CN110472008B

CN110472008B - Intelligent interaction method and device

Info

Publication number: CN110472008B
Application number: CN201910600784.4A
Authority: CN
Inventors: 陈鑫
Original assignee: Advanced New Technologies Co Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2019-07-04
Filing date: 2019-07-04
Publication date: 2023-04-18
Anticipated expiration: 2039-07-04
Also published as: CN110472008A

Abstract

The present specification provides an intelligent interaction method and an intelligent interaction device, wherein the intelligent interaction method includes: receiving an interaction request of a user, wherein the interaction request carries an interaction instruction between the user and the target object; inputting an interactive instruction between the user and the target object into a pre-trained learning algorithm model; outputting adaptation information corresponding to the interaction instruction and related to the target object through the learning algorithm model; and sending the adaptation information to the user. The interactive instruction in the interactive request is input into the pre-trained learning algorithm model, and the corresponding adaptation information is output, so that the interactive effectiveness is favorably improved, and meanwhile, the accuracy of the adaptation information is favorably improved.

Description

Intelligent interaction method and device

Technical Field

The specification relates to the technical field of robots, in particular to an intelligent interaction method. The present specification also relates to an intelligent interaction device, an electronic apparatus, and a computer-readable storage medium.

Background

With the rapid development of economy, the attention of user groups to insurance is higher and higher, and the number of insured people is more and more, so that in order to enable users to complete insurance more quickly and conveniently, the online insurance service platform is different from the traditional online insurance purchasing mode, and a plurality of online insurance service platforms are produced accordingly.

The insurance sales modes are more and more diversified, however, no matter the user applies insurance through an insurance company under the line or applies insurance through an insurance service platform on the line, the user needs to know the insurance through a certain channel before the official insurance application and knows the required dangerous species of the user.

However, in the prior art, the way of positioning answer content required by a user in a knowledge base according to an identification result depends more on the integrity of the knowledge base and the professional knowledge richness of an agent, and certain defects exist in the aspects of accuracy of intelligent information query and recommendation.

Disclosure of Invention

In view of this, the embodiments of the present specification provide an intelligent interaction method. The present specification also relates to an intelligent interaction device, an electronic apparatus, and a computer-readable storage medium, which are used to solve the technical defects in the prior art.

According to a first aspect of embodiments of the present specification, there is provided an intelligent interaction method, including:

receiving an interaction request of a user, wherein the interaction request carries an interaction instruction between the user and the target object;

inputting an interactive instruction between the user and the target object into a pre-trained learning algorithm model;

outputting adaptation information corresponding to the interaction instruction and related to the target object through the learning algorithm model;

and sending the adaptation information to the user.

Optionally, the learning algorithm model is trained by:

acquiring user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object and case data related to the target object;

training a learning algorithm model by taking user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object and case data related to the target object as training samples.

Optionally, after the step of receiving the interaction request of the user is executed, the method further includes:

acquiring the identity identification information of the user according to a preset identity recognition algorithm;

if the user is determined to be the target user according to the identity identification information, learning prompt information is sent to the user;

detecting a response result of the target user to the learning prompt information;

and if the response result is detected to be learning confirmation, displaying information related to the target object for the target user through a system interactive interface.

Optionally, after the step of sending the adaptation information to the user is executed, the method further includes:

collecting an interactive instruction between the user and the target object and adaptation information corresponding to the interactive instruction according to a preset condition;

and adding the interactive instruction between the user and the target object and the adaptive information corresponding to the interactive instruction to the training sample to form a new training sample, and performing model optimization on the learning algorithm model based on the new training sample.

collecting an interactive instruction between the user and the target object, adaptation information corresponding to the interactive instruction and a feedback grading result of the user on the adaptation information corresponding to the interactive instruction according to a preset condition;

and adding the interactive instruction between the user and the target object with the feedback scoring result higher than a preset score threshold value and the adaptive information corresponding to the interactive instruction to the training sample to form a new training sample, and performing model optimization on the learning algorithm model based on the new training sample.

Optionally, the interactive instruction is voice question information;

after the step of receiving the interaction request of the user is executed, the method further comprises the following steps:

carrying out voice recognition on the voice question information to obtain corresponding text information;

the inputting of the interaction instruction between the user and the target object into the pre-trained learning algorithm model comprises:

and inputting the text information into a pre-trained learning algorithm model.

Optionally, before the substep of inputting the text information into the pre-trained learning algorithm model is executed, the method further includes:

performing semantic analysis on the text information to obtain a corresponding semantic analysis result;

the inputting of the text information into a pre-trained learning algorithm model comprises:

and inputting the semantic analysis result into a pre-trained learning algorithm model.

extracting keywords from the text information according to a preset rule;

and inputting the extracted keywords into a pre-trained learning algorithm model.

Optionally, the intelligent interaction method further includes:

acquiring a uniform resource positioning address of an initial webpage according to a preset rule;

extracting a link related to the target object from the initial webpage through a webpage analysis algorithm, and adding the link related to the target object into a uniform resource location address queue to wait for obtaining information;

judging whether a preset condition for stopping executing the information acquisition task is met, if not, repeatedly executing the step of acquiring the uniform resource positioning address of the initial webpage according to a preset rule;

and if so, processing the acquired information and storing the processed data.

