Information recommendation method and device, computer storage medium and electronic equipment
By obtaining and analyzing the portrait information of business promotion personnel and promotion objects, and using the information recommendation model to predict the probability of transaction completion, the problem of low promotion efficiency in the existing technology is solved, and more accurate transaction object identification and promotion effects are achieved.
Patent Information
- Application Number
- CN202411162669.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-29
AI Technical Summary
In the prior art, when business promotion personnel contact multiple promotion objects, it is difficult for them to effectively identify the objects that are most likely to trade with the business service platform, resulting in low promotion efficiency.
By obtaining portrait information of business promotion personnel and promotion objects, using the trained information recommendation model to predict the transaction completion probability between each promotion object and the business service platform, and filtering the target promotion objects based on this probability and recommending them to business promotion personnel.
It improves the accuracy of the probability of transaction completion, helps business promotion personnel identify potential transaction objects, thereby improving promotion efficiency and work targeting.
Smart Images

Figure CN119048161A_ABST
Abstract
Description
Background Art
[0002] With the rapid development and advancement of computer and Internet technologies, the field of relevant information recommendation is also booming.
[0003] In the related art, in order to promote the business development of the business service platform, business promotion is generally carried out by contacting the promotion targets through business promotion personnel. However, how to determine the target that is most likely to conduct transactions with the business service platform from multiple promotion targets that have been contacted in order to improve the promotion efficiency of business promotion personnel has become a focus of attention for relevant technical personnel.
[0004] In view of this, there is an urgent need in this field to develop a new information recommendation method and device.
[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure. Summary of the invention
[0006] The purpose of the present disclosure is to provide an information recommendation method, an information recommendation device, a computer storage medium and an electronic device, thereby overcoming the technical problem of low promotion efficiency of business promoters due to the limitations of related technologies, at least to a certain extent.
[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.
[0008] According to a first aspect of the present disclosure, there is provided an information recommendation method, comprising:
[0009] Acquire first portrait information of a business promoter of a business service platform and second portrait information of each promotion object among a plurality of promotion objects associated with the business promoter;
[0010] Predicting the probability of a transaction between each promotion object and the business service platform using the first portrait information and the second portrait information;
[0011] A target promotion object is screened from the multiple promotion objects according to the multiple transaction achievement probabilities, and information of the target promotion object is recommended to the business promotion personnel.
[0012] In an exemplary embodiment of the present disclosure, the step of predicting the probability of a transaction between each promotion object and the business service platform using the first portrait information and the second portrait information includes:
[0013] The first portrait information and the second portrait information are input into a trained information recommendation model, and the probability of transaction completion between each promotion object and the business service platform is output through the trained information recommendation model.
[0014] In an exemplary embodiment of the present disclosure, the trained information recommendation model includes a first sub-model;
[0015] The step of inputting the first portrait information and the second portrait information into a trained information recommendation model, and outputting the probability of transaction completion between each promotion object and the business service platform through the trained information recommendation model, comprises:
[0016] Counting a first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions;
[0017] If the first number is less than a first preset threshold, inputting the first portrait information and the second portrait information into a first sub-model of the information recommendation model, and outputting a first probability that each promotion object has a preliminary transaction intention through the first sub-model;
[0018] The first probability is determined as a probability of transaction completion between each promotion object and the business service platform.
[0019] In an exemplary embodiment of the present disclosure, the trained information recommendation model further includes a second sub-model;
[0020] After counting the first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions, the method further includes:
[0021] If the first number is greater than or equal to the first preset threshold, counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects;
[0022] If the second number is less than a second preset threshold, inputting the first portrait information and the second portrait information into a first sub-model of the information recommendation model, and outputting a second probability that each promotion object has a preliminary transaction intention through the first sub-model;
[0023] Inputting the first portrait information and the second portrait information into the second sub-model of the information recommendation model, and outputting a third probability that each promotion object belongs to a potential transaction object through the second sub-model;
[0024] The probability of transaction completion between each promotion object and the business service platform is determined according to the product of the second probability and the third probability.
[0025] In an exemplary embodiment of the present disclosure, the trained information recommendation model further includes a third sub-model;
[0026] After counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects, the method further includes:
[0027] If the second number is greater than or equal to the second preset threshold, inputting the first portrait information and the second portrait information into the first sub-model of the information recommendation model, and outputting a fourth probability that each promotion object has a preliminary transaction intention through the first sub-model;
[0028] Inputting the first portrait information and the second portrait information into a second sub-model of the information recommendation model, and outputting a fifth probability that each promotion object belongs to a potential transaction object through the second sub-model;
[0029] Inputting the first portrait information and the second portrait information into a third sub-model of the information recommendation model, and outputting a sixth probability of each promotion object being converted into an actual transaction object through the third sub-model;
[0030] The probability of transaction completion between each promotion object and the business service platform is determined according to the product of the fourth probability, the fifth probability and the sixth probability.
[0031] In an exemplary embodiment of the present disclosure, the information recommendation model to be trained includes a first sub-model to be trained, a second sub-model to be trained, and a third sub-model to be trained;
[0032] The trained information recommendation model is trained in the following way:
[0033] Obtain a training sample set; the training sample set includes multiple training samples and labels corresponding to each of the training samples; each of the training samples includes first sample portrait information of each sample business promoter and second sample portrait information of each historical promotion object associated with each sample business promoter;
[0034] Input the plurality of training samples into the first sub-model to be trained to obtain a first prediction result for each of the training samples, and determine a first loss value according to the degree of difference between the plurality of first prediction results and the plurality of labels; the first prediction result is used to characterize the probability that the historical promotion object in each of the training samples has a preliminary transaction intention;
[0035] Input the plurality of training samples into the second sub-model to be trained to obtain a second prediction result for each of the training samples, determine a plurality of first comprehensive prediction results according to the plurality of first prediction results and the plurality of second prediction results, and determine a second loss value according to the degree of difference between the plurality of first comprehensive prediction results and the plurality of labels; the second prediction result is used to characterize the probability that the historical promotion object in each of the training samples belongs to a potential transaction object;
[0036] Input the plurality of training samples into the third sub-model to be trained to obtain a third prediction result for each of the training samples, determine a plurality of second comprehensive prediction results according to the plurality of the first comprehensive prediction results and the plurality of the third prediction results, and determine a third loss value according to the degree of difference between the plurality of the second comprehensive prediction results and the plurality of the labels; the third prediction result is used to characterize the probability of the historical promotion object in each of the training samples being converted into an actual transaction object;
[0037] Determine an overall loss value according to the first loss value, the second loss value and the third loss value;
[0038] When the overall loss value satisfies a preset first convergence condition, the trained information recommendation model is obtained.
