and the entity is extracted from the historical interaction text, the extracted entities are kitten, Cancer, and younger brother. Then, the target entity whose entity type belongs to the candidate entity type is selected from the extracted entities as Cancer.

[0097] For example, if the candidate entity type is the user's occupation, and the historical interaction text is used to extract entities, and the extracted entities are teacher, XX school, and classmate, then the target entity whose entity type belongs to the candidate entity type is selected from the extracted entities as teacher.

[0098] After obtaining the target entity, the attribute data of the first dimension can be determined based on the target entity and its entity type.

[0099] For example, if the target entity is a puppy and the entity type of the target entity is the user's pet, then based on the target entity and the entity type of the target entity, the attribute data of the first dimension can be determined to be that the user's pet is a puppy.

[0100] For example, if the target entity is Cancer and the entity type of the target entity is the user's zodiac sign, then based on the target entity and the entity type of the target entity, the attribute data of the first dimension can be determined to be the user's zodiac sign as Cancer.

[0101] For example, if the target entity is a teacher and the entity type of the target entity is the user's occupation, then based on the target entity and the entity type of the target entity, the attribute data of the first dimension can be determined to be the user's occupation as a teacher.

[0102] The method provided in this embodiment of the invention determines the attribute data of the first dimension based on the target entity and the entity type of the target entity, thereby ensuring the reliability and accuracy of determining the attribute data of the first dimension.

[0103] Based on the above embodiments, Figure 4 This is a flowchart illustrating step 130 of the user profile construction method provided by the present invention, as shown below. Figure 4 As shown, step 130 includes:

[0104] Step 131: Based on the interaction topics associated with the historical interaction text, count the interaction frequency of each interaction topic.

[0105] Step 132: Determine the attribute data of the second dimension based on the interaction frequency of each interactive topic.

[0106] Specifically, the interaction frequency of each interaction topic can be counted based on the interaction topics associated with the historical interaction text. The interaction frequency of each interaction topic refers to the number of times the user interacts with the intelligent robot based on the current interaction topic. The interaction frequency of each interaction topic can be three months, seven days, or five minutes, etc. The embodiments of the present invention do not make specific limitations on this.

[0107] After obtaining the interaction frequency of each interactive topic, the attribute data of the second dimension can be determined based on the interaction frequency of each interactive topic.

[0108] The attribute data of the second dimension here may include long-term preference attribute data and short-term preference attribute data, etc., and the embodiments of the present invention do not specifically limit this.

[0109] The long-term preference attribute data here refers to a user's stable preferences for various interactive topics over a relatively long period of time, such as several months or even years. The short-term preference attribute data here refers to a user's preferences for various interactive topics in the most recent short period of time, such as seven days or even a few minutes.

[0110] Based on the above embodiments, Figure 5 This is a flowchart illustrating step 132 of the user profile construction method provided by the present invention, as shown below. Figure 5 As shown, step 132 includes:

[0111] Step 1321: Statistically analyze the interaction records based on the historical interaction texts associated with each interaction topic, and statistically analyze the interaction behavior data of each interaction topic;

[0112] Step 1322: Determine the attribute data of the second dimension based on the interaction frequency of each interactive topic and the interaction behavior data of each interactive topic.

[0113] Specifically, interaction records based on historical interaction texts associated with each interaction topic can be statistically analyzed, and interaction behavior data for each interaction topic can be statistically analyzed. Here, interaction behavior data refers to attribute data at the behavioral level during the interaction between the user and the intelligent robot. This may include the commonly used time period of the interaction, the average duration of the interaction, the frequency of the interaction, the depth of the interaction topic (determined by the number of interaction rounds of the same interaction topic), and may also include the commonly used time period of the interaction, the average duration of the interaction, the frequency of the interaction, the depth of the interaction topic, the number of interaction rounds, the degree of cooperation of the interaction topic, the content of the most recent interaction, etc. The embodiments of the present invention do not make specific limitations on this.

[0114] The degree of engagement between interactive topics here refers to the extent to which the interactive topics match the user's expected topics.

[0115] After obtaining the interaction behavior data for each interactive topic, the attribute data for the second dimension can be determined based on the interaction frequency and interaction behavior data for each interactive topic.

[0116] That is, the attribute data of the second dimension here may include interaction behavior data and short-term preference attribute data, or long-term preference attribute data and interaction behavior data, or long-term preference attribute data, short-term preference attribute data and interaction behavior data, etc. The embodiments of the present invention do not make specific limitations in this regard.

