Function information pushing method, device and equipment and computer storage medium

Abstract

The application discloses a function information pushing method and device, equipment and a computer storage medium, the method comprises the steps that a plurality of function information corresponding to a smart home device is acquired; the plurality of function information is input into a convenience decision tree model, and convenience information corresponding to each function information is obtained; the plurality of function information and the convenience information corresponding to each function information are input into a SHAP model, and a contribution value corresponding to each function information is obtained; according to the contribution value, target function information meeting a preset condition is determined from the plurality of function information; and the target function information is pushed to a user.According to the embodiment of the application, the target function information is pushed to the user, so that the function of the smart home device most meeting the demand of the user can be recommended to the user, and the demand of the user for intelligent, personalized and efficient smart home life is met.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of electronic devices, and particularly relates to a function information pushing method and device, an apparatus, and a computer storage medium. BACKGROUND

[0002] With the rapid development of science and technology, home smart devices are gradually integrated into people's daily life and become an indispensable part of modern families. By using various home smart devices with different functions, people's daily life can be greatly facilitated, and the comfort and safety of home life can be improved.

[0003] Further, with the continuous popularization of smart home devices and the continuous improvement of user needs, the functions of smart home devices are gradually enriched. However, too many operation functions not only require a large storage space, but also cause difficulties for users to control various functions of smart home devices. Therefore, the functions of smart home devices can be pushed to users to achieve smart, personalized, and efficient smart home life experience. SUMMARY

[0004] The embodiments of the present application provide a function information pushing method, device, apparatus, and computer storage medium. By obtaining a plurality of function information corresponding to a smart home device, the plurality of function information is input into a convenience decision tree model, and the convenience information corresponding to each function information is obtained. By analyzing the contribution value corresponding to each function information through a SHapley Additive exPlanations (SHAP) model, the target function information meeting the preset condition is pushed to the user, so as to ensure that the function of the smart home device most meeting the needs of the user is recommended to the user, and the needs of the user for smart, personalized, and efficient smart home life are met.

[0005] In a first aspect, the embodiments of the present application provide a function information pushing method, comprising:

[0006] obtaining a plurality of function information corresponding to a smart home device; inputting the plurality of function information into a convenience decision tree model to obtain convenience information corresponding to each function information; inputting the plurality of function information and the convenience information corresponding to each function information into a SHAP model to obtain a contribution value corresponding to each function information; determining target function information meeting a preset condition from the plurality of function information according to the contribution value; and recommending the target function information to a user.

[0007] In an implementation, before the inputting the plurality of function information into the convenience decision tree model to obtain the convenience information corresponding to each function information, the method further comprises: obtaining a first training sample set, the first training sample set comprising a plurality of training samples, each of the training samples comprising a plurality of function information samples corresponding to the smart home device and a convenience label of the smart home device; training a preset decision tree model by using the training samples until a training stop condition is met to obtain the convenience decision tree model.

[0008] In an implementation, the training of the preset decision tree model by using the training samples until the training stop condition is met to obtain the convenience decision tree model comprises: calculating a sample entropy value corresponding to the first training sample set; cyclically performing the following steps A to C until a preset stop condition is met to obtain node quantity information in the decision tree model and function information corresponding to the nodes: step A: classifying the first training sample set based on each of the function information samples to obtain a plurality of second training sample sets; step B: calculating a conditional entropy value of the second training sample set; step C: determining the node quantity information of the preset decision tree model and the function information corresponding to the nodes by using the sample entropy value and the conditional entropy value; and constructing the convenience decision tree model based on the node quantity information and the function information corresponding to each of the nodes.

[0009] In an implementation, the step C comprises: calculating an entropy gain between each of the conditional entropy values and the sample entropy value respectively; determining a target second training sample set with the maximum entropy gain; and taking the function information sample corresponding to the target second training sample set as the function information corresponding to the node of the convenience model.

[0010] In an implementation, after the obtaining of the convenience decision tree model, the method further comprises: verifying the function information corresponding to each of the nodes of the convenience decision tree model in a sequence from low to high of the node levels of the convenience decision tree model; and optimizing and pruning the convenience decision tree model based on the verification result to obtain an optimized convenience decision tree model.

[0011] In an implementation, the function information sample comprises at least one of the following: a function type, usage information, state information, environment information, and interaction information of the smart home device; and the convenience label of the smart home device comprises at least user feedback information.

[0012] In an implementation form, the calculating the contribution value corresponding to each function information by inputting the plurality of function information and the convenience information corresponding to each function information into the Shapley Additive explanations (SHAP) model comprises: combining the plurality of function information to obtain a plurality of function information groups; selecting any one of the plurality of function information as a target function information; for each target function information, performing the following steps: determining the first convenience information corresponding to the target function information in the case that the target function information is contained in the function information group, and calculating the second convenience information in the case that the target function information is not contained in the function information group; performing weighted summation on the first convenience information and the second convenience information to obtain the SHAP value corresponding to the target function information; and calculating the product of the SHAP value corresponding to the target function information and the target function information, and taking the product as the contribution value of the target function information.