[0039] In an exemplary embodiment of the present disclosure, the label corresponding to each of the training samples includes a first label, a second label and a third label; the first label is used to characterize the probability that each of the historical promotion objects has a preliminary transaction intention, the second label is used to characterize the probability that each of the historical promotion objects is a potential transaction object, and the third label is used to characterize the probability that each of the historical promotion objects is converted into an actual transaction object.
[0040] In an exemplary embodiment of the present disclosure, determining the first loss value according to the degree of difference between the plurality of first prediction results and the plurality of labels includes:
[0041] Obtaining a difference degree between each of the first prediction results and each of the first labels to obtain multiple difference degrees;
[0042] The first loss value is determined according to the accumulated values of the multiple difference degrees.
[0043] In an exemplary embodiment of the present disclosure, determining the second loss value according to the degree of difference between the plurality of first comprehensive prediction results and the plurality of labels includes:
[0044] Determine each first comprehensive label according to each first label and each second label;
[0045] Obtaining a difference degree between each of the first comprehensive prediction results and each of the first comprehensive labels to obtain a plurality of difference degrees;
[0046] The second loss value is determined according to the multiple difference degrees.
[0047] In an exemplary embodiment of the present disclosure, determining the third loss value according to the degree of difference between the plurality of the second comprehensive prediction results and the plurality of the labels includes:
[0048] Determine each second comprehensive label according to each of the first labels, each of the second labels and each of the third labels;
[0049] Obtaining a difference degree between each of the second comprehensive prediction results and each of the second comprehensive labels to obtain a plurality of difference degrees;
[0050] The third loss value is determined according to the multiple difference degrees.
[0051] In an exemplary embodiment of the present disclosure, the multiple historical promotion objects associated with each sample business promoter include positive sample objects and negative sample objects, the positive sample objects are promotion objects that have preliminary transaction intentions and / or are converted into actual transaction objects, and the negative sample objects are promotion objects that do not have preliminary transaction intentions and have not been converted into actual transaction objects.
[0052] According to a second aspect of the present disclosure, there is provided an information recommendation device, comprising:
[0053] An information acquisition module, used to acquire first portrait information of a business promoter of a business service platform and second portrait information of each promotion object among a plurality of promotion objects associated with the business promoter;
[0054] A prediction module, used to predict the probability of a transaction between each promotion object and the business service platform by using the first portrait information and the second portrait information;
[0055] The information recommendation module is used to select a target promotion object from the multiple promotion objects according to the multiple transaction completion probabilities, and recommend the information of the target promotion object to the business promotion personnel.
[0056] According to a third aspect of the present disclosure, a computer storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the information recommendation method described in the first aspect is implemented.
[0057] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the information recommendation method described in the first aspect above by executing the executable instructions.
[0058] It can be seen from the above technical solutions that the information recommendation method, information recommendation device, computer storage medium and electronic device in the exemplary embodiments of the present disclosure have at least the following advantages and positive effects:
[0059] In the technical solutions provided by some embodiments of the present disclosure, on the one hand, by obtaining the first portrait information of the business promotion personnel of the business service platform and the second portrait information of each promotion object among the multiple promotion objects associated with the business promotion personnel, and using the first portrait information and the second portrait information to predict the transaction completion probability between each promotion object and the business service platform, it is possible to more accurately understand the characteristics of the business promotion personnel and the promotion objects, thereby helping to improve the accuracy of the subsequent predicted transaction completion probability. On the other hand, according to multiple transaction completion probabilities, target promotion objects are screened from multiple promotion objects, and the information of the target promotion objects is recommended to the business promotion personnel, which can help the business promotion personnel identify those target promotion objects that are more likely to reach a transaction with the business service platform, so that the business promotion personnel can more targetedly carry out accurate and personalized business promotion work and improve their promotion efficiency.
[0060] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.
[0062] Figure 1 A schematic diagram showing a flow chart of an information recommendation method in an embodiment of the present disclosure;
[0063] Figure 2 A schematic diagram showing a flow chart of how to obtain a trained information recommendation model in an embodiment of the present disclosure;
[0064] Figure 3 A flow chart showing how to determine a second loss value according to the degree of difference between a plurality of first comprehensive prediction results and a plurality of labels in an embodiment of the present disclosure;
[0065] Figure 4A flow chart showing how to determine a third loss value according to the degree of difference between a plurality of second comprehensive prediction results and a plurality of labels in an embodiment of the present disclosure;
[0066] Figure 5 A schematic diagram showing a flow chart of how to predict the probability of a transaction between each promotion object and a business service platform in an embodiment of the present disclosure;
[0067] Figure 6 A schematic diagram showing another flow chart of how to determine the probability of a transaction being completed between a promotion object and a business service platform in an embodiment of the present disclosure;
[0068] Figure 7 A schematic diagram showing a flow chart of how to determine the probability of a transaction being completed between each promotion object and the business service platform in another embodiment of the present disclosure;
[0069] Figure 8 A schematic diagram showing how to predict the probability of a transaction between each promotion object and the business service platform in the cold start phase in an embodiment of the present disclosure;
[0070] Fig. 9 A schematic diagram showing the overall process of the information recommendation method in an embodiment of the present disclosure is shown;
[0071] Fig.10 A schematic diagram showing the structure of an information recommendation device in an exemplary embodiment of the present disclosure;
[0072] Fig.11 A schematic structural diagram of an electronic device in an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION
[0073] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.
[0074] The terms "a", "an", "the" and "said" are used in this specification to indicate the presence of one or more elements / components / etc.; the terms "including" and "having" are used to express an open-ended inclusion and mean that additional elements / components / etc. may exist in addition to the listed elements / components / etc.; the terms "first" and "second" etc. are used only as labels and are not intended to limit the quantity of their objects.
[0075] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and their repeated descriptions will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities.
[0076] In the embodiments of the present disclosure, firstly, an information recommendation method is provided, which at least to some extent overcomes the defect of low promotion efficiency of business promoters in the related art.
[0077] Figure 1 A schematic flow chart of an information recommendation method in an embodiment of the present disclosure is shown. The execution subject of the information recommendation method may be a server that performs information recommendation.
[0078] refer to Figure 1 According to an embodiment of the present disclosure, the information recommendation method includes the following steps:
[0079] Step S110, obtaining first portrait information of each business promotion personnel of the business service platform and second portrait information of each promotion object among multiple promotion objects associated with each business promotion personnel;
[0080] Step S120, using the first portrait information and the second portrait information to predict the probability of a transaction being completed between each promotion object and the business service platform;
[0081] Step S130, screening a target promotion object from multiple promotion objects according to the transaction completion probability, and recommending the information of the target promotion object to the business promotion personnel.
