Method and apparatus for generating information

By extracting feature information from product information and using parameter information prediction models and life cycle models, an accurate life cycle curve is generated, and the prediction problem that relies on subjective experience in the existing technology is solved, which improves the scientificity and accuracy of life cycle prediction. , provides a better basis for the procurement and promotion decisions of retail enterprises.

CN110197382BActive Publication Date: 2024-11-26BEIJING JINGDONG SHANGKE INFORMATION TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201810157415.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2018-02-24
Publication Date
2024-11-26
Estimated Expiration
2038-02-24

AI Technical Summary

Technical Problem

In the prior art, retail enterprises rely on the subjective experience of sales personnel when predicting the life cycle of new products, resulting in inaccurate predictions and lack of scientificity and reliability.

Method used

By extracting feature information from the product information of the target product, and using pre-trained parameter information prediction model and life cycle model, the life cycle curve is automatically generated, and combined with machine learning methods and Bass diffusion model, the accurate life cycle curve is fitted to form a .

Benefits of technology

It improves the accuracy of the life cycle curve, provides more scientific and reliable product life cycle prediction, and helps retail enterprises make more reasonable decisions in procurement and promotion.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN110197382B_ABST
    Figure CN110197382B_ABST
Patent Text Reader

Abstract

The present application discloses a method and device for generating information. A specific implementation of the method includes: extracting feature information from product information of a target product; inputting the feature information into a pre-trained parameter information prediction model to obtain parameter information of a life cycle curve of the target product; based on the obtained parameter information, determining the value of the parameter of a pre-established life cycle model to generate a life cycle curve of the target product. This implementation improves the accuracy of the generated life cycle curve.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and in particular to a method and device for generating information. Background Art

[0002] For retail enterprises, the reasonable and effective management of categories is a technical means to improve their market position and gain competitive advantages. Usually, it is necessary to determine the products to be purchased, the frequency of purchases, the quantity of each purchase, the promotion time and other information. Therefore, it is particularly important for the development of retail enterprises to effectively predict the life cycle of new products.

[0003] Existing methods usually predict the market performance of new products based on the subjective experience of buyers and sales personnel, and then artificially determine their life cycle. Summary of the invention

[0004] The embodiments of the present application provide a method and device for generating information.

[0005] In a first aspect, an embodiment of the present application provides a method for generating information, the method comprising: extracting feature information from product information of a target product; inputting the feature information into a pre-trained parameter information prediction model to obtain parameter information of a life cycle curve of the target product, wherein the parameter information prediction model is used to characterize the correspondence between the feature information and the parameter information of the life cycle curve; based on the obtained parameter information, determining the values ​​of the parameters of a pre-established life cycle model to generate a life cycle curve of the target product, wherein the life cycle model is used to characterize the life cycle curve of the product.

[0006] In some embodiments, the parameter information prediction model is trained through the following steps: extracting pre-generated training samples, wherein the training samples include product information of multiple specified products and parameter information of the life cycle curve; extracting feature information from the product information of each specified product; using a machine learning method, taking the feature information extracted from the product information of each specified product as input, and taking the parameter information of the life cycle curve of the specified product as output, to train a parameter information prediction model.

[0007] In some embodiments, the life cycle model is established through the following steps: establishing a first function for characterizing the change of the value of the target attribute of the product over time; establishing a second function for characterizing the relationship between the value of the target attribute of the product and the demand elasticity; and establishing the life cycle model based on the first function, the second function and the Bass diffusion model.

[0008] In some embodiments, training samples are generated by the following steps: extracting operation data of multiple products, wherein the operation data of each of the multiple products includes the value of the target attribute of the product, the time corresponding to the value of the target attribute, and the sales value of the product; selecting operation data whose value of the target attribute is within the target range from the extracted operation data, and performing shrinkage and smoothing processing on the sales value in the selected operation data; for each product involved in the selected operation data, fitting a first function based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product; dividing the operation data of the product according to time intervals; substituting the values ​​of the parameters of the fitted first function into the life cycle model, and substituting the sales value of the product in each divided time interval into the life cycle model respectively, to fit the life cycle curve of the product; verifying the fitted life cycle curve, taking the products that pass the verification as designated products, summarizing the values ​​of the parameters of the life cycle curve of each designated product into the parameter information of the life cycle curve of the designated product, and summarizing the parameter information of the life cycle curve of each designated product and the product information of each designated product into training samples.

[0009] In some embodiments, the fitted life cycle curve is verified, including: for each fitted life cycle curve, determining at least one target time point corresponding to the life cycle curve; determining the sales value corresponding to each target time point in the at least one target time point; determining the amount of operational data used to fit the life cycle curve; and respectively verifying the size relationship of the target time points, the size relationship and the quantity of the sales value corresponding to the target time points.