[0013] In a second aspect, an embodiment of the present application provides a function information pushing device, the device comprising: a first obtaining module configured to obtain a plurality of function information corresponding to a smart home device; a second obtaining module configured to input the plurality of function information into a convenience decision tree model to obtain convenience information corresponding to each function information; a third obtaining module configured to input the plurality of function information and the convenience information corresponding to each function information into a Shapley Additive explanations (SHAP) model to obtain a contribution value corresponding to each function information; a determining module configured to determine a target function information satisfying a preset condition from the plurality of function information according to the contribution value; and a pushing module configured to recommend the target function information to a user.

[0014] In a third aspect, an embodiment of the present application provides a function information pushing device, the device comprising: a processor and a memory storing computer program instructions;

[0015] The processor implements the function information pushing method in the first aspect or any implementation form of the first aspect when executing the computer program instructions.

[0016] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, characterized in that the computer readable storage medium stores computer program instructions, and the computer program instructions are executed by a processor to implement the function information pushing method in the first aspect or any implementation form of the first aspect.

[0017] In a fifth aspect, an embodiment of the present application provides a computer program product, characterized in that instructions in the computer program product are executed by a processor of an electronic device to cause the electronic device to perform the function information pushing method in the first aspect or any implementation form of the first aspect.

[0018] The function information pushing method, device, equipment and computer storage medium provided by the embodiments of the present application obtain a plurality of function information corresponding to the smart home device, input the plurality of function information into a convenience decision tree model, and obtain the convenience information corresponding to each function information. The SHAP model is used to analyze the contribution value corresponding to each function information, and then the target function information meeting the preset condition is pushed to the user, so that the function of the smart home device most meeting the demand of the user can be recommended to the user, and the demand of the user for the intelligent, personalized and efficient smart home life can be met. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments of the present application will be briefly introduced. Those skilled in the art can obtain other drawings according to these drawings without creating any creative labor.

[0020] Figure 1 Fig. 1 shows a schematic diagram of a function information pushing system provided by an embodiment of the present application;

[0021] Figure 2 Fig. 2 is a flowchart of a convenience decision tree model training method provided by an embodiment of the present application;

[0022] Figure 3 Fig. 3 is a flowchart of an implementation of S220 provided by an embodiment of the present application;

[0023] Figure 4 Fig. 4 is a flowchart of another convenience decision tree model training method provided by an embodiment of the present application.

[0024] Figure 5 Fig. 5 shows a flowchart of a function information pushing method provided by an embodiment of the present application.

[0025] Figure 6 Fig. 6 is a flowchart of an implementation of S530 provided by an embodiment of the present application.

[0026] Figure 7 Fig. 7 is a structural diagram of a function information pushing method device provided by another embodiment of the present application;

[0027] Figure 8 Fig. 8 is a structural diagram of a function information pushing equipment provided by another embodiment of the present application. DETAILED DESCRIPTION

[0028] The features and exemplary embodiments of the various aspects of the present application will be described in detail below with reference to the drawings. For the purpose of clarity, the description is divided into the following sections: technical scheme, advantages, and specific embodiments. It should be understood that the specific embodiments described herein are intended to explain the principles of the present application, and are not intended to limit the present application. The present application can be implemented without some of the specific details described below. The following description of the embodiments is merely provided to better illustrate the present application by showing examples of the present application.

[0029] It should be noted that the relative terms such as first and second, etc., are used herein only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a list of elements does not only include those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, method, article or device. Without more limitations, the elements defined by the statement "comprise" do not exclude the presence of additional identical elements in the process, method, article or device that includes the elements.

[0030] With the rapid development of technology, the functions of smart home devices are gradually increasing. However, due to too many functions, it may cause inconvenience to user operation, and affect the user experience of the smart home device.

[0031] As users have higher requirements for the convenience and intelligence level of smart home devices, in order to enable smart home devices to better integrate into the user's home life and achieve intelligent, personalized and efficient smart home life experience, it is essential to push the functions of smart home devices.

[0032] To solve the problems in the prior art, the embodiments of the present application provide a function information pushing method, device, equipment and computer storage medium. It can be understood that the function information pushing method provided by the embodiments of the present application can be implemented in the function information pushing system 100, and the target function information that can be recommended to the user is determined based on the convenience of each function in the smart home device and the contribution value of each function.