[0082] exist Figure 1In the technical solution provided by the illustrated embodiment, on the one hand, by obtaining the first portrait information of the business promotion personnel of the business service platform and the second portrait information of each promotion object among the multiple promotion objects associated with the business promotion personnel, and using the first portrait information and the second portrait information to predict the transaction completion probability between each promotion object and the business service platform, it is possible to more accurately understand the characteristics of the business promotion personnel and the promotion object, thereby helping to improve the accuracy of the subsequent predicted transaction completion probability. On the other hand, by screening the target promotion object from the multiple promotion objects according to the multiple transaction completion probabilities, and recommending the information of the target promotion object to the business promotion personnel, it is possible to help the business promotion personnel identify those target promotion objects that are more likely to reach a transaction with the business service platform, so that the business promotion personnel can carry out more targeted, accurate and personalized business promotion work, thereby improving their promotion efficiency.
[0083] Before step S110, it should be noted that the present disclosure can pre-train an information recommendation model, and subsequently the trained information recommendation model can be used to predict the probability of transaction completion between the promotion object and the business service platform.
[0084] Among them, the business service platform can be a related platform that provides business services, such as a logistics service platform, an e-commerce service platform, etc., which can be set up according to actual conditions. The business service platform can be equipped with business promoters (also known as transaction lead cultivators). The main responsibility of the business promoters is to communicate or contact with the promotion targets to facilitate the establishment of a transaction relationship between the promotion targets and the business service platform.
[0085] It should be noted that the information recommendation model to be trained may include a first sub-model to be trained, a second sub-model to be trained and a third sub-model to be trained. Figure 2 The specific implementation of how to train the information recommendation model to be trained to obtain the trained information recommendation model in the present disclosure is described.
[0086] refer to Figure 2 , Figure 2 A flowchart showing how to obtain a trained information recommendation model in an embodiment of the present disclosure includes steps S201 to S206:
[0087] In step S201, a training sample set is obtained.
[0088] In this step, a training sample set may be obtained, and the training sample set may include multiple training samples and a label corresponding to each training sample.
[0089] Each of the above training samples may include first sample portrait information of each sample business promoter and second sample portrait information of each historical promotion object associated with each sample business promoter (ie, the object for which the sample business promoter promoted business during the historical period).
[0090] It should be noted that the above-mentioned historical promotion objects may include positive sample objects and negative sample objects. Positive sample objects are promotion objects with preliminary transaction intentions and / or converted into actual transaction objects, and negative sample objects are promotion objects without preliminary transaction intentions and not converted into actual transaction objects. Therefore, the present disclosure can solve the technical problems of sample bias and data sparsity caused by only using positive sample objects to train models in related technologies.
[0091] Exemplarily, the first sample portrait information of each sample business promoter may include: gender, age, province, industry information sequence of actual transaction objects facilitated by each sample business promoter within a historical period (for example, the past year, which can be set according to actual conditions, and the present disclosure does not make any special restrictions on this), information on actual transaction objects facilitated by each sample business promoter within a historical period, etc., which can be set according to actual conditions, and the present disclosure does not make any special restrictions on this.
[0092] The second sample portrait information of each historical promotion object may include: the location, industry, telephone activity, number of transactions with the business service platform during the historical period, direct or indirect income obtained by the historical promotion object through transactions with the business service platform during the historical period, etc. of each historical promotion object, which can be set according to actual conditions, and the present disclosure does not make any special limitations on this.
[0093] During or after the communication with each historical promotion object, the sample business promoter can label each historical promotion object according to the communication situation. For example, the label corresponding to each training sample can include a first label, a second label and a third label, the first label is used to characterize the probability that each historical promotion object has a preliminary transaction intention (for example: 0 or 1, 0 means no preliminary transaction intention, 1 means preliminary transaction intention), the second label is used to characterize the probability that each historical promotion object belongs to a potential transaction object (for example: 0 or 1, 0 means not a potential transaction object, 1 means a potential transaction object), and the third label is used to characterize the probability that each historical promotion object is converted into an actual transaction object (for example: 0 or 1, 0 means no contract has been signed and not converted into an actual transaction object, 1 means the contract has been signed and converted into an actual transaction object).
[0094] In step S202, multiple training samples are input into a first sub-model to be trained to obtain a first prediction result for each training sample, and a first loss value is determined according to the degree of difference between the multiple first prediction results and the multiple labels.
[0095] In this step, after obtaining the above-mentioned training sample set, the above-mentioned multiple training samples can be input into the first sub-model to be trained to obtain a first prediction result for each training sample (the first prediction result is used to characterize the probability that the historical promotion object in each training sample has a preliminary transaction intention), and then, the first loss value can be determined according to the degree of difference between the multiple first prediction results and the multiple first labels.
[0096] For example, the difference between each first prediction result and each first label can be calculated, and the obtained multiple difference degrees are accumulated, and then multiplied by the weighting coefficient α to obtain the first loss value. Specifically, the first loss value can be determined based on the following formula 1:
[0097]
[0098] Among them, L(θ IR ) represents the first loss value mentioned above, N represents the total number of training samples, x i Represents each training sample input, y i represents the first label corresponding to each training sample, f(x i θ IR ) represents the first prediction result for each training sample output by the first sub-model to be trained, l(y i ,f(x i θ IR )) represents the degree of difference between each first prediction result and the corresponding first label, Represents the cumulative value of the above multiple differences.
[0099] In step S203, multiple training samples are input into the second sub-model to be trained to obtain a second prediction result for each training sample, multiple first comprehensive prediction results are determined according to the multiple first prediction results and the multiple second prediction results, and a second loss value is determined according to the degree of difference between the multiple first comprehensive prediction results and the multiple labels.
[0100] In this step, the above-mentioned multiple training samples can be input into the second sub-model to be trained to obtain a second prediction result for each training sample (the second prediction result is used to characterize the probability that the sample promotion object and the historical promotion object in each training sample belong to a potential transaction object). Then, each first comprehensive prediction result can be determined according to the product between each first prediction result and each second prediction result to obtain multiple first comprehensive prediction results. After that, the second loss value can be determined according to the degree of difference between the multiple first comprehensive prediction results and the multiple labels.
[0101] Specifically, please refer to Figure 3 , Figure 3 A flow chart showing how to determine the second loss value according to the degree of difference between multiple first comprehensive prediction results and multiple labels in the embodiment of the present disclosure includes steps S301 to S303:
[0102] In step S301, each first comprehensive label is determined according to each first label and each second label.
[0103] In this step, the first comprehensive label can be determined based on the following method: if both the first label and the second label are 1, the first comprehensive label is 1; if either the first label or the second label is 0, the first comprehensive label is 0.