Optionally, after the sub-step of processing the acquired information is executed, the method further includes:

adding the processed content to the training samples to form new training samples, and performing model optimization on the learning algorithm model based on the new training samples.

Optionally, after the step of outputting, by the learning algorithm model, the adaptation information related to the target object corresponding to the interaction instruction is executed, and before the step of sending the adaptation information to the user is executed, the method further includes:

analyzing the adaptive information and determining a language type corresponding to the adaptive information;

generating voice information corresponding to the adaptation information through a text conversion technology based on a pre-stored voice sample library and the language type;

the sending the adaptation information to the user comprises:

and playing the voice information.

According to another aspect of embodiments of the present specification, there is provided an intelligent interaction device, including:

the interactive request receiving module is configured to receive an interactive request of a user, wherein the interactive request carries an interactive instruction between the user and the target object;

an interactive instruction input module configured to input an interactive instruction between the user and the target object into a pre-trained learning algorithm model;

an adaptation information output module configured to output adaptation information related to the target object corresponding to the interaction instruction through the learning algorithm model;

an adaptation information sending module configured to send the adaptation information to the user.

Optionally, the intelligent interaction device further includes:

an acquisition module configured to acquire user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object, and case data related to the target object;

the model training module is configured to train a learning algorithm model by taking the user attribute information, the target object attribute information, the management mechanism information of the target object, the basic knowledge information related to the target object and the case data related to the target object as training samples.

Optionally, the intelligent interaction device further includes:

the identity information acquisition module is configured to acquire identity identification information of the user according to a preset identity recognition algorithm;

a prompt message sending module configured to send learning prompt messages to the user if the user is determined to be the target user according to the identity information;

a detection module configured to detect a result of a response of the target user to the learning prompt information;

and the display module is configured to display information related to the target object for the target user through a system interactive interface if the response result is detected to be the confirmation learning.

Optionally, the intelligent interaction apparatus further includes:

the first adaptation information collection module is configured to collect an interaction instruction between the user and the target object and adaptation information corresponding to the interaction instruction according to a preset condition;

and the first model optimization module is configured to add an interactive instruction between the user and the target object and adaptation information corresponding to the interactive instruction to the training sample to form a new training sample, and perform model optimization on the learning algorithm model based on the new training sample.

Optionally, the intelligent interaction apparatus further includes:

the second adaptive information collection module is configured to collect an interactive instruction between the user and the target object, adaptive information corresponding to the interactive instruction and a feedback grading result of the user on the adaptive information corresponding to the interactive instruction according to a preset condition;

and the second model optimization module is configured to add the interactive instruction between the user and the target object with the feedback scoring result higher than the preset score threshold value and the adaptive information corresponding to the interactive instruction to the training sample to form a new training sample, and perform model optimization on the learning algorithm model based on the new training sample.

Optionally, the interaction request receiving module includes:

the interactive request receiving submodule is configured to receive an interactive request of a user, and the interactive request carries voice question information of the user for the target object;

the voice recognition sub-module is configured to perform voice recognition on the voice question information to obtain corresponding text information;

the interactive instruction input module comprises:

an information input sub-module configured to input the text information into a pre-trained learning algorithm model.

Optionally, the interaction request receiving module further includes:

the semantic analysis submodule is configured to perform semantic analysis on the text information to obtain a corresponding semantic analysis result;

the information input sub-module is further configured to: and inputting the semantic analysis result into a pre-trained learning algorithm model.

Optionally, the interaction request receiving module further includes:

the keyword extraction submodule is configured to extract keywords from the text information according to a preset rule;

the information input sub-module is further configured to: and inputting the extracted keywords into a pre-trained learning algorithm model.

Optionally, the intelligent interaction device further includes:

an information acquisition module configured to:

extracting a link related to the target object from the initial webpage through a webpage analysis algorithm, and adding the link related to the target object to a uniform resource positioning address queue to wait for obtaining information;

judging whether a preset condition for stopping executing the information obtaining task is met, if not, repeatedly executing the step of obtaining the uniform resource positioning address of the initial webpage according to a preset rule;

and if so, processing the acquired information and storing the processed data.

Optionally, the information obtaining module includes:

a third model optimization submodule configured to add the processed content to the training samples to form new training samples, and perform model optimization on the learning algorithm model based on the new training samples.

Optionally, the adaptation information sending module further includes:

the language type analysis sub-module is configured to analyze the adaptation information and determine a language type corresponding to the adaptation information;

the voice information conversion module is configured to determine voice information corresponding to the adaptation information through a text conversion technology based on a preset voice sample library and the language type;

and the voice information playing sub-module is configured to play the voice information.

According to another aspect of embodiments of the present specification, there is provided an electronic device comprising a memory, a processor, and computer instructions stored on the memory and executable on the processor, the processor implementing the steps of the intelligent interaction method when executing the instructions.