[0033] In the embodiments of the present application, Figure 1 A schematic diagram of a function information pushing system provided by an embodiment of the present application is shown. As shown in FIG. 1, the function information pushing system 100 includes a function information pushing device 110 and a smart home device 120. Figure 1As shown, the function information pushing system 100 can include a data collection module 110, an Internet of Things platform 120, and an upper application 130. Among them, the data collection module includes multiple smart home modules, including a smart home module 111, a smart home module 112, etc., and each smart home includes multiple smart home devices. It can be understood that the smart home devices included in each smart home can be the same or can be different. And each smart home device can record the user's use data in the user's use process. The Internet of Things platform 120 can be used to obtain the corresponding data and transmit the corresponding data to the upper application 130. In the upper application 130, the received data can be used to train the decision tree model based on the convenience decision tree training module 131 to obtain the convenience decision tree model. Further, the SHAP model in the model analysis module 132 can be used to quantitatively analyze the convenience, so as to determine the device function information of the intelligent device that needs to be recommended to the user, and through the scheduling module 133, the corresponding device function information is recommended to the user.

[0034] It should be noted that in the function information pushing method provided by the embodiments of the present application, the convenience decision tree model is used to determine the convenience of the intelligent device. And the SHAP model analyzes the contribution value of each function of each smart home device to the convenience, so as to determine the target function information that can be pushed to the user. Therefore, first, the specific implementation of the training method of the convenience decision tree model provided by the embodiments of the present application described in combination with the accompanying drawings is described below.

[0035] As shown in Figure 2 , Figure 2 is a flowchart of the training method of the convenience decision tree model provided by an embodiment of the present application. The training method of the convenience decision tree model includes:

[0036] S210, obtaining a first training sample set.

[0037] S220, training the to-be-trained decision tree model using the first training sample until the training stop condition is met, to obtain the convenience decision tree model.

[0038] Among them, the first training sample set can include multiple training samples. And each training sample includes multiple function information samples corresponding to the smart home device and the convenience label of the smart home device.

[0039] In the embodiments of the present application, the first training sample set can be the use information of the smart home device deployed in the user's home. In an example, the Internet of Things platform 120 and the plurality of smart home modules 111 can communicate through the Message Queuing Telemetry Transport (MQTT) protocol to realize data transmission. It can be understood that data transmission through the MQTT protocol can ensure the real-time and high efficiency of the data. In addition, when using the MQTT protocol for interaction, a suitable topic structure can be used to facilitate subsequent data management by the Internet of Things platform; setting a real-time monitoring mechanism is beneficial to ensure the timeliness of data uploading; establishing an abnormal data processing flow can record, notify and correct abnormal data in a timely manner.

[0040] It can be understood that the data format uploaded needs to comply with the specifications of the Internet of Things platform, including data types, units, etc. The label and field of the data can be defined in advance for each smart home device to facilitate subsequent data processing. In addition, encryption and other means can be used to ensure the security of data transmission. Considering user privacy, desensitization or anonymization measures are adopted.

[0041] Further, the data of the smart home device can be obtained through the data types and data collection methods shown in Table 1, and used as the first training sample:

[0042] Table 1

[0043]

[0044] It can be understood that when collecting data, attention should be paid to the timing and compliance of the data. The timing of the data can be ensured by using a timestamp. In an example, the timestamp can be recorded in a date and time format.

[0045] Further, the obtained data can be preprocessed, including standardization, missing value processing, etc.

[0046] In an example, the standardization of the data can be performed by using Z-score standardization to adjust the values of different ranges to the same scale to prevent some features from having too much influence on the model. Normalization can be used to ensure that all feature values are between 0 and 1 to speed up the training process of the model. In addition, enumeration, string and other types of data can be digitally encoded.

[0047] In an example, the missing value processing of the data can be performed by interpolation, filling or deleting the missing values. If d types of data information are selected as feature data, then for the i-th sample x i The expression is as follows:

[0048] x i =(x i1 ,x i2 ,x i3 ......x id )

[0049] wherein, d is the dimension of sample data. It can be understood that x i may represent that the i-th smart home device function information contains d feature data, including x i1 , x i2 , x i3 ......x id .

[0050] In the embodiment of the application, the function information sample of the smart home device includes at least one of the following: function type of the smart home device, usage information, state information, environment information, interaction information. The convenience label of the smart home device at least includes user feedback information.

[0051] In the embodiment of the application, the preset decision tree model can be trained by using the first training sample until the training stop condition is met, and the convenience decision tree model is obtained.

[0052] In the embodiment of the application, the first training sample set is obtained to train the decision tree model to be trained, so that the convenience decision tree model is obtained after training, so that the convenience of the smart home device can be determined by using the trained convenience model.

[0053] In the embodiment of the application, Figure 3 is a flowchart of an implementation of S220 provided by an embodiment of the application, as Figure 3 shown, the training method of the convenience decision tree model includes:

[0054] S221, the sample entropy value corresponding to the first training sample set is calculated.

[0055] In the embodiment of the application, the sample entropy value of the first training sample can represent the degree of confusion of the first training sample, which can be calculated by the following formula:

[0056] H(S)=-∑P k *log(P k ) (1)

[0057] wherein, S is the first sample set. If the output value of the convenience decision tree model has K types, k belongs to K, P k is the probability that the convenience evaluation result is the k-th category.