[0104] In step S302, multiple difference degrees are obtained according to the difference degree between each first comprehensive prediction result and each first comprehensive label.
[0105] In this step, the degree of difference between each first comprehensive prediction result and each first comprehensive label may be calculated to obtain multiple degrees of difference.
[0106] In step S303, a second loss value is determined according to a plurality of difference degrees.
[0107] In this step, the multiple difference degrees obtained may be accumulated and then multiplied by a weighting coefficient β to obtain a second loss value.
[0108] Specifically, the second loss value may be determined based on the following formula 2:
[0109]
[0110] Among them, L(θ OR ) represents the second loss value, N represents the total number of training samples, x i Represents each training sample input, y i Represents the first label corresponding to each training sample, z i Represents the second label corresponding to each training sample, y i &zi Represents each first comprehensive label, f(x i θ IR ) represents the first prediction result for each training sample output by the first sub-model to be trained, f(x i θ OR ) represents the second prediction result for each training sample output by the second sub-model to be trained, f(x i θ IR )*f(x i θ OR ) represents each first comprehensive prediction result, l(y i &z i ,f(x i θ IR )*f(x i θ OR )) represents the difference between each first comprehensive prediction result and the corresponding first comprehensive label, Represents the cumulative value of the above multiple differences.
[0111] In step S204, multiple training samples are input into the third sub-model to be trained to obtain a third prediction result for each training sample, multiple second comprehensive prediction results are determined according to the multiple first comprehensive prediction results and the multiple third prediction results, and a third loss value is determined according to the degree of difference between the multiple second comprehensive prediction results and the multiple labels.
[0112] In this step, the above-mentioned multiple training samples can be input into the third sub-model to be trained to obtain a third prediction result for each training sample (the third prediction result is used to characterize the probability of converting the historical promotion object in each training sample into an actual transaction object). Afterwards, each second comprehensive prediction result can be determined according to the product between each first comprehensive prediction result and each third prediction result to obtain multiple second comprehensive prediction results. Afterwards, the third loss value can be determined according to the degree of difference between the multiple second comprehensive prediction results and the multiple labels.
[0113] Specifically, refer to Figure 4 , Figure 4 A flow chart showing how to determine the third loss value according to the degree of difference between the plurality of second comprehensive prediction results and the plurality of labels in the embodiment of the present disclosure includes steps S401 to S403:
[0114] In step S401, each second comprehensive label is determined according to each first label, each second label and each third label.
[0115] In this step, the second comprehensive label can be determined based on the following method: if the first label, the second label and the third label are all 1, the second comprehensive label is 1; if any of the first label, the second label or the third label is 0, the second comprehensive label is 0.
[0116] In step S402, the difference degree between each second comprehensive prediction result and each second comprehensive label is obtained to obtain multiple difference degrees.
[0117] In this step, the degree of difference between each second comprehensive prediction result and each second comprehensive label may be calculated first to obtain multiple degrees of difference.
[0118] In step S403, a third loss value is determined according to the multiple difference degrees.
[0119] In this step, the multiple difference degrees obtained may be accumulated and then multiplied by a weighting coefficient γ to obtain a second loss value.
[0120] Specifically, the third loss value may be determined based on the following formula 3:
[0121]
[0122] Among them, L(θ SR ) represents the third loss value mentioned above, N represents the total number of training samples, x i Represents each training sample input, y i Represents the first label corresponding to each training sample, z i Represents the second label corresponding to each training sample, k i Represents the third label corresponding to each training sample, y i &z i represents each second comprehensive label, f(x i θ IR ) represents the first prediction result for each training sample output by the first sub-model to be trained, f(x i θ OR ) represents the second prediction result for each training sample output by the second sub-model to be trained, f(x i θ SR ) represents the third prediction result for each training sample output by the third sub-model to be trained, f(x i θ IR )*f(x i θ OR )*f(x i θ SR ) represents each second comprehensive prediction result, l(y i &z i &k i,f(x i θ IR )*f(x i θ OR )*f(x i θ SR )) represents the degree of difference between each second comprehensive prediction result and the corresponding second comprehensive label, Represents the cumulative value of the above multiple differences.
[0123] It should be noted that in order to avoid the adverse effects of too high or too low values of weighting coefficients on model accuracy, the present invention determines the values of the above-mentioned weighting coefficients α=0.3, β=0.5, γ=0.2 by weighting according to the relevant real parameters of the business service platform in the historical period.
[0124] In step S205, an overall loss value is determined according to the first loss value, the second loss value and the third loss value.
[0125] In this step, the overall loss value can be determined based on the cumulative value of the first loss value, the second loss value and the third loss value, so that the overall loss value L(θ IR ,θ OR ,θ SR ) can be expressed as the following formula 4:
[0126] L(θ IR ,θ OR ,θ SR )
[0127]
[0128] In step S206, when the overall loss value satisfies a preset first convergence condition, a trained information recommendation model is obtained.
[0129] In this step, the back propagation algorithm can be used to optimize the model parameters, so that when the above-mentioned overall loss value meets the preset first convergence condition (for example, less than the preset convergence threshold), a trained information recommendation model can be obtained.
[0130] After the above information recommendation model is trained, the model application phase can be entered to perform the process of recommending relevant information using the trained information recommendation model.
[0131] Next reference Figure 1 In step S110, first portrait information of a business promoter of a business service platform and second portrait information of each promotion object among a plurality of promotion objects associated with the business promoter are obtained.
[0132] In this step, the first portrait information of the business promotion personnel of the business service platform can be obtained. By way of example, the first portrait information of the business promotion personnel may include: gender, age, province, industry information sequence of actual transaction objects facilitated by each business promotion personnel within a historical period (for example, the past year, which can be set according to actual conditions, and the present disclosure does not make special restrictions on this), information on actual transaction objects facilitated by each business promotion personnel within a historical period, etc., which can be set according to actual conditions, and the present disclosure does not make special restrictions on this.
[0133] The second portrait information of each promotion object may include: the location, industry, telephone activity, number of transactions with the business service platform during the historical period, direct or indirect income obtained by the promotion object through transactions with the business service platform during the historical period, etc., of each promotion object. It can be set according to actual conditions, and this disclosure does not make any special limitations on this.
[0134] In step S120, the first portrait information and the second portrait information are used to predict the probability of transaction completion between each promotion object and the business service platform.
[0135] In this step, the first portrait information and the second portrait information mentioned above can be input into a trained information recommendation model, and the probability of transaction completion between each promotion object and the business service platform can be output through the trained information recommendation model.
[0136] Optionally, given that the information recommendation model in the present disclosure includes three sub-models, the present disclosure can also be configured at different application stages of the model to decide whether to select the entire model or a partial model to execute the prediction process of the transaction completion probability according to different situations.