[0010] In some embodiments, verifying the fitted life cycle curve includes: for each fitted life cycle curve, verifying the goodness of fit and the mean absolute error of the life cycle curve.

[0011] In a second aspect, an embodiment of the present application provides a device for generating information, the device comprising: a first extraction unit, configured to extract feature information from product information of a target product; an input unit, configured to input the feature information into a pre-trained parameter information prediction model to obtain parameter information of a life cycle curve of the target product, wherein the parameter information prediction model is used to characterize the correspondence between the feature information and the parameter information of the life cycle curve; a generation unit, configured to determine the values ​​of the parameters of a pre-established life cycle model based on the obtained parameter information, and generate a life cycle curve of the target product, wherein the life cycle model is used to characterize the life cycle curve of the product.

[0012] In some embodiments, the device also includes: a second extraction unit, configured to extract pre-generated training samples, wherein the training samples include product information of multiple specified products and parameter information of the life cycle curve; a third extraction unit, configured to extract feature information from the product information of each specified product; and a training unit, configured to use a machine learning method, taking the feature information extracted from the product information of each specified product as input and the parameter information of the life cycle curve of the specified product as output, to train a parameter information prediction model.

[0013] In some embodiments, the device also includes: a first establishing unit, configured to establish a first function for characterizing the change of the value of the target attribute of the product over time; a second establishing unit, configured to establish a second function for characterizing the relationship between the value of the target attribute of the product and the demand elasticity; and a third establishing unit, configured to establish a life cycle model based on the first function, the second function and the Bass diffusion model.

[0014] In some embodiments, the device also includes: a fourth extraction unit, configured to extract operation data of multiple products, wherein the operation data of each of the multiple products includes the value of the target attribute of the product, the time corresponding to the value of the target attribute, and the sales value of the product; a selection unit, configured to select operation data whose value of the target attribute is within the target range from the extracted operation data, and perform tail-folding and smoothing processing on the sales value in the selected operation data; a fitting unit, configured to fit a first function for each product involved in the selected operation data based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product; divide the operation data of the product according to time intervals; substitute the values ​​of the parameters of the fitted first function into the life cycle model, substitute the sales value of the product in each divided time interval into the life cycle model respectively, and fit the life cycle curve of the product; a summary unit, configured to verify the fitted life cycle curve, use the products that pass the verification as designated products, summarize the values ​​of the parameters of the life cycle curve of each designated product into the parameter information of the life cycle curve of the designated product, and summarize the parameter information of the life cycle curve of each designated product and the product information of each designated product as training samples.

[0015] In some embodiments, the summary unit is further configured to: for each fitted life cycle curve, determine at least one target time point corresponding to the life cycle curve; determine the sales value corresponding to each target time point in at least one target time point; determine the amount of operational data used to fit the life cycle curve; and respectively verify the size relationship of the target time points, the size relationship and the quantity of the sales value corresponding to the target time points.

[0016] In some embodiments, the summarizing unit is further configured to: for each fitted life cycle curve, verify the goodness of fit and mean absolute error of the life cycle curve.

[0017] In a third aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by one or more processors, the one or more processors implement a method such as any embodiment of the method for generating information.

[0018] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method as in any embodiment of the method for generating information.

[0019] The method and device for generating information provided in the embodiments of the present application extract feature information from the product information of the target product, and then input the feature information into a pre-trained parameter information prediction model to obtain parameter information of the life cycle curve of the target product. Finally, based on the obtained parameter information, the values ​​of the parameters of the pre-established life cycle model are determined to generate the life cycle curve of the target product. Thus, the life cycle curve can be automatically generated based on the product information, the parameter information prediction model and the life cycle model of the product, which improves the accuracy of the generated life cycle curve compared to existing methods such as manually determining the life cycle based on the subjective experience of procurement and sales personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0021] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0022] Figure 2 is a flow chart of an embodiment of a method for generating information according to the present application;

[0023] Figure 3 is a flow chart of a method for generating training samples according to the present application;

[0024] Figure 4 is a schematic diagram of an application scenario of the method for generating information according to the present application;

[0025] Figure 5 is a structural schematic diagram of an embodiment of a device for generating information according to the present application;

[0026] Figure 6It is a structural diagram of a computer system suitable for implementing an electronic device of an embodiment of the present application. DETAILED DESCRIPTION

[0027] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0028] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0029] Figure 1 An exemplary system architecture 100 to which the method for generating information or the apparatus for generating information of the present application can be applied is shown.

[0030] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0031] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Terminal devices 101, 102, 103 can be installed with various communication client applications, such as product information management applications, web browser applications, shopping applications, search applications, etc.