[0058] In the embodiments of the present application, the first training sample set can be a training set. It can be understood that the entire preprocessed data set can be divided to obtain the training set and the validation set, so as to ensure that the model can effectively generalize. In an example, the entire data set can be divided by 80% as a training set and 20% as a validation set in a random division manner.

[0059] S222, based on each function information sample, respectively classifying the first training sample set to obtain a plurality of second training sample sets.

[0060] In the embodiments of the present application, the first training sample set can be classified based on the function information sample of the smart home device to obtain a plurality of second training sample sets.

[0061] In the embodiments of the present application, the function information sample can be the function information contained in the smart home device. Each smart home device can contain one or more function information. And there can be one or more smart home devices.

[0062] S223, calculating the conditional entropy value of the second training sample set.

[0063] In the embodiments of the present application, the conditional entropy value of the second training sample set can represent the confusion degree of the second sample set. It can be understood that the confusion degree of the second sample set can represent the category information of each sample label in each sub-training set obtained by dividing the first training sample set through a certain function information sample.

[0064] In the embodiments of the present application, the conditional entropy value corresponding to each second training sample can be calculated by the following formula:

[0065] H(S|A)=∑P(x ij )*H(S j ) (2)

[0066] Wherein, H(S|A) represents the conditional entropy value of the second training sample set after being divided by a certain feature A; if m is the value number of the feature A, then j∈m, that is, j belongs to m; P(x ij ) represents the probability that the feature A takes the jth value. H(S j ) represents the entropy of the set S when the feature A takes the value x ij .

[0067] S224, determining the node number information of the to-be-trained decision tree model and the function information corresponding to the node through the sample entropy value and the conditional entropy value.

[0068] In the embodiments of the present application, the node quantity information of the to-be-trained decision tree model and the function information corresponding to each node can be determined through the sample entropy value and the conditional entropy value of each second training sample corresponding to different function information samples.

[0069] In the embodiments of the present application, the target second training sample set with the maximum entropy gain can be determined by respectively calculating the entropy gain between each conditional entropy value and the sample entropy value. The function information sample corresponding to the target second training sample set is taken as the function information corresponding to the node of the convenience model.

[0070] In the embodiments of the present application, the entropy gain can be calculated by the following formula:

[0071] Gain(S, A) = H(S) - H(S|A) (3)

[0072] Wherein, Gain(S, A) represents the information entropy gain of the input feature A of the convenience decision tree model.

[0073] In the embodiments of the present application, by calculating the information entropy gain corresponding to each function information sample, the function information with the maximum information entropy gain can be selected for the construction of the decision tree, so as to better segment the samples and improve the discrimination ability of the model.

[0074] S225, determine whether the preset stop condition is met, if yes, execute S226; if no, return to step S222.

[0075] In the embodiments of the present application, when it is determined that the node of the decision tree meets the preset stop condition, the training is stopped, and the decision tree is constructed based on the function information corresponding to each node of the decision tree completed by the training and the node quantity. If it is determined that the node of the decision tree does not meet the preset stop condition, the decision tree is continuously trained, that is, the node quantity information of the decision tree and the function information corresponding to each node are determined.

[0076] In the embodiments of the present application, the preset stop condition can be that the number of samples contained in the node is less than a set threshold, or the maximum depth of the decision tree is reached. It can be understood that the training process of the decision tree can be understood as a recursive process. Once the stop condition is met, the recursion stops, the current node becomes a leaf node, represents a class or result, and the construction process of the entire decision tree ends.

[0077] S226, construct the convenience decision tree model based on the node quantity information and the function information corresponding to each node.

[0078] In the embodiments of the present application, the first training sample set is classified according to the function information sample, thereby obtaining a plurality of second training sample sets, and the conditional entropy value of each second training sample set is calculated. Based on the sample entropy value corresponding to the first training sample set and the plurality of conditional entropy values, the node number information of the decision tree model and the function information corresponding to each node are determined. The convenience decision tree is constructed through the function information, which can accurately capture the influence of each intelligent device function on user experience. Moreover, the problem of mutual interference between features in the traditional model can be effectively avoided, so that the convenience decision tree more accurately reflects the relationship between device function and convenience.

[0079] In order to avoid overfitting problem of the decision tree model, as another implementation manner of the present application, another implementation manner of the convenience decision tree model training method is provided, please refer to the following embodiments.

[0080] Figure 4 is a flowchart of another convenience decision tree model training method provided by an embodiment of the present application, please refer to Figure 4 , the method comprises the following steps:

[0081] S410, obtaining a first training sample set.

[0082] S420, training the to-be-trained decision tree model by using the first training sample until a training stop condition is met, and obtaining a convenience decision tree model.

[0083] S430, verifying the function information corresponding to each node of the convenience decision tree model in the order from low to high of the node level of the convenience decision tree model.