[0137] Specifically, in the initial application stage of the above-mentioned information recommendation model (for example, within half a year after the model is put into use, which can also be called the cold start stage), you can choose to directly use the above-mentioned information recommendation model or use some sub-models in the above-mentioned information recommendation model to execute the prediction process of the transaction probability according to the actual situation of each business promotion personnel. After the above-mentioned initial application stage, you can directly use the above-mentioned information recommendation model to execute the prediction process of the transaction probability without distinguishing the actual situation of each business promotion personnel.
[0138] The following is a description of the specific implementation method of how to perform the prediction process of the transaction completion probability in the above initial application stage:
[0139] refer to Figure 5 , Figure 5 A flowchart showing how to predict the probability of a transaction between each promotion object and a business service platform in an embodiment of the present disclosure includes steps S501 to S503:
[0140] In step S501, a first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions is counted.
[0141] In this step, the first number of promotion objects associated with the above-mentioned business promotion personnel and having preliminary transaction intentions can be counted. num .
[0142] In step S502, if the first number is less than the first preset threshold, the first portrait information and the second portrait information are input into the first sub-model of the information recommendation model, and the first sub-model outputs the first probability that each promotion object has a preliminary transaction intention.
[0143] In this step, if the first number intuition num If the value of the first portrait information and the second portrait information is less than a first preset threshold value A (for example, A can be 500, or it can be set according to actual conditions, and the present disclosure does not make any special limitation on this), the first portrait information and the second portrait information can be input into the first sub-model of the information recommendation model, and the first sub-model can output the first probability that each promotion object has a preliminary transaction intention.
[0144] In step S503, the first probability is determined as a probability of transaction completion between each promotion object and the business service platform.
[0145] In this step, the first probability may be determined as a probability of transaction completion between each promotion object and the business service platform.
[0146] It should be noted that after the above step S501 (counting the first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions), if the above first number intuition num is greater than or equal to the first preset threshold A, then you can refer to Figure 6 , Figure 6 A flowchart showing another method of determining the transaction completion probability between the promotion object and the business service platform in an embodiment of the present disclosure includes steps S601 to S604:
[0147] In step S601, a second number of promotion objects associated with the business promotion personnel and which are potential transaction objects is counted.
[0148] In this step, the second number of promotion objects associated with the business promotion personnel and belonging to potential transaction objects may be counted. num .
[0149] In step S602, if the second number is less than the second preset threshold, the first portrait information and the second portrait information are input into the first sub-model of the information recommendation model, and the second probability that each promotion object has a preliminary transaction intention is output through the first sub-model.
[0150] In this step, if the second number chance num If the probability of each promotion object having a preliminary transaction intention is smaller than a second preset threshold value B (for example, 50, which can be set according to actual conditions and is not specifically limited in the present disclosure), the first portrait information and the second portrait information can be input into the first sub-model of the information recommendation model, and the first sub-model can be used to predict the second probability that each promotion object has a preliminary transaction intention.
[0151] In step S603, the first portrait information and the second portrait information are input into the second sub-model of the information recommendation model, and the third probability that each promotion object belongs to a potential transaction object is output through the second sub-model.
[0152] In this step, the first portrait information and the second portrait information may also be input into the second sub-model of the information recommendation model, and the third probability that each promotion object belongs to a potential transaction object may be output through the second sub-model.
[0153] In step S604, the probability of transaction completion between each promotion object and the business service platform is determined according to the product of the second probability and the third probability.
[0154] In this step, the probability of a transaction being completed between each promotion object and the business service platform can be determined based on the product of the second probability and the third probability.
[0155] It should be noted that after step S601 (counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects), if the second number chance num is greater than or equal to the second preset threshold B, then you can refer to Figure 7 , Figure 7 A flowchart showing how to determine the probability of a transaction between each promotion object and the business service platform in an embodiment of the present disclosure includes steps S701 to S704:
[0156] In step S701, the first portrait information and the second portrait information are input into the first sub-model of the information recommendation model, and the fourth probability that each promotion object has a preliminary transaction intention is output through the first sub-model.
[0157] In this step, the first portrait information and the second portrait information may be input into the first sub-model of the information recommendation model, and the fourth probability that each promotion object has a preliminary transaction intention may be output through the first sub-model.
[0158] In step S702, the first portrait information and the second portrait information are input into the second sub-model of the information recommendation model, and the fifth probability that each promotion object belongs to a potential transaction object is output through the second sub-model.
[0159] In this step, the first portrait information and the second portrait information may be input into the second sub-model of the information recommendation model, and the fifth probability that each promotion object belongs to a potential transaction object may be output through the second sub-model.
[0160] In step S703, the first portrait information and the second portrait information are input into the third sub-model of the information recommendation model, and the sixth probability of each promotion object being converted into an actual transaction object is output through the third sub-model.
[0161] In this step, the first portrait information and the second portrait information may be input into the third sub-model of the information recommendation model, and the sixth probability of each promotion object being converted into an actual transaction object may be output through the third sub-model.
[0162] In step S704, the probability of transaction completion between each promotion object and the business service platform is determined according to the product of the fourth probability, the fifth probability and the sixth probability.
[0163] In this step, the probability of achieving a transaction between each promotion object and the business service platform can be determined based on the product of the fourth probability, the fifth probability and the sixth probability.
[0164] After obtaining the transaction completion probability between each promotion object and the business service platform, you can then refer to Figure 1 In step S130, a target promotion object is screened from multiple promotion objects according to multiple transaction achievement probabilities, and information of the target promotion object is recommended to the business promotion personnel.
[0165] In this step, the target promotion object can be screened from the above-mentioned multiple promotion objects according to multiple transaction completion probabilities. For example, the promotion object with the highest transaction completion probability can be determined as the target promotion object, or the M promotion objects with transaction completion probabilities in the first M (M is an integer greater than 1) positions can be determined as the above-mentioned target promotion objects. Then, the information of the target promotion object is recommended to the business promotion personnel so that the business promotion personnel can focus on following up on it, thereby promoting transactions and cooperation between the target promotion object and the business service platform.