[0032] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0033] The server 105 may be a server that provides various services, such as a backend server for managing products in an e-commerce platform. The backend server may store product information of various products. The backend server may analyze the product information of the target product (e.g., a new product that is put on sale for the first time), generate processing results (e.g., a life cycle curve), and may also feed back the processing results to the terminal devices 101, 102, 103 so that the terminal devices 101, 102, 103 display the processing results.

[0034] It should be noted that the method for generating information provided in the embodiment of the present application is generally executed by the server 105 , and accordingly, the device for generating information is generally set in the server 105 .

[0035] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0036] Continue to refer Figure 2 , shows a process 200 of an embodiment of a method for generating information according to the present application. The method for generating information comprises the following steps:

[0037] Step 201: extract feature information from product information of a target product.

[0038] In this embodiment, the method for generating information is executed on an electronic device (eg Figure 1 The server 105 shown in the figure can extract feature information from the product information of the target product. Among them, the above-mentioned target product can be a new product (for example, a product sold for the first time on a certain e-commerce platform), or it can be a product that needs to generate a life cycle curve in advance specified by a technician. The above-mentioned product information can be various information related to the above-mentioned target product, such as product name, product number, inventory, product description information, attribute information (which can include the attributes of the above-mentioned target product and the attribute values ​​of each attribute), etc. In practice, the attributes of a product can refer to the inherent properties of the product itself, which are the properties that distinguish the product from other products, such as color, brand, model, size, storage conditions, capacity, etc. It should be noted that the above-mentioned feature information can be various forms of information used to characterize the characteristics of the target product (for example, it can be characterized in the form of a vector or matrix). Among them, the characteristics of the target product can include the attribute values ​​of multiple specified attributes (such as color, model, size, capacity, etc.) of the above-mentioned target product (for example, the attribute value of the color attribute can be "red", "blue", "green", etc.). Here, the above-mentioned electronic device can use various feature information extraction methods to extract feature information from the above-mentioned product information.

[0039] In some optional implementations of the present embodiment, for each attribute, the electronic device may store codes corresponding to the attribute values ​​of the attribute. As an example, the attribute values ​​of the attribute TV resolution may be "HD", "Full HD", and "UHD", which may correspond to codes "1", "2", and "3", respectively. Since the product information of the target product may include attribute information, the electronic device may directly extract the attribute values ​​of the multiple specified attributes from the attribute information; then, the codes corresponding to the attribute values ​​of each specified attribute may be retrieved; finally, the retrieved codes may be summarized in the form of vectors or matrices to obtain the characteristic information of the target product.

[0040] In some optional implementations of this embodiment, the electronic device can extract the attribute values ​​of the above-mentioned multiple specified attributes from the above-mentioned attribute information; then, the one-hot encoding corresponding to the attribute value of each attribute can be determined; finally, each one-hot encoding can be summarized in the form of a vector or a matrix to obtain the characteristic information of the above-mentioned target product. In practice, one-hot encoding is also called one-bit effective encoding, and its method is to use N (N is a positive integer) bit state register to encode N states, each state has its own independent register bit, and at any time, only one of them is valid. For example, six states are encoded: the natural sequence code is 000, 001, 010, 011, 100, 101, then the one-hot encoding can be 000001, 000010, 000100, 001000, 010000, 100000. Generally, one-hot encoding can be used to process discrete features of text, and to a certain extent, it also plays a role in expanding features. One-hot encoding can be represented in the form of a vector. It should be noted that the above-mentioned one-hot encoding method is a well-known technology that is currently widely studied and applied, and will not be described in detail here.

[0041] In some optional implementations of the present embodiment, the electronic device may first extract the attribute values ​​of the multiple specified attributes from the attribute information, and determine the encoding (or unique hot encoding) corresponding to the attribute values ​​of each attribute extracted. If the attribute values ​​of the multiple specified attributes are incomplete, features may be extracted from the product description information of the target product to supplement them. Specifically, the product description information may be first segmented, and then the word vectors of each word after segmentation may be determined using an existing word vector generation tool (such as an open source word vector generation tool word2vec). Afterwards, each encoding and word vector may be summarized in the form of a vector or matrix to obtain the feature information of the target product.

[0042] Step 202: input the feature information into a pre-trained parameter information prediction model to obtain parameter information of the life cycle curve of the target product.

[0043] In this embodiment, the electronic device may input the feature information into a pre-trained parameter information prediction model to obtain the parameter information of the life cycle curve of the target product. The parameter information prediction model is used to characterize the correspondence between the feature information and the parameter information of the life cycle curve. As an example, the parameter information prediction model may be a correspondence table between the feature information and the parameter information of the life cycle curve that is pre-formulated by a technician based on a large amount of data statistics. It should be noted that the life cycle curve may be a curve used to characterize the product life cycle, which may be used to characterize the changes in the sales of the product over time. In practice, the product life cycle may be the entire process from the conception of the product (here, it may refer to the time when the product starts to be sold) to the disappearance of the product (here, it may refer to the time when the product stops being sold). The product life cycle can be divided into four main stages, namely, the introduction stage, the growth stage, the maturity stage and the decline stage. It should be noted that the parameter information of the life cycle curve may be a vector composed of the parameter values ​​of each parameter in the expression of the life cycle curve.