[0084] S440, based on the verification result, optimizing and pruning the convenience decision tree model to obtain an optimized convenience decision tree model.

[0085] In the embodiments of the present application, steps S410-S420 are consistent with steps S210-S220, and will not be described in detail here.

[0086] In the embodiments of the present application, for the convenience decision tree which has completed training, each node of the convenience decision tree model can be evaluated recursively from the bottom up, that is, the function information corresponding to each node of the convenience decision tree model can be verified in the order from low to high of the node level of the convenience decision tree model. In an example, the verification can be performed in the cross-validation manner.

[0087] In the embodiments of the present application, through the verification structure, it is judged whether removing a certain branch can improve the generalization ability of the whole decision tree model, that is, the optimization pruning of the decision tree model is realized.

[0088] In the embodiments of the present application, the convenience decision tree model after optimization pruning can be used as a model for judging whether the smart home device is convenient.

[0089] In the embodiments of the present application, by optimizing and pruning the trained convenience decision tree model, overfitting of the convenience decision tree model can be avoided, and the generalization ability of the convenience decision tree model can be improved.

[0090] The above is a specific implementation of the convenience decision tree model training method provided by the embodiments of the present application. The convenience decision tree model trained by the above method can be used in the function information pushing method provided by the following embodiments.

[0091] First, the function information pushing method provided by the embodiments of the present application is introduced.

[0092] Figure 5 The flowchart of the function information pushing method provided by an embodiment of the present application is shown. As shown in Figure 5 The function information pushing method includes:

[0093] S510, obtaining a plurality of function information corresponding to a smart home device.

[0094] In the embodiments of the present application, the smart home device can include function information of the smart home device in the user's home. It can be understood that the user's home can include a plurality of smart home devices, and each smart home device includes a plurality of corresponding function information.

[0095] S520, inputting the plurality of function information into the convenience decision tree model to obtain the convenience information corresponding to each function information.

[0096] In the embodiments of the present application, the plurality of function information corresponding to one smart home device can be input into the convenience decision tree model, so as to determine the convenience information of each function for the smart home device. Alternatively, the plurality of function information corresponding to one or more smart home devices can be input into the convenience decision tree model at the same time, so as to determine the convenience information of each function for the respective smart home device.

[0097] S530, inputting the plurality of function information and the convenience information corresponding to each function information into the SHAP model to obtain the contribution value corresponding to each function information.

[0098] In the embodiments of the present application, the SHAP model can be used to analyze the plurality of function information and the convenience information corresponding to each function information, so as to determine the contribution value corresponding to each function information, that is, through the corresponding contribution value, it can be determined which function information has a greater influence on user satisfaction and convenience.

[0099] In the embodiments of the present application, the Shaply value corresponding to each function information can be calculated, and the Shaply value is taken as the contribution value of each function information. However, since the amount of data corresponding to the smart home device and its functions is large, in the embodiments of the present application, the Monte Carlo Sampling method can be used to approximately calculate the Shaply value of each function information.

[0100] In the embodiments of the present application, Figure 6 is a flowchart of one implementation of S530 provided by an embodiment of the present application, as shown in Figure 6 The method comprises the following steps.

[0101] S531, the plurality of function information is combined to obtain a plurality of function information groups.

[0102] In the embodiments of the present application, the function information group can be obtained by sampling and combining the plurality of function information based on a preset sampling strategy.

[0103] S532, any one of the plurality of function information is selected as the target function information.

[0104] S533, in the case of adding the target function information to the function information group, the first convenience information corresponding to the target function information is determined, and in the case of not adding the target function information to the function information group, the second convenience information is calculated.

[0105] In the embodiments of the present application, the first convenience information in the case of adding the target function information to the obtained function information group and the second convenience information in the case of not adding the target function information to the function information group can be calculated respectively.

[0106] S534, the first convenience information and the second convenience information are weighted and summed to obtain the SHAP value corresponding to the target function information.

[0107] In the embodiments of the present application, the first convenience information and the second convenience information are weighted and summed to obtain the SHAP value corresponding to the target function information. It can be understood that the SHAP value output by the SHAP model can be represented by the Shaply value.

[0108] In the embodiments of the present application, the SHAP value corresponding to the target function information can be calculated by the following formula:

[0109]

[0110] Among them, x can represent the function information x ijcorresponding Shaply value. M is the sampling number. S m is the function information set of the mth sampling. f(x ij ) and respectively represent the prediction result when the ith feature x ij is included and the prediction result when the ith feature x ij is not included. SamplingWeight(S i , i) represents the weight given to the ith feature x ij when the function information set S m is added in the sampling process, which can be used to reflect the weighted property of the Shapley value theory formula.

[0111] S535, calculates the product of the SHAP value corresponding to the target function information and the target function information, and takes the product as the contribution value of the target function information.