According to another aspect of embodiments of the present specification, there is provided a computer-readable storage medium storing computer instructions which, when executed by a processor, implement the steps of the intelligent interaction method.

In the embodiment of the specification, user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object and case data related to the target object are acquired; training a learning algorithm model by taking user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object and case data related to the target object as training samples; receiving an interactive request of a user, inputting an interactive instruction between the user and the target object carried in the interactive request into the learning algorithm model, acquiring adaptation information corresponding to the interactive instruction and related to the target object, and sending the adaptation information to the user.

In the embodiment of the specification, the interactive instruction in the interactive request is input into the pre-trained learning algorithm model, and the corresponding adaptation information is output, so that the efficiency of interaction is improved, and the accuracy of the adaptation information is improved.

Drawings

FIG. 1 is a flowchart of an intelligent interaction method provided by an embodiment of the present application;

FIG. 2 is a schematic diagram of a model development deployment call link provided by an embodiment of the present application;

FIG. 3 is a schematic diagram of an intelligent interaction method applied to an insurance scenario according to an embodiment of the present application;

FIG. 4 is a schematic structural diagram of an intelligent interaction device provided in an embodiment of the present application;

fig. 5 is a block diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit of this application and is therefore not limited to the specific implementations disclosed below.

The terminology used in the description of the one or more embodiments is for the purpose of describing the particular embodiments only and is not intended to be limiting of the description of the one or more embodiments. As used in one or more embodiments of the present specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It will be understood that, although the terms first, second, etc. may be used herein in one or more embodiments to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first can also be referred to as a second and, similarly, a second can also be referred to as a first without departing from the scope of one or more embodiments of the present description. The word "if," as used herein, may be interpreted as "at … …" or "at … …" or "in response to a determination," depending on the context.

The embodiment of the specification provides an intelligent interaction method. This specification also relates to an intelligent interaction device, an electronic apparatus, and a computer-readable storage medium, which are described in detail in the following embodiments one by one.

Fig. 1 shows a flowchart of an intelligent interaction method according to an embodiment of the present specification, including steps 102 to 108.

Step 102: and receiving an interaction request of a user, wherein the interaction request carries an interaction instruction between the user and the target object.

In one embodiment provided by the present specification, the interactive request includes a project information consultation request, a project join request, or other form of interactive request; the items comprise insurance items, mutual aid items, public service items and the like; the interaction instruction can be sent by voice, text information or clickable button/option information of a click robot interaction interface; if the user sends the interactive instruction through voice information, the robot receives an interactive request of the user and then carries out voice recognition on the voice information to obtain corresponding text information; and if the user sends the interactive instruction by clicking a clickable button/option of the robot interactive interface, the robot acquires corresponding text information through the button/option information.

In addition, after receiving an interactive request of a user, the identity identification information of the user can be acquired according to a preset identity recognition algorithm; the identity recognition algorithm is a face recognition algorithm and/or a fingerprint recognition algorithm. If the user is determined to be the target user according to the identity identification information, learning prompt information is sent to the user; detecting a response result of the target user to the learning prompt information; and if the response result is detected to be learning confirmation, displaying information related to the target object for the target user through a system interactive interface.

Taking the item as an insurance item as an example, assuming that the received interaction request of the user is 'i want to know health risks', after receiving the interaction request of the user, the robot can determine whether the user is an M user through a facial recognition algorithm, and if the user is determined to be the M user and the content to be learned is detected to exist in the database, the robot sends learning prompt information to the M user; detecting a response result of the M users to the learning prompt information; specifically, the M user can select whether to learn or not by clicking a touch button of the robot interactive interface; and if the response result is detected to be learning confirmation, displaying information related to the content to be learned for the M user through a robot interactive interface.

Step 104: and inputting the interactive instruction between the user and the target object into a pre-trained learning algorithm model.

In one embodiment provided by the present specification, the learning algorithm model is trained by:

In an embodiment provided by this specification, model training is performed based on an artificial intelligence learning system (TensorFlow), and a link for developing, deploying and calling a learning algorithm model includes 6 links, which can be specifically realized through the following steps:

1) Sample preparation

The general TensorFlow application code contains the definition of the Graph (Graph) and the operation of Session control (Session), the code quantity is not large, and the code quantity can be packaged into a file. Before training, sample data and test data need to be prepared, and a data file is a space or comma separated file (CSV) generally.

2) Feature engineering

The characteristic engineering refers to screening better data characteristics from original data by a series of engineering modes so as to improve the training effect of the model. The feature engineering generally comprises the steps of data preprocessing, feature selection, dimension reduction and the like.

3) Model training

Model training is a process of adjusting model parameters through training data to improve the fitting degree of a model to the data.

4) Model deployment

The artificial intelligence learning service system (TensorFlow Serving) is a high-performance open source library for machine learning model services (Serving). The method can deploy a trained machine learning model on line, and use a remote procedure call system (gPC) as an interface to accept external calls. In addition, it also supports model hot updates and automatic model version management. Specifically, after model training is completed, model deployment can be achieved only through a simple program.