[0044] In some optional implementations of this embodiment, the parameter information prediction model can be trained by machine learning. Specifically, it can be trained by the following steps:

[0045] The first step is to extract pre-generated training samples, wherein the training samples may include product information of multiple specified products and parameter information of the life cycle curve. In practice, the parameter information of the life cycle curve may be predetermined based on the actual sales of the specified products over time.

[0046] The second step is to extract characteristic information from the product information of each designated product. Here, the method of extracting characteristic information can be the same as the method of extracting characteristic information in step 201, which will not be repeated here.

[0047] The third step is to use a machine learning method to take the feature information extracted from the product information of each specified product as input, and the parameter information of the life cycle curve of the specified product as output, and train to obtain a parameter information prediction model. Here, the model used to train the above-mentioned parameter information prediction model can be a gradient boosting regression tree (GradientBoosting Regression Tree, GBRT). In practice, the gradient boosting regression tree can also be a gradient boosting decision tree (Gradient Boost Decision Tree, GBDT), a multiple additive regression tree (Multiple Additive Regression Tree, MART) or a decision tree network (Tree Net). The gradient boosting regression tree is an iterative decision tree algorithm, which consists of multiple decision trees. The conclusions of all trees are accumulated to obtain the final result. The tree in the gradient boosting regression tree is a regression tree, and the gradient boosting regression tree is usually used for regression prediction. Therefore, the gradient boosting regression tree can be used for machine learning, and the trained gradient boosting regression tree is used as a parameter information prediction model. It should be noted that the gradient boosting regression tree is a well-known technology that is currently widely studied and applied, and will not be repeated here.

[0048] In some optional implementations of this embodiment, the step of generating the above training samples can refer to Figure 3 . Figure 3 A flow chart of a method for generating a training sample according to the present application is given. The method 300 for generating a training sample may include the following steps:

[0049] Step 301, extracting operation data of multiple products.

[0050] Here, the operational data of each of the above-mentioned multiple products may include the value of the target attribute of the product (such as price), the time corresponding to the value of the above-mentioned target attribute, and the sales value of the product. It should be noted that there may be multiple operational data for each product, which are divided by day. The daily operational data of the product may include the value of the target attribute of the product, the time corresponding to the value of the target attribute (such as date), and the sales value of the product. In practice, the operational data may be stored locally on the above-mentioned electronic device, or may be stored in other devices connected to the above-mentioned electronic device, which is not limited here.

[0051] Step 302: Select the operating data whose target attribute value is within the target range from the extracted operating data, and perform tail-folding and smoothing processing on the sales value in the selected operating data.

[0052] Here, the above-mentioned target range can be a numerical range pre-specified by the technician, and different products can specify different numerical ranges. It should be noted that various tailing processing methods can be used to perform tailing processing on the sales values ​​in the selected operating data. For example, for each of the above-mentioned multiple products, the sales value in the operating data of the product that is less than a specified value can be updated to the specified value; the sales value in the operating data of the product that is greater than a specified value can also be updated to the specified value. It should be pointed out that various smoothing methods can be used to smooth the sales values ​​in the selected operating data, such as first-order exponential smoothing, second-order exponential smoothing, etc.

[0053] Optionally, the above target range can also be a numerical range determined according to the following steps: first, for each of the above multiple products, the target median (e.g., three-quarter median) of the value of the target attribute in the operating data of the product can be determined; then, the product of the above target median and a first preset value (e.g., 0.3) can be used as the first value, and the product of the above target median and a second preset value (e.g., 2) can be used as the second value; finally, the numerical range greater than the above first value and not greater than the above second value can be used as the target range. In practice, quantiles are also called quantiles, which refer to numerical points that divide the probability distribution range of a random variable into several equal parts. The above three-quarters point can be the value at the 75% position after the values ​​of the target attributes in the operating data of the product are arranged in order from small to large. It should be noted that quantiles are a well-known technology that is currently widely studied and applied, and will not be repeated here.

[0054] Optionally, the above-mentioned tail-contraction processing can be performed according to the following steps: first, for each of the above-mentioned multiple products, the first target quantile (for example, 0.01 quantile) and the second target quantile (for example, 0.98 quantile) of the sales value in the operating data of the product can be determined; then, the sales value in the operating data of the product that is not greater than the above-mentioned first target quantile can be updated to the above-mentioned first target quantile, and the sales value in the operating data of the product that is greater than the above-mentioned second target quantile can be updated to the above-mentioned second target quantile, thereby realizing the tail-contraction processing of the sales value in the operating data.