[0112] In the embodiments of the present application, the product of the calculated shaply value of the target function information and the target function information can be taken as the contribution value of the target function information.

[0113] In the embodiments of the present application, the contribution value corresponding to each function information can be calculated by the following formula:

[0114]

[0115] wherein x ij is the input function information of the convenience decision tree model. φ ij represents the corresponding Shapley value when the function information is x ij . φ0 is the model intercept.

[0116] S536, judges whether the contribution values corresponding to all function information are calculated, if yes, step S537 is executed; if not, step S532 is returned.

[0117] S537, outputs each function information and the corresponding contribution value.

[0118] In the embodiments of the present application, the Shaply value of each function information is calculated by combining the Monte Carlo estimation method, so as to reduce the calculation complexity and efficiently estimate the Shaply value corresponding to each function information in each intelligent device.

[0119] S540, according to the contribution value, determines the target function information meeting the preset condition from the plurality of function information.

[0120] In the embodiments of the present application, the contribution value corresponding to different function information can be used to determine the function information that can be pushed to the customer in the plurality of function information, that is, to determine the target function information. It can be understood that the contribution value can include positive and negative values. Among them, the positive value represents that the function information has a positive impact on improving convenience, and the negative value can represent that the function information has a negative impact on improving convenience. In an example, the related technical personnel can set the function information with a contribution value greater than a preset threshold as the target function information.

[0121] In the embodiments of the present application, the data output by the SHAP model has many characteristics, which can be explained by the function information in the smart home device in Table 2 below, and the contribution to the overall convenience and satisfaction of the user.

[0122] Table 2

[0123]

[0124] S550, pushing the target function information to the user.

[0125] In the embodiments of the present application, by obtaining a plurality of function information corresponding to the smart home device, and inputting the plurality of function information into the convenience decision tree model, the convenience information corresponding to each function information is obtained. By analyzing the contribution value corresponding to each function information through the SHAP model, the target function information meeting the preset condition is pushed to the user, so as to ensure that the function of the smart home device most meeting the needs of the user is recommended, and the needs of the user for intelligent, personalized and efficient smart home life are met.

[0126] Based on the function information pushing method provided in the above embodiments, the present application also provides a specific implementation mode of the function information pushing device. Please refer to the following embodiments.

[0127] Firstly, referring to Figure 7 The function information pushing device 700 provided by the embodiments of the present application includes the following modules:

[0128] The first acquisition module 701 is configured to acquire a plurality of function information corresponding to the smart home device;

[0129] The second acquisition module 702 is configured to input the plurality of function information into the convenience decision tree model to obtain the convenience information corresponding to each function information;

[0130] The third acquisition module 703 is configured to input the plurality of function information and the convenience information corresponding to each function information into the SHAP model to obtain the contribution value corresponding to each function information;

[0131] The determining module 704 is configured to determine, from the plurality of function information, target function information satisfying a preset condition according to the contribution value.

[0132] The pushing module 705 is configured to push the target function information to the user.

[0133] The function information pushing apparatus provided by the embodiment of the present application obtains a plurality of function information corresponding to the smart home device, and inputs the plurality of function information into the convenience decision tree model, so as to obtain the convenience information corresponding to each function information. The SHAP model is used to analyze the contribution value corresponding to each function information, and then the target function information satisfying the preset condition is pushed to the user, so as to ensure that the function of the smart home device most meeting the demand of the user can be recommended to the user, and the demand of the user for the smart home life in the aspects of intelligence, individualization and high efficiency can be met.

[0134] As an implementation manner of the present application, before the plurality of function information is input into the convenience decision tree model to obtain the convenience information corresponding to each function information, the apparatus further comprises a training module 706 configured to obtain a first training sample set, the first training sample set comprising a plurality of training samples, each training sample comprising a plurality of function information samples corresponding to the smart home device and a convenience label of the smart home device; and train the to-be-trained decision tree model by using the first training sample until a training stop condition is met, so as to obtain the convenience decision tree model.

[0135] As an implementation manner of the present application, the training module trains the preset decision tree model by using the training sample in the following manner until the training stop condition is met, so as to obtain the convenience decision tree model: calculating a sample entropy value corresponding to the first training sample set; and performing the following steps A to C in a loop until a preset stop condition is met, so as to obtain the node quantity information of the decision tree model and the function information corresponding to the nodes: step A: classifying the first training sample set based on each function information sample, so as to obtain a plurality of second training sample sets; step B: calculating a conditional entropy value of the second training sample set; and step C: determining the node quantity information of the decision tree model and the function information corresponding to the nodes by using the sample entropy value and the conditional entropy value; and constructing the convenience decision tree model based on the node quantity information and the function information corresponding to each node.

[0136] As an implementation manner of the present application, the training module determines the node quantity information of the decision tree model and the function information corresponding to the nodes by using the sample entropy value and the conditional entropy value in the following manner: calculating the entropy gain between each conditional entropy value and the sample entropy value; determining a target second training sample set with the maximum entropy gain; and taking the function information sample corresponding to the target second training sample set as the function information corresponding to the node of the convenience model.