[0166] refer to Figure 8 , Figure 8 A schematic diagram showing how to predict the probability of a transaction between each promotion object and the business service platform in the cold start phase in an embodiment of the present disclosure is shown as follows: Figure 8 As shown:
[0167] For the first stage (when the first number of promotion objects associated with the business promoter and having preliminary transaction intention is less than the first preset threshold A), the first portrait information and the second portrait information can be input into the first sub-model of the information recommendation model, and the first probability P of each promotion object having preliminary transaction intention output by the first sub-model is calculated as IR Determine the probability of reaching a transaction between each promotion object and the business service platform;
[0168] For the second stage (when the first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions is greater than or equal to the first preset threshold A, and the second number of promotion objects associated with the business promotion personnel and being potential transaction objects is less than the second preset threshold B), the first portrait information and the second portrait information can be input into the first sub-model of the information recommendation model, and the second probability that each promotion object has a preliminary transaction intention is output through the first sub-model. At the same time, the first portrait information and the second portrait information are input into the second sub-model of the information recommendation model, and the third probability P that each promotion object is a potential transaction object is output through the second sub-model. OR , and then, the product of the second probability and the third probability P IOR Determine the probability of reaching a transaction between each promotion object and the business service platform;
[0169] For the third stage (when the first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions is greater than or equal to the first preset threshold A, and the second number of promotion objects associated with the business promotion personnel and being potential transaction objects is greater than or equal to the second preset threshold B), the first portrait information and the second portrait information can be input into the first sub-model of the information recommendation model, and the fourth probability that each promotion object has a preliminary transaction intention is output through the first sub-model, the first portrait information and the second portrait information are input into the second sub-model of the information recommendation model, and the fifth probability that each promotion object is a potential transaction object is output through the second sub-model, and the first portrait information and the second portrait information are input into the third sub-model of the information recommendation model, and the sixth probability P of each promotion object being converted into an actual transaction object is output through the third sub-model. SR , and then, the product P between the fourth probability, the fifth probability and the sixth probability IOSR , determine the probability of transaction completion between each promotion object and the business service platform;
[0170] In addition, by using a shared Embedding layer, the present disclosure enables the subtask of estimating the probability of transaction completion to be learned from samples that are interested but not signed, which can alleviate the problem of sparse training data.
[0171] refer to Fig. 9 , Fig. 9The overall flow chart of the information recommendation method in the embodiment of the present disclosure is shown as follows: Fig. 9 As shown:
[0172] First, the first portrait information of the business promotion personnel can be extracted, and the second portrait information of each promotion object can be extracted;
[0173] Next, feature fusion extraction can be performed;
[0174] Then, we enter the cold start phase of the model;
[0175] In the cold start phase of the model, the probability of each promotion object and the business service platform reaching a transaction can be determined based on one or more of the following estimated results (the probability that each promotion object has a preliminary transaction intention, the probability that each promotion object is a potential transaction object, and the probability that each promotion object is converted into an actual transaction object);
[0176] Afterwards, the target promotion objects can be screened according to the probability of transaction completion, and the business promotion personnel can focus on following up the target promotion objects to encourage the target promotion objects to complete transactions with the business service platform (sign contracts).
[0177] Based on the above technical solutions, the present disclosure has at least the following technical effects:
[0178] First, it fully considers the correlation between the promotion object from having initial transaction intention, to becoming a potential transaction object, to converting into an actual transaction object, thus helping to improve the accuracy of the prediction of the probability of transaction completion;
[0179] Second, the profile information of both business promoters and promoters is used to predict the probability of transaction completion, thereby comprehensively considering various personnel characteristics and helping to improve the accuracy of transaction completion probability prediction;
[0180] Third, three schemes are provided for predicting the probability of transaction completion between promotion objects and business service platforms in the cold start phase. Therefore, different models can be used to predict the probability of transaction completion in different situations, and ultimately provide business promoters with information on promotion objects that are most likely to achieve transactions, helping business promoters to generate transaction leads that are more likely to be signed, solving the problem of business promoters' substandard business volume and improving their work efficiency.
[0181] The present disclosure also provides an information recommendation device, Fig.10 A schematic diagram showing the structure of an information recommendation device in an exemplary embodiment of the present disclosure is shown; Fig.10 As shown, the information recommendation device 1000 may include an information acquisition module 1010, a prediction module 1020 and an information recommendation module 1030. Among them:
[0182] An information acquisition module 1010 is used to acquire first portrait information of a business promoter of a business service platform and second portrait information of each promotion object among a plurality of promotion objects associated with the business promoter;
[0183] Prediction module 1020, used to predict the probability of transaction completion between each promotion object and the business service platform by using the first portrait information and the second portrait information;
[0184] The information recommendation module 1030 is used to select a target promotion object from the multiple promotion objects according to the multiple transaction completion probabilities, and recommend the information of the target promotion object to the business promotion personnel.
[0185] In an exemplary embodiment of the present disclosure, the prediction module 1020 predicts the probability of transaction completion between each promotion object and the business service platform using the first portrait information and the second portrait information, including:
[0186] The first portrait information and the second portrait information are input into a trained information recommendation model, and the probability of transaction completion between each promotion object and the business service platform is output through the trained information recommendation model.
[0187] In an exemplary embodiment of the present disclosure, the trained information recommendation model includes a first sub-model;
[0188] The prediction module 1020 inputs the first portrait information and the second portrait information into a trained information recommendation model, and outputs the transaction completion probability between each promotion object and the business service platform through the trained information recommendation model, including:
[0189] Counting a first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions;
[0190] If the first number is less than a first preset threshold, inputting the first portrait information and the second portrait information into a first sub-model of the information recommendation model, and outputting a first probability that each promotion object has a preliminary transaction intention through the first sub-model;
[0191] The first probability is determined as a probability of transaction completion between each promotion object and the business service platform.
[0192] In an exemplary embodiment of the present disclosure, the trained information recommendation model further includes a second sub-model;
[0193] After counting the first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions, the prediction module 1020 is configured to:
[0194] If the first number is greater than or equal to the first preset threshold, counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects;
[0195] If the second number is less than a second preset threshold, inputting the first portrait information and the second portrait information into a first sub-model of the information recommendation model, and outputting a second probability that each promotion object has a preliminary transaction intention through the first sub-model;
[0196] Inputting the first portrait information and the second portrait information into the second sub-model of the information recommendation model, and outputting a third probability that each promotion object belongs to a potential transaction object through the second sub-model;
[0197] The probability of transaction completion between each promotion object and the business service platform is determined according to the product of the second probability and the third probability.
[0198] In an exemplary embodiment of the present disclosure, the trained information recommendation model further includes a third sub-model;
[0199] After counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects, the prediction module 1020 is configured to:
[0200] If the second number is greater than or equal to the second preset threshold, inputting the first portrait information and the second portrait information into the first sub-model of the information recommendation model, and outputting a fourth probability that each promotion object has a preliminary transaction intention through the first sub-model;
[0201] Inputting the first portrait information and the second portrait information into a second sub-model of the information recommendation model, and outputting a fifth probability that each promotion object belongs to a potential transaction object through the second sub-model;
[0202] Inputting the first portrait information and the second portrait information into a third sub-model of the information recommendation model, and outputting a sixth probability of each promotion object being converted into an actual transaction object through the third sub-model;
[0203] The probability of transaction completion between each promotion object and the business service platform is determined according to the product of the fourth probability, the fifth probability and the sixth probability.