5) Model invocation

After a model is trained, the results of the model are usually saved for later reuse. If Tensorflow is used to implement the neural network, all the weight values in the neural network are stored. And when the model is used in the later period, the saved model can be directly called through the code.

6) Log reflow

Logging reflow is the logging and reflow of the feature.

In an embodiment provided by the present specification, the model development deployment call link forms a closed loop through 6 links of sample preparation, feature engineering, model training, model deployment, model calling, and log reflow, as shown in fig. 2. The meaning of forming the closed loop is to record the online scoring characteristics, so that the generation of samples and characteristic data is facilitated, and the strong consistency of the data is ensured.

In the training process of the learning algorithm model, user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object and case data related to the target object are used as training samples and input into the learning algorithm model for training.

In an embodiment provided by this specification, the intelligent interaction method may be applied to interaction between a user and an intelligent robot in any scene, taking the application of the intelligent interaction method to an insurance consultation scene as an example, the target object is an insurance, the target object attribute information is insurance attribute information, a management mechanism of the target object is an insurance management mechanism, and the basic knowledge information related to the target object and the case data related to the target object are respectively basic knowledge information related to insurance and case data related to insurance.

By the training method of the learning algorithm model of the embodiment, the learning algorithm model is trained by taking the user attribute information, the target object attribute information, the management mechanism information of the target object, the basic knowledge information related to the target object and the case data related to the target object as training samples, so that the association among the user attribute information, the target object attribute information, the management mechanism information of the target object, the basic knowledge information related to the target object and the case data related to the target object is realized. In addition, the model is trained by using the user attribute information, the target object attribute information, the management mechanism information of the target object, the basic knowledge information related to the target object and the case data related to the target object, so that the relevance between the user and the adaptation information output by the model can be more reflected in the use process of the learning algorithm model.

In an embodiment provided by the present specification, the interaction instruction between the user and the target object may be sent through voice, text information, or information of clickable buttons/options of the robot interaction interface. Assuming that a user sends an interactive instruction of 'insurance application condition of health insurance' to an intelligent robot through voice, after the intelligent robot receives the interactive instruction of the user, voice recognition is carried out on voice information of the user to obtain corresponding text information, namely the obtained text information is 'insurance application condition of health insurance', and then the intelligent robot inputs the text information into a pre-trained learning algorithm model.

Specifically, before the text information is input into the pre-trained learning algorithm model, semantic analysis can be performed on the text information to obtain a corresponding semantic analysis result, and the semantic analysis result is input into the pre-trained learning algorithm model.

Still taking the application of the intelligent interaction method to an insurance consultation scene as an example, suppose that a user sends an interactive instruction of 'insurance conditions of health insurance' to an intelligent robot through voice, after the intelligent robot receives the interactive instruction of the user, voice recognition is carried out on voice information of the user to obtain corresponding text information, namely the obtained text information is 'insurance conditions of health insurance', then the intelligent robot carries out semantic analysis on the text information to obtain a corresponding semantic analysis result which is 'insurance conditions of health insurance', and the semantic analysis result is input into a pre-trained learning algorithm model.

In addition, before the text information is input into the pre-trained learning algorithm model, keywords can be extracted from the text information according to a preset rule, and the extracted keywords are input into the pre-trained learning algorithm model.

Along the above example, it is assumed that a user sends an interactive instruction of a "health insurance application condition" to an intelligent robot through voice, after the intelligent robot receives the interactive instruction of the user, voice recognition is performed on voice information of the user to obtain corresponding text information, that is, the obtained text information is the "health insurance application condition", then the intelligent robot extracts keywords from the text information, and the extracted keywords are assumed to be the "health insurance and application condition", and the extracted keywords are input into a pre-trained learning algorithm model.

Step 106: and outputting adaptation information corresponding to the interaction instruction and relevant to the target object through the learning algorithm model.

In one embodiment provided in this specification, the adaptation information, that is, answer information related to an interaction request of a user, is obtained by training a learning algorithm model using user attribute information, target object attribute information, management mechanism information of a target object, basic knowledge information related to the target object, and case data related to the target object as training samples, so as to implement association between the user attribute information, the target object attribute information, the management mechanism information of the target object, the basic knowledge information related to the target object, and the case data related to the target object.

Still taking the application of the intelligent interaction method to an insurance consultation scene as an example, suppose that a user sends an interaction instruction of 'i want to know health risks' to an intelligent robot through voice, the intelligent robot receives the interaction instruction of the user, performs voice recognition on voice information of the user to obtain corresponding text information, performs semantic analysis on the text information to obtain a corresponding semantic analysis result of 'health risks', inputs the semantic analysis result into a pre-trained learning algorithm model, and outputs adaptation information corresponding to the interaction instruction and related to the target object through the learning algorithm model.

The learning algorithm model is supposed to process the semantic analysis result "health insurance" to obtain an output result which is the health insurance, which is a Chinese abbreviation of health insurance, and means that an insurance company pays insurance money for loss caused by health reasons in the modes of disease insurance, medical insurance, incapability income loss insurance, nursing insurance and the like.