[0137] As an implementation form of the present application, the training module is further configured to verify the function information corresponding to each node of the convenience decision tree model in a sequence from low to high of the node levels of the convenience decision tree model, and optimize and prune the convenience decision tree model based on the verification result to obtain an optimized convenience decision tree model.

[0138] As an implementation form of the present application, the function information sample includes at least one of the following: a function type, usage information, state information, environment information, and interaction information of the smart home device; and the convenience label of the smart home device includes at least user feedback information.

[0139] As an implementation form of the present application, the third obtaining module calculates the contribution value corresponding to each function information in the following manner: combining the plurality of function information to obtain a plurality of function information groups; selecting any one of the plurality of function information as a target function information; for each target function information, performing the following steps: determining first convenience information corresponding to the target function information in the case of adding the target function information to the function information group, and calculating second convenience information in the case of not adding the target function information to the function information group; performing weighted summation on the first convenience information and the second convenience information to obtain a SHAP value corresponding to the target function information; and calculating a product of the SHAP value corresponding to the target function information and the target function information, and taking the product as the contribution value of the target function information.

[0140] Figure 8 A hardware structure 800 schematic diagram of the function information pushing provided by the embodiments of the present application is shown.

[0141] The function information pushing device can include a processor 801 and a memory 802 storing computer program instructions.

[0142] Specifically, the processor 801 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or can be configured to implement one or more integrated circuits of the embodiments of the present application.

[0143] The memory 802 can include mass storage for data or instructions. As an example and not by way of limitation, the memory 802 can include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disc (e.g., a compact disc (CD) or a digital versatile disc (DVD)), a solid-state drive (SSD), a USB drive, or a combination of two or more of these. Where appropriate, the memory 802 can include removable or non-removable (or fixed) media, where appropriate. The memory 802 can be internal or external to the integrated gateway disaster recovery appliance. In particular embodiments, the memory 802 is non-volatile, solid-state memory.

[0144] The memory can include read-only memory (ROM), random-access memory (RAM), magnetic disk storage mediums, optical storage mediums, flash memory devices, electrical, optical, or other physical / tangible memory storage devices. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software that, when executed (by one or more processors), is operable to access the data and / or instructions that enable the functionality described with respect to the methods according to an aspect of the present disclosure.

[0145] The processor 801 implements any of the functional information push methods described above by reading and executing computer program instructions stored in the memory 802.

[0146] In one example, the functional information push device can further include a communication interface 803 and a bus 810. As shown, the processor 801, the memory 802, the communication interface 803 are connected through the bus 810 and complete the communication between each other. Figure 8

[0147] The communication interface 803 is mainly used to realize the communication between each module, device, unit and / or equipment in the embodiments of the present application.

[0148] ​Bus 810 includes a hardware, software, or both that couples components of the online data traffic billing device to each other. As an example and not by way of limitation, bus can include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand (IB) interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association local (VLB) bus, or another suitable bus or a combination of two or more of these. Where appropriate, bus 810 can include one or more buses. Although this application describes and shows a particular bus, this application contemplates any suitable bus or interconnect.

[0149] The function information pushing device can execute the function information pushing method in the embodiments of the application based on the function information corresponding to the current smart home device of the user, thereby realizing the function information pushing method and device described in combination Figure 2 and Figure 7 with the above embodiments.

[0150] In addition, in combination with the function information pushing method in the above embodiments, the embodiments of the application can provide a computer storage medium for implementation. The computer storage medium has computer program instructions stored thereon; the computer program instructions are executed by a processor to implement any one of the online data traffic billing methods in the above embodiments.

[0151] It needs to be clear that the application is not limited to the specific configurations and processes described above and shown in the drawings. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between steps, after understanding the spirit of the application.

[0152] The functions noted in the description of the structural block diagrams above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, and the like. When implemented in software, the elements of the present application are program or code segments that are used to perform the required tasks. The program or code segments can be stored in a machine-readable medium, or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. A "machine-readable medium" includes any medium that can store or transport information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, and the like. The code segments can be downloaded via computer networks such as the Internet, intranets, and the like.

[0153] It is also important to note that the examples mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the steps mentioned above, that is, the steps can be performed in the order mentioned in the examples, or in an order different from the examples, or several steps can be performed simultaneously.

[0154] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer program instructions can also be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other processing device to operate in a particular manner, such that the instructions stored in the computer readable medium produce an article of manufacture including instructions which implement the function / act specified in the flowchart and / or block diagram block or blocks. The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer program instructions can also be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other processing device to operate in a particular manner, such that the instructions stored in the computer readable medium produce an article of manufacture including instructions which implement the function / act specified in the flowchart and / or block diagram block or blocks.