[0204] In an exemplary embodiment of the present disclosure, the information recommendation model to be trained includes a first sub-model to be trained, a second sub-model to be trained, and a third sub-model to be trained;
[0205] The information recommendation device 1000 further includes a model training module 1040 (not shown in the figure), and the model training module 1040 is configured to:
[0206] Obtain a training sample set; the training sample set includes multiple training samples and labels corresponding to each of the training samples; each of the training samples includes first sample portrait information of each sample business promoter and second sample portrait information of each historical promotion object associated with each sample business promoter;
[0207] Input the plurality of training samples into the first sub-model to be trained to obtain a first prediction result for each of the training samples, and determine a first loss value according to the degree of difference between the plurality of first prediction results and the plurality of labels; the first prediction result is used to characterize the probability that the historical promotion object in each of the training samples has a preliminary transaction intention;
[0208] Input the plurality of training samples into the second sub-model to be trained to obtain a second prediction result for each of the training samples, determine a plurality of first comprehensive prediction results according to the plurality of first prediction results and the plurality of second prediction results, and determine a second loss value according to the degree of difference between the plurality of first comprehensive prediction results and the plurality of labels; the second prediction result is used to characterize the probability that the historical promotion object in each of the training samples belongs to a potential transaction object;
[0209] Input the plurality of training samples into the third sub-model to be trained to obtain a third prediction result for each of the training samples, determine a plurality of second comprehensive prediction results according to the plurality of the first comprehensive prediction results and the plurality of the third prediction results, and determine a third loss value according to the degree of difference between the plurality of the second comprehensive prediction results and the plurality of the labels; the third prediction result is used to characterize the probability of the historical promotion object in each of the training samples being converted into an actual transaction object;
[0210] Determine an overall loss value according to the first loss value, the second loss value and the third loss value;
[0211] When the overall loss value satisfies a preset first convergence condition, the trained information recommendation model is obtained.
[0212] In an exemplary embodiment of the present disclosure, the label corresponding to each of the training samples includes a first label, a second label and a third label; the first label is used to characterize the probability that each of the historical promotion objects has a preliminary transaction intention, the second label is used to characterize the probability that each of the historical promotion objects is a potential transaction object, and the third label is used to characterize the probability that each of the historical promotion objects is converted into an actual transaction object.
[0213] In an exemplary embodiment of the present disclosure, the model training module 1040 determines a first loss value according to the degree of difference between the plurality of first prediction results and the plurality of labels, including:
[0214] Obtaining a difference degree between each of the first prediction results and each of the first labels to obtain multiple difference degrees;
[0215] The first loss value is determined according to the accumulated values of the multiple difference degrees.
[0216] In an exemplary embodiment of the present disclosure, determining the second loss value according to the degree of difference between the plurality of first comprehensive prediction results and the plurality of labels includes:
[0217] Determine each first comprehensive label according to each first label and each second label;
[0218] Obtaining a difference degree between each of the first comprehensive prediction results and each of the first comprehensive labels to obtain a plurality of difference degrees;
[0219] The second loss value is determined according to the multiple difference degrees.
[0220] In an exemplary embodiment of the present disclosure, the model training module 1040 determines the third loss value according to the degree of difference between the plurality of the second comprehensive prediction results and the plurality of the labels, including:
[0221] Determine each second comprehensive label according to each of the first labels, each of the second labels and each of the third labels;
[0222] Obtaining a difference degree between each of the second comprehensive prediction results and each of the second comprehensive labels to obtain a plurality of difference degrees;
[0223] The third loss value is determined according to the multiple difference degrees.
[0224] In an exemplary embodiment of the present disclosure, the multiple historical promotion objects associated with each sample business promoter include positive sample objects and negative sample objects, the positive sample objects are promotion objects that have preliminary transaction intentions and / or are converted into actual transaction objects, and the negative sample objects are promotion objects that do not have preliminary transaction intentions and have not been converted into actual transaction objects.
[0225] The specific details of each module in the above information recommendation device have been described in detail in the corresponding information recommendation method, so they will not be repeated here.
[0226] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.
[0227] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.
[0228] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.
[0229] The present application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device.
[0230] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.
[0231] Computer-readable storage media can send, propagate or transmit programs for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.
[0232] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiments.
[0233] In addition, an electronic device capable of implementing the above method is also provided in an embodiment of the present disclosure.
[0234] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".
[0235] Refer to the following Fig.11 1100 according to this embodiment of the present disclosure is described. Fig.11 The electronic device 1100 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.
[0236] like Fig.11 As shown, the electronic device 1100 is in the form of a general computing device. The components of the electronic device 1100 may include, but are not limited to: the at least one processing unit 1110, the at least one storage unit 1120, a bus 1130 connecting different system components (including the storage unit 1120 and the processing unit 1110), and a display unit 1140.
[0237] The storage unit stores a program code, which can be executed by the processing unit 1110, so that the processing unit 1110 performs the steps described in the "Exemplary Method" section of the present specification according to various exemplary embodiments of the present disclosure. For example, the processing unit 1110 can perform the following steps: Figure 1As shown in: Step S110, obtaining the first portrait information of each business promotion personnel of the business service platform and the second portrait information of each promotion object among the multiple promotion objects associated with each business promotion personnel; Step S120, using the first portrait information and the second portrait information to predict the probability of transaction completion between each promotion object and the business service platform; Step S130, screening the target promotion object from the multiple promotion objects according to the transaction completion probability, and recommending the information of the target promotion object to the business promotion personnel.
[0238] The storage unit 1120 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 11201 and / or a cache storage unit 11202 , and may further include a read-only storage unit (ROM) 11203 .
[0239] The storage unit 1120 may also include a program / utility 11204 having a set (at least one) of program modules 11205, such program modules 11205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.
[0240] Bus 1130 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.
[0241] The electronic device 1100 may also communicate with one or more external devices 1200 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1100, and / or communicate with any device that enables the electronic device 1100 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1150. Furthermore, the electronic device 1100 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1160. As shown, the network adapter 1160 communicates with other modules of the electronic device 1100 via a bus 1130. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1100, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.
[0242] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and examples are to be considered as exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.
Claims
1. An information recommendation method, characterized in that: include: Acquire first portrait information of a business promoter of a business service platform and second portrait information of each promotion object among a plurality of promotion objects associated with the business promoter; Predicting the probability of a transaction between each promotion object and the business service platform using the first portrait information and the second portrait information; A target promotion object is screened from the multiple promotion objects according to the multiple transaction achievement probabilities, and information of the target promotion object is recommended to the business promotion personnel.