It should be noted that the above describes preferred embodiments of the present disclosure, and some steps are not necessary for implementing the present disclosure. Once the model is built, the model can be repeatedly used on line for a period of time, and in order to ensure that the adaptation information output by the model is more accurate, updated relevant data can be periodically selected to rebuild the model, but the steps are not necessary for providing the adaptation information for the user every time.

Step 108: and sending the adaptation information to the user.

In an embodiment provided by this specification, the adaptation information output by the learning algorithm model is in a text form, and the adaptation information may be sent to the user in a text manner, or may be sent in a voice, picture, or table manner.

Assuming that the adaptation information is transmitted by voice, a specific information conversion process may be implemented by the following steps:

and generating voice information corresponding to the adaptation information through a text conversion technology based on a pre-stored voice sample library and the language type.

And after the adaptation information is converted into voice information, sending the adaptation information to the user, namely playing the voice information through the intelligent robot.

Still taking the application of the intelligent interaction method to an insurance consultation scenario as an example, assuming that a robot receives an interaction request of a user as voice information "insurance application condition" and after receiving the interaction request of the user, voice recognition is performed on the voice information of the user to obtain corresponding text information, then keyword extraction is performed on the text information, assuming that the extracted keyword is "insurance application condition", the extracted keyword is input into a pre-trained learning algorithm model, assuming that the learning algorithm model processes the keyword as "insurance application condition" to obtain an output result that "the applicant must have corresponding right ability and behavior ability", otherwise, the established insurance contract does not have legal effectiveness ", that is, the adaptive information corresponding to the interaction instruction and related to the target object is output through the learning algorithm model, that" the applicant must have corresponding right ability and behavior ability ", otherwise, the established insurance contract does not have legal effectiveness".

After the adaptation information is determined, analyzing the adaptation information that the applicant has to have corresponding right capability and behavior capability, otherwise the established insurance contract does not have legal effectiveness, determining that the language type corresponding to the adaptation information is Chinese, generating the voice information corresponding to the adaptation information through a text conversion technology based on a pre-stored voice sample library and the language type, and playing the adaptation information through the voice of the intelligent robot that the applicant has to have corresponding right capability and behavior capability, otherwise the established insurance contract does not have legal effectiveness.

In an embodiment provided by this specification, after the learning model training is completed, in order to ensure accuracy of adaptation information output by the model, updated relevant data may be periodically selected for model optimization, and a specific model optimization process may be implemented through the following steps:

Specifically, the preset condition is a preset period, and assuming that the preset period is 10 days, the interactive instruction between the user and the target object and the adaptation information corresponding to the interactive instruction are collected according to the preset condition, that is, the interactive instruction between the user and the target object and the adaptation information corresponding to the interactive instruction within 10 days are collected. And after information collection is completed, adding the collected information to the training samples to form new training samples, and performing model optimization on the learning algorithm model based on the new training samples.

In practical applications, the preset condition may also be the number of the adaptation information, which is not limited in the present invention.

Besides, the model optimization process can be realized by the following steps:

Specifically, assuming that the interactive instruction of the user is "insurance application condition", and the adaptation information corresponding to the interactive instruction is "the applicant must have corresponding right ability and behavior ability, otherwise, the contracted insurance contract does not have legal effectiveness", and the feedback score of the user on the adaptation information is 8; assuming that the interactive instruction of the user is 'i want to know about health insurance', the adaptive information corresponding to the interactive instruction is 'health insurance', which is Chinese short for health insurance, and means that an insurance company pays insurance money for loss caused by health reasons in the modes of disease insurance, medical insurance, disability, income, loss, insurance, nursing insurance and the like, and the feedback score of the user on the adaptive information is 9. After information collection is completed, adding an interactive instruction between the user and the target object and adaptive information corresponding to the interactive instruction, of which the feedback scoring result is higher than a preset score threshold value, to the training sample to form a new training sample, and performing model optimization on the learning algorithm model based on the new training sample.

In an embodiment provided by this specification, since the basic knowledge related to the target object and the management mechanism information of the target object are continuously updated, in order to ensure the accuracy of the adaptation information output by the model, the updated information may be periodically obtained to perform model optimization, and the specific information obtaining process may be implemented by:

and if so, processing the acquired information and storing the processed data.

Specifically, the acquired information is processed, the processed content is added to the training sample to form a new training sample, and model optimization is performed on the learning algorithm model based on the new training sample.

In an embodiment provided in this specification, for example, if the preset rule is "acquire information included in a webpage with a click rate exceeding 1 ten thousand times in 24 hours", assuming that the target object is an english article, and the webpages meeting the preset rule are a webpage B, a webpage C, and a webpage D, acquiring URLs of the webpage B, the webpage C, and the webpage D, extracting links related to the english article from the webpage B, the webpage C, and the webpage D by using a webpage analysis algorithm, and adding the links to a URL queue to wait for acquisition of information; after a certain number of English articles are acquired, if the number of the acquired English articles reaches 100 preset articles, the condition that the execution of the preset information acquisition task is stopped can be determined, the acquisition of the English articles is stopped, and the acquired English articles are processed and stored.