[0055] Optionally, the above-mentioned smoothing process can be performed according to the following steps: First, for each of the above-mentioned multiple products, the sales value in the operating data of the product after the tailing process is arranged according to the operating data (n represents the number of operating data, that is, the number of sales values) in the corresponding time sequence to form a sales value sequence. Here, S[i] can be used to represent the i-th sales value in the above-mentioned sales value sequence, and S'[i] can be used to represent the i-th sales value after smoothing, where i is an integer not less than 1 and not greater than n. Then, the sales value S[i] in the above-mentioned sales value sequence is extracted in turn, and it is determined whether i is equal to 1. In response to determining that i is equal to 1, S'[i]=S[i]; in response to determining that i is not equal to 1, S[i] is smoothed according to the following formula:

[0056] S'[i]=a×S[i]+(1-a)×S'[i-1]

[0057] Thus, the sales value of the product after smoothing is obtained, where a is a smoothing coefficient (eg, 0.5) preset by the technician.

[0058] Step 303, for each product involved in the selected operation data, fit the first function based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product; divide the operation data of the product according to time intervals; substitute the values ​​of the parameters of the fitted first function into the life cycle model, and substitute the sales value of the product in each divided time interval into the life cycle model respectively, to fit the life cycle curve of the product.

[0059] It should be noted that the above life cycle model can be pre-established through the following steps:

[0060] The first step is to establish a first function (represented by pr(t)) for characterizing the value of the target attribute of the product (represented by pr) over time. The first function is as follows:

[0061]

[0062] in, η and b are the parameters to be fitted, t is the time, and e is the natural constant.

[0063] The second step is to establish a second function (expressed by g(pr)) for characterizing the relationship between the value of the above target attribute of the product and the demand elasticity. The above second function is as follows:

[0064] g(pr)=pr E

[0065] Among them, E is the demand elasticity, which is the parameter to be fitted. In practice, demand elasticity refers to the degree to which the relative change in the demand for a product over a certain period of time responds to the relative change in the value of the target attribute of the product (such as price).

[0066] The third step is to establish a life cycle model based on the first function, the second function and the Bass diffusion model. Here, the first function and the second function can be combined with the existing Bass diffusion model to obtain an improved Bass diffusion model, which is used as a life cycle model for characterizing the life cycle curve of the product.

[0067] In practice, the Bass Diffusion Model is an existing model used to predict product sales, so the Bass Model can be used to fit the product life cycle curve. Here, since the first function and the second function are both functions related to the target attribute, the first function and the second function are combined with the Bass Diffusion Model to obtain a life cycle model, which can take the target attribute into consideration as a factor affecting sales volume, further improving the accuracy of the fitted life cycle curve.

[0068] Here, the basic form of the existing Bass diffusion model is as follows:

[0069]

[0070] Where m is the market capacity (also known as the maximum market potential and the total number of adopters, where adopters can refer to sales). p is the innovation diffusion coefficient (also known as the external influence coefficient), and q is the imitation diffusion coefficient (also known as the internal influence coefficient). m, p, and q are all parameters of the existing Bass diffusion model, which are common knowledge and will not be described here. N(t) is the cumulative number of adopters at time t, which can be used here to represent the cumulative sales at time t.

[0071] The life cycle model obtained by combining the first and second functions with the Bass diffusion model is as follows:

[0072]

[0073] It should be noted that, since the solution of the life cycle model is complex, the first function and the second function can be directly combined with the solution of the existing Bass diffusion model to obtain the following solution:

[0074]

[0075]

[0076] in, F(t) represents the cumulative proportion of adopters at time t. f(t) represents the probability density function of the proportion of the number of adopters at time t to the total number of adopters.

[0077] Here, f(t) can be used as an expression of the life cycle curve, that is, the life cycle model.

[0078] Continuing with step 303, for each product involved in the selected operation data, a life cycle curve of the product is fitted, and specifically the following steps may be performed:

[0079] The first step is to fit the first function based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product. Specifically, the first function established in advance can be extracted first. Then, the initial value of the preset parameter (for example, The initial value of is 100, the initial value of η is 0.1, the initial value of b is max(pr), where max(pr) represents the maximum value of the target attribute of the product) and the preset parameter value range (for example, The value range of is [0, max(pr)], the value range of η is [0, +∞], and the value range of b is [0, max(pr)]). Finally, the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product can be substituted into the above first function, and the first function corresponding to the product can be fitted according to the initial value and value range of the extracted parameters using the trust region algorithm, that is, the parameter value of the first function corresponding to the product can be obtained ( In practice, the trust region algorithm is a method for solving nonlinear optimization problems. It is an iterative algorithm, that is, starting from a given initial solution, it continuously improves through step-by-step iteration until a satisfactory near-optimal solution is obtained.