[0117] The method provided in this invention determines the attribute data of the second dimension based on the interaction frequency of each interactive topic and the interaction behavior data of each interactive topic, thereby further improving the reliability and accuracy of determining the attribute data of the second dimension.

[0118] Based on the above embodiments, Figure 6 This is a flowchart illustrating the human-computer interaction process for missing attributes provided by the present invention, as shown below. Figure 6 As shown, step 140, followed by:

[0119] Step 141: Determine the missing attributes in the user profile;

[0120] Step 142: Perform human-computer interaction based on the missing attributes.

[0121] Specifically, after constructing user profiles for users corresponding to historical interactive voice messages, missing attributes in the user profiles can be identified. Here, missing attributes refer to attributes that are missing from the user profiles.

[0122] After identifying the missing attributes in the user profile, human-computer interaction can be performed based on the missing attributes, and the user voice obtained based on the human-computer interaction can be used as new historical interaction voice to complete the attribute data of the missing attributes in the user profile.

[0123] For example, if the missing attribute in a user profile is the user's name, the intelligent robot can ask the user "What is your name?" After the user answers with their name, the user's name can be collected and stored in the user profile.

[0124] The method provided in this invention identifies missing attributes in a user profile and then performs human-computer interaction based on these missing attributes, thereby completing the missing attributes in the user profile, improving the completeness of user profile construction, and further enhancing the accuracy and reliability of user profile construction.

[0125] Based on any of the above embodiments, the present invention provides a human-computer interaction method. Figure 7 This is a flowchart illustrating the human-computer interaction method provided by the present invention, as shown below. Figure 7 As shown, the method includes:

[0126] Step 710: Obtain the current interactive voice;

[0127] Specifically, the current interactive voice can be acquired. The current interactive voice refers to the interactive voice obtained during the current human-computer interaction process. The current interactive voice can be acquired through a sound pickup device, which can be a smart home camera, a smart speaker, a smart air conditioner, etc. After the sound pickup device acquires the current interactive voice through a microphone array, it can also amplify and reduce noise. This embodiment of the invention does not specifically limit this.

[0128] The current interactive voice here can also be the interactive voice from the previous round or multiple rounds in a human-computer interaction, and the embodiments of the present invention do not specifically limit this.

[0129] Step 720: Determine attribute data related to the current interactive voice from the user profile;

[0130] Step 730: Perform human-computer interaction based on the attribute data and the current interactive voice;

[0131] The user profile is determined based on the user profile construction method described above.

[0132] Specifically, after obtaining the current interactive voice, attribute data related to the current interactive voice can be determined from the user profile.

[0133] The user profile here refers to tagged user data abstracted from information such as user attributes, preferences, lifestyle habits, and behaviors. It includes attribute data from both the first and second dimensions.

[0134] The attribute data of the first dimension here may include basic attribute data and social attribute data, or interest preference attribute data and content preference attribute data, or basic attribute data, social attribute data, interest preference attribute data and content preference attribute data, etc. The embodiments of the present invention do not specifically limit this.

[0135] The attribute data of the second dimension here may include long-term preference attribute data and short-term preference attribute data, or long-term preference attribute data and interaction behavior data, or long-term preference attribute data, short-term preference attribute data and interaction behavior data, etc. The embodiments of the present invention do not make specific limitations in this regard.

[0136] The attribute data in the first dimension can be obtained by transcribing historical interactive speech to obtain historical interactive text, and then extracting it from the historical interactive text based on the semantic information of the historical interactive text.

[0137] The attribute data for the second dimension can be obtained by statistically analyzing the interactive topics associated with historical interactive texts.

[0138] Understandably, user profiles that encompass attribute data from both the first and second dimensions will be continuously updated over time. During subsequent interactions, user profiles, such as those related to friends, teachers, pets, long-term preferences, and short-term preferences, will change. Thus, user profiles are continuously updated as human-computer interaction progresses, enabling human-computer interaction based on the constructed user profiles to meet users' psychological needs or closely approximate the content users want to communicate, thereby improving the anthropomorphism of the interaction and satisfying users' needs for a more human-like experience.

[0139] Accordingly, the attribute data related to the current interactive voice determined from the user profile may include basic attribute data, social attribute data, long-term preference attribute data, short-term preference attribute data, and interactive behavior data, or may include basic attribute data, social attribute data, interest preference attribute data, content preference attribute data, long-term preference attribute data, short-term preference attribute data, and interactive behavior data, etc. The embodiments of the present invention do not specifically limit this.