[0155] The above merely describes a specific implementation of the present application. Those skilled in the art can clearly understand the specific working processes of the system, modules and units described above for the convenience and brevity of description, and can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein again. It should be understood that the protection scope of the present application is not limited to this, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. A method of pushing functional information, characterized by, The method comprises the following steps: obtaining a plurality of function information corresponding to a smart home device; inputting the plurality of function information into a convenience decision tree model to obtain convenience information corresponding to each function information, wherein the convenience information of the smart home device at least includes user feedback information; inputting the plurality of function information and the convenience information corresponding to each function information into a SHAP model to obtain a contribution value corresponding to each function information; determining target function information meeting a preset condition from the plurality of function information according to the contribution value; pushing the target function information to a user; the step of inputting the plurality of function information and the convenience information corresponding to each function information into the SHAP model to calculate the contribution value corresponding to each function information comprises the following steps: combining the plurality of function information to obtain a plurality of function information groups; selecting any one of the plurality of function information as target function information; for each target function information, the following steps are executed: determining first convenience information corresponding to the target function information in the case of adding the target function information to the function information group, and calculating second convenience information in the case of not adding the target function information to the function information group; weighting and summing the first convenience information and the second convenience information to obtain a SHAP value corresponding to the target function information; calculating the product of the SHAP value corresponding to the target function information and the target function information, and taking the product as the contribution value of the target function information.

2. The functional information push method according to claim 1, characterized by, Before the step of inputting the plurality of function information into the convenience decision tree model to obtain the convenience information corresponding to each function information, the method further comprises the following steps: obtaining a first training sample set, wherein the first training sample set comprises a plurality of training samples, and each training sample comprises a plurality of function information samples corresponding to the smart home device and convenience information of the smart home device; training a to-be-trained decision tree model by using the first training sample set until a training stop condition is met to obtain the convenience decision tree model.

3. The method of claim 2, wherein, The step of training the preset decision tree model by using the training sample until the training stop condition is met to obtain the convenience decision tree model comprises the following steps: calculating a sample entropy value corresponding to the first training sample set; recursively executing the following steps A to C until a preset stop condition is met to obtain node quantity information in the decision tree model and function information corresponding to each node: step A: classifying the first training sample set based on each function information sample to obtain a plurality of second training sample sets; step B: calculating a conditional entropy value of the second training sample set; step C: determining node quantity information of the decision tree model and function information corresponding to each node by using the sample entropy value and the conditional entropy value; constructing the convenience decision tree model based on the node quantity information and the function information corresponding to each node.

4. The method of claim 3, wherein, The step C comprises the following steps: calculating an entropy gain between each conditional entropy value and the sample entropy value, respectively; determining a target second training sample set with the maximum entropy gain; Corresponding functional information of a target second training sample set is taken as functional information corresponding to a node of the convenience decision tree model.

5. The method of claim 2, wherein, After the convenience decision tree model is obtained, the method further comprises: According to a node level of the convenience decision tree model from low to high, the functional information corresponding to each node of the convenience decision tree model is verified; Based on the verification result, the convenience decision tree model is optimized and pruned to obtain an optimized convenience decision tree model.

6. The method of claim 1, wherein, The functional information comprises at least one of the following: a function type, usage information, state information, environment information, and interaction information of the smart home device.

7. A functional information push device characterized by comprising: The device comprises: A first obtaining module is configured to obtain a plurality of functional information corresponding to a smart home device; A second obtaining module is configured to input the plurality of functional information into a convenience decision tree model to obtain convenience information corresponding to each functional information, wherein the convenience information of the smart home device at least comprises user feedback information; A third obtaining module is configured to input the plurality of functional information and the convenience information corresponding to each functional information into a SHAP model to obtain a contribution value corresponding to each functional information, and is further configured to combine the plurality of functional information to obtain a plurality of functional information groups; select any one of the plurality of functional information as a target functional information; for each target functional information, perform the following steps: determine first convenience information corresponding to the target functional information in a case where the target functional information is added to the functional information group, and calculate second convenience information in a case where the target functional information is not added to the functional information group; perform weighted summation on the first convenience information and the second convenience information to obtain a SHAP value corresponding to the target functional information; calculate a product of the SHAP value corresponding to the target functional information and the target functional information, and take the product as the contribution value of the target functional information; A determining module is configured to determine a target functional information satisfying a preset condition from the plurality of functional information according to the contribution value; A pushing module is configured to push the target functional information to a user.

8. A function information push device characterized by comprising: The device comprises a processor and a memory storing computer program instructions; The processor executes the computer program instructions to implement the functional information pushing method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer program instructions, and the computer program instructions are executed by the processor to implement the functional information pushing method according to any one of claims 1-6.

10. A computer program product, characterised in that, The instructions in the computer program product are executed by the processor of the electronic device to enable the electronic device to perform the functional information pushing method according to any one of claims 1-6.

Images