In one embodiment provided by the specification, after an interactive request of a user is received, identity information of the user is identified according to an identity identification algorithm, a learning function is provided for a target user, the target user is helped to learn latest basic knowledge information and management mechanism information related to the target object in time, learning cost is effectively reduced, and learning efficiency is improved; meanwhile, after the model training is finished, model optimization is carried out according to continuously updated related information, so that the accuracy of the model output information is ensured; the method and the system automatically capture programs or scripts of the world wide web information according to certain rules, automatically acquire all accessible page contents to acquire or update the contents and retrieval modes of websites, and are favorable for ensuring that the information related to the target object in the database can be updated in time.

Fig. 3 shows an intelligent interactive method according to an embodiment of the present specification, which is described by taking an insurance scenario as an example, and includes steps 302 to 318.

Step 302: receiving an interactive request of a user, wherein the interactive request carries consultation information of the user on insurance projects.

Step 304: and acquiring the identity identification information of the user according to a facial recognition algorithm.

Step 306: judging whether the user is an insurance agent, if so, executing step 308; if not, go to step 314.

Step 308: and sending learning prompt information to the user.

Step 310: checking whether the response result is a learning confirmation result, if yes, executing step 312; if not, go to step 314.

Specifically, it is detected whether the response result of the target user to the learning prompt information is learning confirmation, if yes, step 312 is executed; if not, go to step 314.

Step 312: and displaying information related to insurance for the target user through a system interactive interface.

Specifically, assuming that the received interaction request of the user is 'i want to know health risks', after the interaction request of the user is received, the robot can determine whether the user is a user A through a facial recognition algorithm, and if the user is determined to be the user A and the fact that the content to be learned exists in the database is detected, the robot sends learning prompt information to the user A; detecting a response result of the user A to the learning prompt information; specifically, the user A can select whether to learn or not by clicking a touch button of the robot interactive interface; and if the response result is detected to be learning confirmation, displaying information related to the content to be learned for the user A through a robot interactive interface.

Specifically, the interaction request may be sent by voice, text information, or by clicking clickable button/option information of the robot interaction interface; if the user sends the interaction request through voice, the robot receives the interaction request of the user and then carries out voice recognition on the voice information to obtain corresponding text information; and if the user sends the interactive instruction through a clickable button/option of the robot interactive interface, the robot acquires corresponding text information through the button/option information.

Step 314: and inputting the consultation information of the user on the insurance project into a pre-trained learning algorithm model.

acquiring user attribute information, insurance management mechanism information, basic knowledge information related to insurance and case data related to insurance;

and training a learning algorithm model by taking the user attribute information, the insurance management mechanism information, the basic knowledge information related to insurance and the case data related to insurance as training samples.

By the training method of the learning algorithm model of the embodiment, the learning algorithm model is trained by taking the user attribute information, the insurance management mechanism information, the basic knowledge information related to insurance and the case data related to insurance as training samples, so that the association among the user attribute information, the insurance management mechanism information, the basic knowledge information related to insurance and the case data related to insurance is realized. In addition, the model is trained by using the user attribute information, the insurance management mechanism information, the basic knowledge information related to insurance and the case data related to insurance, so that the relevance between the user and the adaptive information output by the model can be more reflected in the using process of the learning algorithm model.

In one embodiment provided by the present specification, the user's consultation information for insurance items may be sent by voice, text message, or clicking a clickable button/option message on the robot interactive interface. Assuming that a user sends an interactive request of 'i want to know health risks' to an intelligent robot through voice, after the intelligent robot receives the interactive request of the user, voice recognition is carried out on voice information of the user to obtain corresponding text information, namely the obtained text information is 'i want to know health risks', and then the intelligent robot inputs the text information into a pre-trained learning algorithm model.

Assuming that the text information obtained by performing voice recognition on the voice information of the user by the intelligent robot is 'i want to know health risks', then extracting keywords from the text information by the intelligent robot, and assuming that the extracted keywords are 'health risks', inputting the extracted keywords into a pre-trained learning algorithm model.

Step 316: and outputting adaptation information corresponding to the interaction request through the learning algorithm model.

According to the above example, the intelligent robot receives an interaction request of a user, performs voice recognition on voice information of the user to obtain corresponding text information, then performs keyword extraction on the text information to obtain a corresponding extraction result of 'health risk', inputs the extracted keyword into a pre-trained learning algorithm model, and assumes that the learning algorithm model processes according to the keyword 'health risk' to obtain an output result of 'basic information of health risk'.

Step 318: and sending the adaptation information to the user.

And if the adaptation information is sent in a voice mode, the adaptation information is converted into voice information, and then the voice information is sent to the user, namely the voice information is played through the intelligent robot, namely the basic knowledge of the health risk is transmitted to the user through the intelligent robot in a voice playing mode.

In an embodiment provided in this specification, after the training of the learning model is completed, in order to ensure the accuracy of the adaptation information output by the model, updated relevant data may be periodically selected for model optimization, and a specific model optimization process may refer to an implementation manner described in the method in fig. 1, which is not described herein again.