[0080] The second step is to divide the operation data of the product according to the time interval. For example, the operation data of the product can be divided by day, week or month. It should be noted that if the unit is day, the value of the target attribute and the sales value in the original operation data can be retained, and each day corresponds to one operation data. If the unit is week, the value of the target attribute of each day of the week can be summed up and the average value can be calculated, and the average value of the value of the target attribute can be used as the value of the target attribute in the operation data corresponding to the week; at the same time, the sales value of each day of the week can be summed up, and the summed sales value can be used as the sales value in the operation data corresponding to the week. Thus, each week corresponds to an operation data after data statistics. If the unit is month, the value of the target attribute of each day of each month can be summed up and the average value can be calculated, and the average value of the target attribute can be used as the value of the target attribute in the operation data corresponding to the month; at the same time, the sales value of each day of each month can be summed up, and the summed sales value can be used as the sales value in the operation data corresponding to the month. Thus, each month corresponds to an operation data after data statistics. Therefore, each time interval divided according to the time interval corresponds to an operation data.

[0081] The third step is to set the value of the parameter of the first function to be fitted ( The electronic device can be fitted as follows: First, the preset initial values ​​of the parameters of the above life cycle model can be extracted. For example, the initial value of m is max(susales), where max(susales) represents the total sales volume of the product, that is, the sum of the sales volumes in each operating data after the time interval division. The initial value of p is 0.03, the initial value of q is 0.38, and the initial value of E is 0.1). At the same time, the preset value ranges of the parameters of the above life cycle model can be extracted. For example, the value range of m is [0, +∞], the value range of p is [0, 2], the value range of q is [0, 2], and the value range of E is [-100, +100]. Afterwards, the values ​​of the parameters of the first function being fitted ( The values ​​of η, b) are substituted into the above life cycle model, and the sales volume values ​​of the product in each divided time interval are substituted into the pre-established life cycle model respectively. According to the initial values ​​and value ranges of the parameters of the life cycle model extracted, the trust region method is used to fit the values ​​of the parameters of the life cycle model of the product (that is, the values ​​of the parameters of the expression of the life cycle curve of the product), that is, the life cycle curve of the product is fitted.

[0082] Step 304, verify the fitted life cycle curve, take the products that pass the verification as designated products, summarize the values ​​of the parameters of the life cycle curve of each designated product into the parameter information of the life cycle curve of the designated product, and summarize the parameter information of the life cycle curve of each designated product and the product information of each designated product as training samples.

[0083] For each product involved in the selected operation data, the corresponding life cycle curve can be obtained through the above step 303. Here, various verification methods can be used to verify the fitted life cycle curves. For example, the size relationship of the sales value corresponding to multiple time points in the life cycle curve can be verified, the size relationship of multiple time points in the life cycle curve can be verified, and the number of operation data used for fitting the life cycle curve can be verified.

[0084] Then, the products that have passed the verification can be used as designated products, and the values ​​of the parameters of the life cycle curve of each designated product (including m, p, q, E, The parameter information of the life cycle curve of each specified product and the product information of each specified product (the product information of each product can be pre-stored in the above-mentioned electronic device or in a database, which is not limited here) are summarized as training samples.

[0085] Optionally, for each fitted life cycle curve, the following steps can be used for verification: Step 1, determine at least one target time point corresponding to the above life cycle curve. Specifically, the first-order derivative of the life cycle curve can be set to 0, and the time point t1 corresponding to the vertex of the life cycle curve can be obtained; then, the second-order derivative of the life cycle curve can be set to 0, and the time point t2 and the time point t3 can be obtained, where t2 < t3; then, 0.5 × (t1 + t2) is used as the time point t4, and 0.5 × (t1 + t3) is used as the time point t5; finally, determine the time point t0 corresponding to the starting point of the life cycle curve and the time point t end The second step is to determine the sales value corresponding to each target time point in the at least one target time point. end The corresponding sales values ​​are Q1, Q2, Q3, Q4, Q5, Q0, Q end. The third step is to determine the amount of operational data used to fit the life cycle curve. The fourth step is to verify the size relationship of the target time point, the size relationship of the sales value corresponding to the target time point, and the above quantity. Here, it can be determined that the life cycle curve passes the verification when the following conditions are met at the same time: t0<t2<t1<t3<t end 、Q1>Q5>Q3>Q end , Q1>Q4>Q0, the above number is greater than a preset value (for example, 10). It should be noted that determining whether the life cycle curve passes the verification may also include other judgment conditions, in addition to the above, such as whether the value range of the time point is within the preset value range, etc., which will not be repeated here.