[0140] In addition, actions can also be generated when human-computer interaction is based on attribute data. Here, actions refer to the user's expressive actions (body language), the user's actions towards the intelligent robot (such as touching, patting, etc.), and the intelligent robot's actions towards the human (body expressions, device movements, etc.).

[0141] After obtaining the attribute data, human-computer interaction can be performed based on the attribute data and the current interactive voice. That is, human-computer interaction can include voice interaction, expressing actions, etc. This embodiment of the invention does not make specific limitations on this.

[0142] In addition, attribute data can be updated based on attribute data and current interactive voice to ensure the timeliness of attribute data (i.e., the timeliness of user profiles), thereby ensuring the accuracy of human-computer interaction.

[0143] During this stage, the intelligent robot has accumulated relatively rich attribute data, and the relationship between the intelligent robot and the user gradually becomes closer. The intelligent robot can interact with the user in the user's comfort zone based on the user profile.

[0144] Throughout the interaction process, the system gradually adjusts as the user profile becomes more comprehensive. The intelligent robot improves its interaction strategies by mimicking how people gradually get to know each other, become familiar with each other, and develop closer relationships. This enhances the "human-like" responses of the intelligent robot.

[0145] The method provided in this invention determines attribute data related to the current interactive voice from the user profile, and then performs human-computer interaction based on the attribute data and the current interactive voice. This can meet the user's psychological needs or closely approximate the content the user wants to communicate, improve the degree of humanization of the interaction, and meet the user's need for a more human-like interaction.

[0146] Based on any of the above embodiments, a user profile construction method includes the following steps:

[0147] The first step is to transcribe the historical interactive speech to obtain the historical interactive text.

[0148] The second step is to classify the historical interactive text based on its semantic information to obtain the content type of the historical interactive text, and then determine the candidate entity type related to the content type. Next, entity extraction is performed on the historical interactive text, and target entities whose entity type belongs to the candidate entity type are selected from the extracted entities. Finally, based on the target entity and its entity type, the attribute data of the first dimension is determined.

[0149] The third step is to count the interaction frequency of each interaction topic based on the interaction topics associated with the historical interaction texts, then count the interaction records of each interaction topic based on the historical interaction texts associated with the interaction topics, and finally, determine the attribute data of the second dimension based on the interaction frequency and the interaction behavior data of each interaction topic.

[0150] The fourth step is to construct user profiles for users corresponding to historical interactive voice messages, based on the attribute data of the first dimension and the attribute data of the second dimension.

[0151] The fifth step is to identify the missing attributes in the user profile.

[0152] The sixth step is to conduct human-computer interaction based on the missing attributes to complete the user profile.

[0153] The user profile building apparatus provided by the present invention is described below. The user profile building apparatus described below and the user profile building method described above can be referred to in correspondence.

[0154] Based on any of the above embodiments, the present invention provides a user profile construction device. Figure 8 This is a schematic diagram of the user profile building device provided by the present invention, as shown below. Figure 8 As shown, the device includes:

[0155] The speech transcription unit 810 is used to transcribe historical interactive speech to obtain historical interactive text.

[0156] Extraction unit 820 is used to extract attribute data of the first dimension from the historical interaction text based on the semantic information of the historical interaction text;

[0157] The statistical unit 830 is used to statistically analyze the attribute data of the second dimension based on the interactive topics associated with the historical interactive text.

[0158] The construction unit 840 is used to construct a user profile of the user corresponding to the historical interactive voice based on the attribute data of the first dimension and the attribute data of the second dimension.

[0159] The device provided in this embodiment of the invention uses first-dimensional attribute data extracted from historical interactive text based on semantic information, and second-dimensional attribute data statistically derived from interactive topics associated with the historical interactive text. The combination of the first-dimensional and second-dimensional attribute data provides more comprehensive and complete user profile reference information corresponding to historical interactive voice, improving the accuracy and reliability of user profile construction. Furthermore, since the user profile is constructed based on historical interactive voice, it eliminates the need for an interactive interface, reducing the cost of user profile construction. Subsequent human-computer interaction based on the constructed user profile can meet the user's psychological needs or closely approximate the content the user wants to communicate, improving the anthropomorphism of the interaction and satisfying the user's need for a more human-like experience.