In one embodiment provided by the specification, after an interaction request of a user is received, identity information of the user is identified according to a facial recognition algorithm, a learning function is provided for a target user, the target user is helped to learn latest basic knowledge information and management mechanism information related to insurance in time, learning cost is effectively reduced, and learning efficiency is improved; meanwhile, after the model training is finished, model optimization is carried out according to continuously updated related information, so that the accuracy of the model output information is guaranteed; by continuously and automatically acquiring new information, the information related to the target object in the database can be timely updated.

Corresponding to the above method embodiment, the present specification further provides an intelligent interaction device embodiment, and fig. 4 shows a schematic structural diagram of the intelligent interaction device according to an embodiment of the present specification. As shown in fig. 4, the apparatus includes:

an interaction request receiving module 402, configured to receive an interaction request of a user, where the interaction request carries an interaction instruction between the user and the target object;

an interactive instruction input module 404 configured to input an interactive instruction between the user and the target object into a pre-trained learning algorithm model;

an adaptation information output module 406 configured to output adaptation information related to the target object corresponding to the interaction instruction through the learning algorithm model;

an adaptation information sending module 408 configured to send the adaptation information to the user.

Optionally, the intelligent interaction device further includes:

a model training module configured to train a learning algorithm model for a training sample using the user attribute information, the target object attribute information, the management mechanism information of the target object, the basic knowledge information related to the target object, and the case data related to the target object.

Optionally, the intelligent interaction device further includes:

and the display module is configured to display information related to the target object for the target user through a system interactive interface if the response result is detected to be the confirmed learning.

Optionally, the intelligent interaction device further includes:

the first adaptive information collection module is configured to collect an interactive instruction between the user and the target object and adaptive information corresponding to the interactive instruction according to a preset condition;

a first model optimization module configured to add an interaction instruction between the user and the target object and adaptation information corresponding to the interaction instruction to the training sample to form a new training sample, and perform model optimization on the learning algorithm model based on the new training sample.

Optionally, the intelligent interaction apparatus further includes:

Optionally, the interaction request receiving module includes:

the interactive instruction input module comprises:

an information input sub-module configured to input the textual information into a pre-trained learning algorithm model.

Optionally, the interaction request receiving module further includes:

the information input sub-module further configured to: and inputting the semantic analysis result into a pre-trained learning algorithm model.

Optionally, the interaction request receiving module further includes:

Optionally, the intelligent interaction device further includes:

an information acquisition module configured to:

acquiring a Uniform Resource Locator (URL) of an initial webpage according to a preset rule;

extracting a link related to the target object from the initial webpage through a webpage analysis algorithm, and adding the link related to the target object to a URL queue to wait for obtaining information;

judging whether a preset information acquisition task stop execution condition is reached, if not, repeatedly executing the step of acquiring a Uniform Resource Locator (URL) of the initial webpage according to a preset rule;

and if so, processing the acquired information and storing the processed data.

Optionally, the information obtaining module includes:

a third model optimization sub-module configured to add the processed content to the training samples to form new training samples, and perform model optimization on the learning algorithm model based on the new training samples.

Optionally, the adaptation information sending module further includes:

the language type analysis submodule is configured to analyze the adaptation information and determine a language type corresponding to the adaptation information;

the voice information conversion module is configured to determine voice information corresponding to the adaptation information through a text conversion technology (TTS) based on a preset voice sample library and the language type;

Fig. 5 shows a block diagram of an electronic device 500 according to an embodiment of the present description. The components of the electronic device 500 include, but are not limited to, a memory 510 and a processor 520. Processor 520 is coupled to memory 510 via bus 530, and database 550 is used to store data.

The electronic device 500 also includes an access device 540, the access device 540 enabling the electronic device 500 to communicate via one or more networks 560. Examples of such networks include the Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. The access device 540 may include one or more of any type of network interface, e.g., a Network Interface Card (NIC), wired or wireless, such as an IEEE802.11 Wireless Local Area Network (WLAN) wireless interface, a worldwide interoperability for microwave access (Wi-MAX) interface, an ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.

In one embodiment of the present description, the above-mentioned components of the electronic device 500 and other components not shown in fig. 5 may also be connected to each other, for example, through a bus. It should be understood that the block diagram of the electronic device shown in fig. 5 is for exemplary purposes only and is not intended to limit the scope of the present disclosure. Those skilled in the art may add or replace other components as desired.

The electronic device 500 may be any type of stationary or mobile electronic device, including a mobile computer or mobile electronic device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), a mobile phone (e.g., smartphone), a wearable electronic device (e.g., smartwatch, smartglasses, etc.), or other type of mobile device, or a stationary electronic device such as a desktop computer or PC. The electronic device 500 may also be a mobile or stationary server.

Wherein, the processor 520 is used for executing the following steps of the intelligent interaction method which can be executed by the computer.

The above is a schematic scheme of an electronic device of the present embodiment. It should be noted that the technical solution of the electronic device and the technical solution of the intelligent interaction method belong to the same concept, and details that are not described in detail in the technical solution of the electronic device can be referred to the description of the technical solution of the intelligent interaction method.