[0086] Optionally, for each fitted life cycle curve, the goodness of fit and mean absolute deviation of the life cycle curve can also be verified. For example, the verification is determined to be passed when the goodness of fit of the life cycle curve is greater than a specified value (such as 0) and the mean absolute deviation is less than a specified value (such as 0.8). In practice, goodness of fit refers to the degree of fit of the regression line to the observed values. The statistic that measures the goodness of fit is the coefficient of determination (also known as the coefficient of determination). The maximum value of the coefficient of determination is 1. The closer the value of the coefficient of determination is to 1, the better the degree of fit of the regression line to the observed values; conversely, the smaller the value of the coefficient of determination, the worse the degree of fit of the regression line to the observed values. The mean absolute error, also called the mean absolute deviation, is the average of the absolute values ​​of the deviations of all individual observations from the arithmetic mean.

[0087] Optionally, for each fitted life cycle curve, the expression of the life cycle curve can be integrated to obtain a probability distribution function (i.e., the above-mentioned F(t)). Then, the goodness of fit and mean absolute error of the life cycle curve can be verified, and the goodness of fit and mean absolute error of the probability distribution function can be verified. For example, it is determined that the life cycle curve verification passes when the following conditions are met: the goodness of fit of the life cycle curve is greater than 0, the mean absolute error of the life cycle curve is less than 0.8, the goodness of fit of the probability distribution function is greater than 0.7, and the mean absolute error of the probability distribution function is less than 0.

[0088] Optionally, for each fitted life cycle curve, the size relationship of the target time points of the life cycle curve, the size relationship of the sales value corresponding to the target time points, the number of operating data used to fit the life cycle curve, the goodness of fit and mean absolute error of the life cycle curve, and the goodness of fit and mean absolute error of the probability distribution function obtained after integrating the life cycle curve can be verified. If all items are verified, it is determined that the life cycle curve has passed the verification.

[0089] Since the above training samples are composed of the parameter information of the life cycle curves of each specified product and the product information of each specified product, and the specified products are all products whose life cycle curves have been verified, the parameter information prediction model is trained using the training samples, and the parameter information generated by the trained parameter information prediction model is more effective and accurate.

[0090] Continue to see step 203.

[0091] Step 203: Based on the obtained parameter information, the values ​​of the parameters of the pre-established life cycle model are determined to generate a life cycle curve of the target product.

[0092] In this embodiment, the electronic device can determine the values ​​of the parameters of the pre-established life cycle model based on the obtained parameter information, and generate the life cycle curve of the target product, wherein the life cycle model can be used to characterize the life cycle curve of the product. For example, the life cycle model can be an existing Bass diffusion model. Here, since the parameter information can be a vector composed of the parameter values ​​of each parameter in the expression of the life cycle curve, the parameters corresponding to the numerical values ​​of different orders in the vector can be determined to obtain the values ​​of each parameter of the pre-established life cycle model.

[0093] In some optional implementations of the present embodiment, the above-mentioned life cycle model can be pre-established through the following steps: the first step is to establish a first function for characterizing the change of the value of the target attribute of the product over time. The second step is to establish a second function for characterizing the relationship between the value of the above-mentioned target attribute of the product and the demand elasticity. The third step is to establish a life cycle model based on the above-mentioned first function, the above-mentioned second function and the Bass diffusion model. Here, the above-mentioned first function and the second function can be combined with the existing Bass diffusion model to obtain an improved Bass diffusion model, and the improved Bass diffusion model is used as a life cycle model for characterizing the life cycle curve of the product. It should be noted that the method for establishing the life cycle model here has been described in detail in step 303 and will not be repeated here.

[0094] Therefore, after obtaining the life cycle curve of the target product, it is possible to determine, based on the life cycle curve, after the target product goes online, which stage of the life cycle the target product is in and what its future trends will be, etc. This provides strong assistance for accurately selecting the appropriate time points for activities such as replenishment and promotions. It can also combine the sales forecast of the target product to evaluate the market performance of the product.

[0095] Continue to see Figure 4 , Figure 4 FIG. 1 is a schematic diagram of an application scenario of the method for generating information according to this embodiment. Figure 4 In the application scenario, the backend server for managing products in the e-commerce platform first extracts feature information 402 from the product information 401 of the new product to be put on the shelf; then, the feature information 402 is input into a pre-trained parameter information prediction model to obtain parameter information 403 of the life cycle curve of the target product in the form of a vector; finally, based on the obtained parameter information 403, the values ​​of the parameters of the pre-established life cycle model are determined to generate a life cycle curve 404 of the target product.

[0096] The method provided by the above-mentioned embodiment of the present application extracts feature information from the product information of the target product, and then inputs the feature information into a pre-trained parameter information prediction model to obtain the parameter information of the life cycle curve of the target product. Finally, based on the obtained parameter information, the values ​​of the parameters of the pre-established life cycle model are determined to generate the life cycle curve of the target product. In this way, the life cycle curve can be automatically generated based on the product information, the parameter information prediction model and the life cycle model of the product, which improves the accuracy of the generated life cycle curve compared to the existing method of manually determining the life cycle based on the subjective experience of procurement and sales personnel.