[0160] Based on any of the above embodiments, the extraction unit 820 is specifically used for:

[0161] The content classification unit is used to classify the historical interactive text based on the semantic information of the historical interactive text to obtain the content type of the historical interactive text;

[0162] A first dimension unit is determined, which is used to extract entities related to the content type from the historical interactive text, and to determine the attribute data of the first dimension based on the entities.

[0163] Based on any of the above embodiments, the first dimension unit is specifically used for:

[0164] Identify candidate entity types related to the content type;

[0165] Entity extraction is performed on the historical interactive text, and target entities whose entity type belongs to the candidate entity type are selected from the extracted entities;

[0166] Based on the target entity and the entity type of the target entity, the attribute data of the first dimension is determined.

[0167] Based on any of the above embodiments, the statistical unit 830 is specifically used for:

[0168] The interaction frequency counting unit is used to count the interaction frequency of each interaction topic based on the interaction topics associated with the historical interaction text.

[0169] A second dimension unit is determined, which is used to determine the attribute data of the second dimension based on the interaction frequency of each interactive topic.

[0170] Based on any of the above embodiments, the second dimension unit is specifically used for:

[0171] The interaction records based on the historical interaction text associated with each interaction topic are statistically analyzed, and the interaction behavior data of each interaction topic are statistically analyzed.

[0172] Based on the interaction frequency of each interactive topic and the interaction behavior data of each interactive topic, the attribute data of the second dimension is determined.

[0173] Based on any of the above embodiments, the construction unit 840 is further configured to:

[0174] Identify the missing attributes in the user profile;

[0175] Human-computer interaction is based on the missing attributes.

[0176] The human-computer interaction device provided by the present invention is described below. The human-computer interaction device described below and the human-computer interaction method described above can be referred to in correspondence.

[0177] Based on any of the above embodiments, the present invention provides a human-computer interaction device. Figure 9 This is a structural schematic diagram of the human-computer interaction device provided by the present invention, as shown below. Figure 9 As shown, the device includes:

[0178] Acquisition unit 910 is used to acquire the current interactive voice;

[0179] The determining unit 920 is used to determine attribute data related to the current interactive voice from the user profile;

[0180] The human-computer interaction unit 930 is used to perform human-computer interaction based on the attribute data and the current interactive voice.

[0181] The user profile is determined based on the user profile construction method described above.

[0182] The device provided in this embodiment of the invention determines attribute data related to the current interactive voice from the user profile, and then performs human-computer interaction based on the attribute data and the current interactive voice. This can meet the user's psychological needs or approximate the content the user wants to communicate, improve the degree of humanization of the interaction, and meet the user's need for a more human-like interaction.

[0183] Figure 10 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 10 As shown, the electronic device may include a processor 1010, a communications interface 1020, a memory 1030, and a communication bus 1040, wherein the processor 1010, the communications interface 1020, and the memory 1030 communicate with each other via the communication bus 1040. The processor 1010 can call logical instructions in the memory 1030 to execute a user profile construction method, which includes: transcribing historical interactive speech to obtain historical interactive text; extracting first-dimensional attribute data from the historical interactive text based on the semantic information of the historical interactive text; calculating second-dimensional attribute data based on the interactive topics associated with the historical interactive text; and constructing a user profile of the user corresponding to the historical interactive speech based on the first-dimensional attribute data and the second-dimensional attribute data.

[0184] The processor 1010 can also call logical instructions in the memory 1030 to execute a human-computer interaction method, which includes: acquiring the current interactive voice; determining attribute data related to the current interactive voice from a user profile; performing human-computer interaction based on the attribute data and the current interactive voice; the user profile is determined based on the aforementioned user profile construction method. Furthermore, the logical instructions in the memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0185] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the user profile construction method provided by the above methods. The method includes: transcribing historical interactive speech to obtain historical interactive text; extracting first-dimensional attribute data from the historical interactive text based on the semantic information of the historical interactive text; calculating second-dimensional attribute data based on the interactive topics associated with the historical interactive text; and constructing a user profile of the user corresponding to the historical interactive speech based on the first-dimensional attribute data and the second-dimensional attribute data.

[0186] When the computer program is executed by the processor, the computer can execute the human-computer interaction methods provided by the above methods, which include: acquiring the current interactive voice; determining attribute data related to the current interactive voice from the user profile; performing human-computer interaction based on the attribute data and the current interactive voice; the user profile is determined based on the above user profile construction method.