An embodiment of the present specification further provides a computer readable storage medium storing computer instructions, which when executed by a processor, implement the steps of the intelligent interaction method as described above.

The above is an illustrative scheme of a computer-readable storage medium of the present embodiment. It should be noted that the technical solution of the storage medium belongs to the same concept as the technical solution of the intelligent interaction method, and details that are not described in detail in the technical solution of the storage medium can be referred to the description of the technical solution of the intelligent interaction method.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The computer instructions comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, or the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, read-Only Memory (ROM), random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.

It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present application is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

The preferred embodiments of the present application disclosed above are intended only to aid in the explanation of the application. Alternative embodiments are not exhaustive and do not limit the invention to the precise embodiments described. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the application and its practical application, to thereby enable others skilled in the art to best understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. An intelligent interaction method, comprising:

outputting adaptation information related to the target object corresponding to the interaction instruction through the learning algorithm model, wherein the adaptation information is answer information related to the interaction request of the user;

and sending the adaptation information to the user.

2. The method of claim 1, wherein the learning algorithm model is trained by:

3. The method of claim 1, wherein after the step of receiving the user's interactive request is performed, the method further comprises:

4. The method of claim 1, wherein after the step of sending the adaptation information to the user is performed, the method further comprises:

and adding the interactive instruction between the user and the target object and the adaptive information corresponding to the interactive instruction into a training sample to form a new training sample, and performing model optimization on the learning algorithm model based on the new training sample.

5. The method of claim 1, wherein after the step of sending the adaptation information to the user is performed, the method further comprises:

and adding the interaction instruction between the user and the target object with the feedback scoring result higher than a preset score threshold value and the adaptation information corresponding to the interaction instruction to a training sample to form a new training sample, and performing model optimization on the learning algorithm model based on the new training sample.

6. The method of claim 1, wherein the interactive instruction is voice question information;

and inputting the text information into a pre-trained learning algorithm model.

7. The method of claim 6, wherein before said entering the textual information into a pre-trained learning algorithm model substep is performed, further comprising:

the inputting the text information into a pre-trained learning algorithm model comprises:

8. The method of claim 6, wherein before the entering the text message into a pre-trained learning algorithm model substep is performed, further comprising:

extracting keywords from the text information according to a preset rule;

9. The method of claim 1, further comprising:

and if so, processing the acquired information and storing the processed data.

10. The method of claim 9, wherein after the substep of processing the retrieved information is performed, the method further comprises:

adding the processed content to a training sample to form a new training sample, and performing model optimization on the learning algorithm model based on the new training sample.

11. The method according to claim 1, wherein after the step of outputting the adapted information related to the target object corresponding to the interactive instruction through the learning algorithm model is executed and before the step of sending the adapted information to the user is executed, the method further comprises:

the sending the adaptation information to the user comprises:

and playing the voice information.

12. An intelligent interaction device, comprising:

the interactive instruction input module is configured to input an interactive instruction between the user and a target object into a pre-trained learning algorithm model;

an adaptation information output module configured to output adaptation information related to the target object corresponding to the interaction instruction through the learning algorithm model, wherein the adaptation information is answer information related to the interaction request of the user;

13. The apparatus of claim 12, further comprising:

14. The apparatus of claim 12, further comprising:

a prompt information sending module configured to send learning prompt information to the user if the user is determined to be the target user according to the identity information;

15. The apparatus of claim 12, further comprising:

and the first model optimization module is configured to add an interactive instruction between the user and the target object and adaptation information corresponding to the interactive instruction to a training sample to form a new training sample, and perform model optimization on the learning algorithm model based on the new training sample.

16. The apparatus of claim 12, further comprising:

and the second model optimization module is configured to add an interactive instruction between the user and the target object and adaptive information corresponding to the interactive instruction, of which the feedback scoring result is higher than a preset score threshold value, to a training sample to form a new training sample, and perform model optimization on the learning algorithm model based on the new training sample.

17. The apparatus of claim 12, wherein the interaction request receiving module comprises:

the voice recognition submodule is configured to perform voice recognition on the voice question information to obtain corresponding text information;

the interactive instruction input module comprises:

18. The apparatus of claim 17, wherein the interaction request receiving module further comprises:

19. The apparatus of claim 17, wherein the interaction request receiving module further comprises:

20. The apparatus of claim 12, further comprising:

an information acquisition module configured to:

and if so, processing the acquired information and storing the processed data.

21. The apparatus of claim 20, wherein the information obtaining module comprises:

a third model optimization submodule configured to add the processed content to training samples to form new training samples, and perform model optimization on the learning algorithm model based on the new training samples.

22. The apparatus of claim 12, wherein the adaptation information sending module further comprises:

23. An electronic device, comprising:

a memory, a processor;

the memory to store computer-executable instructions, the processor to execute the computer-executable instructions:

receiving an interaction request of a user, wherein the interaction request carries an interaction instruction between the user and a target object;

and sending the adaptation information to the user.

24. A computer-readable storage medium storing computer instructions, which when executed by a processor, perform the steps of the method of any one of claims 1 to 11.