2. The method according to claim 1, characterized in that: The using the first portrait information and the second portrait information to predict the probability of a transaction between each promotion object and the business service platform includes: The first portrait information and the second portrait information are input into a trained information recommendation model, and the probability of transaction completion between each promotion object and the business service platform is output through the trained information recommendation model.
3. The method according to claim 2, characterized in that The trained information recommendation model includes a first sub-model; The step of inputting the first portrait information and the second portrait information into a trained information recommendation model, and outputting the transaction completion probability between each promotion object and the business service platform through the trained information recommendation model, comprises: Counting a first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions; If the first number is less than a first preset threshold, inputting the first portrait information and the second portrait information into a first sub-model of the information recommendation model, and outputting a first probability that each promotion object has a preliminary transaction intention through the first sub-model; The first probability is determined as a probability of transaction completion between each promotion object and the business service platform.
4. The method according to claim 3, characterized in that The trained information recommendation model also includes a second sub-model; After counting the first number of promotion objects associated with the business promotion personnel and having preliminary transaction intentions, the method further includes: If the first number is greater than or equal to the first preset threshold, counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects; If the second number is less than a second preset threshold, inputting the first portrait information and the second portrait information into a first sub-model of the information recommendation model, and outputting a second probability that each promotion object has a preliminary transaction intention through the first sub-model; Inputting the first portrait information and the second portrait information into the second sub-model of the information recommendation model, and outputting a third probability that each promotion object belongs to a potential transaction object through the second sub-model; The probability of transaction completion between each promotion object and the business service platform is determined according to the product of the second probability and the third probability.
5. The method according to claim 4, characterized in that The trained information recommendation model also includes a third sub-model; After counting the second number of promotion objects associated with the business promotion personnel and being potential transaction objects, the method further includes: If the second number is greater than or equal to the second preset threshold, inputting the first portrait information and the second portrait information into the first sub-model of the information recommendation model, and outputting a fourth probability that each promotion object has a preliminary transaction intention through the first sub-model; Inputting the first portrait information and the second portrait information into a second sub-model of the information recommendation model, and outputting a fifth probability that each promotion object belongs to a potential transaction object through the second sub-model; Inputting the first portrait information and the second portrait information into a third sub-model of the information recommendation model, and outputting a sixth probability of each promotion object being converted into an actual transaction object through the third sub-model; The probability of transaction completion between each promotion object and the business service platform is determined according to the product of the fourth probability, the fifth probability and the sixth probability.
6. The method according to claim 2, characterized in that The information recommendation model to be trained includes a first sub-model to be trained, a second sub-model to be trained, and a third sub-model to be trained; The trained information recommendation model is trained in the following way: Obtain a training sample set; the training sample set includes multiple training samples and labels corresponding to each of the training samples; each of the training samples includes first sample portrait information of each sample business promoter and second sample portrait information of each historical promotion object associated with each sample business promoter; Input the plurality of training samples into the first sub-model to be trained to obtain a first prediction result for each of the training samples, and determine a first loss value according to the degree of difference between the plurality of first prediction results and the plurality of labels; the first prediction result is used to characterize the probability that the historical promotion object in each of the training samples has a preliminary transaction intention; Input the plurality of training samples into the second sub-model to be trained to obtain a second prediction result for each of the training samples, determine a plurality of first comprehensive prediction results according to the plurality of first prediction results and the plurality of second prediction results, and determine a second loss value according to the degree of difference between the plurality of first comprehensive prediction results and the plurality of labels; the second prediction result is used to characterize the probability that the historical promotion object in each of the training samples belongs to a potential transaction object; Input the plurality of training samples into the third sub-model to be trained to obtain a third prediction result for each of the training samples, determine a plurality of second comprehensive prediction results according to the plurality of the first comprehensive prediction results and the plurality of the third prediction results, and determine a third loss value according to the degree of difference between the plurality of the second comprehensive prediction results and the plurality of the labels; the third prediction result is used to characterize the probability of the historical promotion object in each of the training samples being converted into an actual transaction object; Determine an overall loss value according to the first loss value, the second loss value and the third loss value; When the overall loss value satisfies a preset first convergence condition, the trained information recommendation model is obtained.
7. The method according to claim 6, characterized in that The labels corresponding to each of the training samples include a first label, a second label and a third label; the first label is used to characterize the probability that each of the historical promotion objects has a preliminary transaction intention, the second label is used to characterize the probability that each of the historical promotion objects is a potential transaction object, and the third label is used to characterize the probability that each of the historical promotion objects is converted into an actual transaction object.
8. The method according to claim 7, characterized in that The determining a first loss value according to the degree of difference between the plurality of first prediction results and the plurality of labels comprises: Obtaining a difference degree between each of the first prediction results and each of the first labels to obtain multiple difference degrees; The first loss value is determined according to the accumulated values of the multiple difference degrees.
9. The method according to claim 7, characterized in that: The determining of the second loss value according to the degree of difference between the plurality of the first comprehensive prediction results and the plurality of the labels comprises: Determine each first comprehensive label according to each first label and each second label; Obtaining a difference degree between each of the first comprehensive prediction results and each of the first comprehensive labels to obtain a plurality of difference degrees; The second loss value is determined according to the multiple difference degrees.
10. The method according to claim 7, characterized in that The determining of the third loss value according to the degree of difference between the plurality of the second comprehensive prediction results and the plurality of the labels comprises: Determine each second comprehensive label according to each of the first labels, each of the second labels and each of the third labels; Obtaining a difference degree between each of the second comprehensive prediction results and each of the second comprehensive labels to obtain a plurality of difference degrees; The third loss value is determined according to the multiple difference degrees.
11. The method according to any one of claims 6 to 10, characterized in that: The multiple historical promotion objects associated with each sample business promoter include positive sample objects and negative sample objects. The positive sample objects are promotion objects that have preliminary transaction intentions and / or are converted into actual transaction objects, and the negative sample objects are promotion objects that do not have preliminary transaction intentions and have not been converted into actual transaction objects.
12. An information recommendation device, characterized in that: include: An information acquisition module, used to acquire first portrait information of a business promoter of a business service platform and second portrait information of each promotion object among a plurality of promotion objects associated with the business promoter; A prediction module, used to predict the probability of a transaction between each promotion object and the business service platform by using the first portrait information and the second portrait information; The information recommendation module is used to select a target promotion object from the multiple promotion objects according to the multiple transaction completion probabilities, and recommend the information of the target promotion object to the business promotion personnel.
13. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the information recommendation method according to any one of claims 1 to 11 is implemented.
14. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the information recommendation method according to any one of claims 1 to 11 by executing the executable instructions.