[0187] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the user profile construction method provided by the above methods. The method includes: transcribing historical interactive speech to obtain historical interactive text; extracting attribute data of a first dimension from the historical interactive text based on the semantic information of the historical interactive text; calculating attribute data of a second dimension based on the interactive topics associated with the historical interactive text; and constructing a user profile of the user corresponding to the historical interactive speech based on the attribute data of the first dimension and the attribute data of the second dimension.

[0188] When executed by a processor, the computer program implements the human-computer interaction methods provided by the above methods, the method comprising: acquiring current interactive voice; determining attribute data related to the current interactive voice from a user profile; performing human-computer interaction based on the attribute data and the current interactive voice; the user profile being determined based on the above user profile construction method.

[0189] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0190] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A user portrait construction method, characterized in that, The method comprises the following steps: transcribing historical interaction voice into historical interaction text; extracting first-dimension attribute data from the historical interaction text based on semantic information of the historical interaction text; the first-dimension attribute data comprises basic attribute data, social attribute data, interest preference attribute data and content preference attribute data; statistically determining second-dimension attribute data based on interaction topics associated with the historical interaction text; the second-dimension attribute data comprises long-term preference attribute data, short-term preference attribute data and interaction behavior data; constructing a user portrait of a user corresponding to the historical interaction voice based on the first-dimension attribute data and the second-dimension attribute data; after constructing the user portrait of the user corresponding to the historical interaction voice based on the first-dimension attribute data and the second-dimension attribute data, the method further comprises the following steps: determining missing attributes in the user portrait; performing human-computer interaction based on the missing attributes. 2.The user portrait construction method of claim 1, wherein, The method of extracting first-dimension attribute data from the historical interaction text based on semantic information of the historical interaction text comprises the following steps: performing content classification on the historical interaction text based on the semantic information of the historical interaction text to obtain a content type of the historical interaction text; extracting an entity related to the content type from the historical interaction text, and determining the first-dimension attribute data based on the entity. 3.The user portrait construction method of claim 2, wherein, The method of extracting an entity related to the content type from the historical interaction text, and determining the first-dimension attribute data based on the entity comprises the following steps: determining a candidate entity type related to the content type; performing entity extraction on the historical interaction text, and selecting a target entity whose entity type belongs to the candidate entity type from the extracted entities; determining the first-dimension attribute data based on the target entity and the entity type of the target entity. 4.The user portrait construction method of claim 1, wherein, The method of statistically determining second-dimension attribute data based on interaction topics associated with the historical interaction text comprises the following steps: statistically determining interaction frequencies of the interaction topics based on the interaction topics associated with the historical interaction text; determining the second-dimension attribute data based on the interaction frequencies of the interaction topics. 5.The user portrait construction method of claim 4, wherein, The method of determining the second-dimension attribute data based on the interaction frequencies of the interaction topics comprises the following steps: statistically determining interaction behavior data of the interaction topics based on interaction records of the historical interaction text associated with the interaction topics; determining the second-dimension attribute data based on the interaction frequencies of the interaction topics and the interaction behavior data of the interaction topics.

6. A human-machine interaction method characterized by, The method comprises the following steps: obtaining current interaction voice; determining attribute data related to the current interaction voice from a user portrait; performing human-computer interaction based on the attribute data and the current interaction voice; the user portrait is determined based on the user portrait construction method in any one of claims 1 to 5. 7.A user profiling apparatus, characterized by comprising: The method comprises the following steps: a voice transcription unit configured to transcribe historical interaction voice into historical interaction text; The extraction unit is configured to extract attribute data of a first dimension from the historical interaction text based on semantic information of the historical interaction text, and the attribute data of the first dimension includes basic attribute data, social attribute data, interest preference attribute data, and content preference attribute data. The statistics unit is configured to count attribute data of a second dimension based on an interaction topic associated with the historical interaction text, and the attribute data of the second dimension includes long-term preference attribute data, short-term preference attribute data, and interaction behavior data. The construction unit is configured to construct a user portrait of a user corresponding to the historical interaction voice based on the attribute data of the first dimension and the attribute data of the second dimension. The interaction unit is further configured to: determine a missing attribute in the user portrait; and perform human-computer interaction based on the missing attribute.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the user portrait construction method according to any one of claims 1 to 6 when executing the program. 9.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program implements the user portrait construction method according to any one of claims 1 to 6 when executed by the processor.

Images