[0097] Further references Figure 5 As an implementation of the methods shown in the above figures, the present application provides an embodiment of a device for generating information, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0098] like Figure 5As shown, the above-mentioned device 500 for generating information in this embodiment includes: a first extraction unit 501, configured to extract feature information from the product information of the target product; an input unit 502, configured to input the above-mentioned feature information into a pre-trained parameter information prediction model to obtain parameter information of the life cycle curve of the above-mentioned target product, wherein the above-mentioned parameter information prediction model is used to characterize the correspondence between the feature information and the parameter information of the life cycle curve; a generation unit 503, configured to determine the value of the parameter of the pre-established life cycle model based on the obtained parameter information, and generate the life cycle curve of the above-mentioned target product, wherein the above-mentioned life cycle model is used to characterize the life cycle curve of the product.

[0099] In some optional implementations of this embodiment, the device may further include a second extraction unit and a third extraction unit (not shown in the figure). The second extraction unit may be configured to extract pre-generated training samples, wherein the training samples include product information of multiple specified products and parameter information of the life cycle curve. The third extraction unit may be configured to extract feature information from the product information of each specified product; the training unit is configured to use a machine learning method, taking the feature information extracted from the product information of each specified product as input and the parameter information of the life cycle curve of the specified product as output, to train a parameter information prediction model.

[0100] In some optional implementations of this embodiment, the device may further include a first establishing unit, a second establishing unit, and a third establishing unit (not shown in the figure). The first establishing unit may be configured to establish a first function for characterizing the change of the value of the target attribute of the product over time. The second establishing unit may be configured to establish a second function for characterizing the relationship between the value of the target attribute of the product and the demand elasticity. The third establishing unit may be configured to establish a life cycle model based on the first function, the second function, and the Bass diffusion model.

[0101] In some optional implementations of this embodiment, the device may further include a fourth extraction unit, a selection unit, a fitting unit and a summary unit (not shown in the figure). The fourth extraction unit may be configured to extract the operation data of multiple products, wherein the operation data of each of the multiple products includes the value of the target attribute of the product, the time corresponding to the value of the target attribute, and the sales value of the product. The selection unit may be configured to select the operation data whose value of the target attribute is within the target range from the extracted operation data, and perform shrinkage processing and smoothing processing on the sales value in the selected operation data. The fitting unit may be configured to fit the first function for each product involved in the selected operation data based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product; divide the operation data of the product according to time intervals; substitute the values ​​of the parameters of the fitted first function into the life cycle model, and substitute the sales value of the product in each divided time interval into the life cycle model, and fit the life cycle curve of the product. The above-mentioned summary unit can be configured to verify the fitted life cycle curve, take the products that pass the verification as designated products, summarize the values ​​of the parameters of the life cycle curve of each designated product into the parameter information of the life cycle curve of the designated product, and summarize the parameter information of the life cycle curve of each designated product and the product information of each designated product as training samples.

[0102] In some optional implementations of the present embodiment, the summary unit may be further configured to determine, for each fitted life cycle curve, at least one target time point corresponding to the life cycle curve; determine the sales value corresponding to each target time point among the at least one target time point; determine the amount of operational data used to fit the life cycle curve; and verify the size relationship of the target time points, the size relationship of the sales values ​​corresponding to the target time points, and the above quantities, respectively.

[0103] In some optional implementations of this embodiment, the above-mentioned summarizing unit may be further configured to verify the goodness of fit and the mean absolute error of each fitted life cycle curve.

[0104] The device provided by the above-mentioned embodiment of the present application extracts feature information from the product information of the target product through the first extraction unit 501, and then the input unit 502 inputs the feature information into a pre-trained parameter information prediction model to obtain the parameter information of the life cycle curve of the target product. Finally, the generation unit 503 determines the values ​​of the parameters of the pre-established life cycle model based on the obtained parameter information, and generates the life cycle curve of the target product. In this way, the life cycle curve can be automatically generated based on the product information, the parameter information prediction model and the life cycle model of the product, which improves the accuracy of the generated life cycle curve compared to the existing method of manually determining the life cycle based on the subjective experience of procurement and sales personnel.

[0105] Reference below Figure 6 , which shows a schematic diagram of the structure of a computer system 600 suitable for implementing an electronic device of an embodiment of the present application. Figure 6 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0106] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0107] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0108] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above functions defined in the method of the present application are executed. It should be noted that the computer-readable medium described in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, - but not limited to - an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections with one or more conductors, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device, or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, device, or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0109] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0110] The units involved in the embodiments described in the present application may be implemented by software or hardware. The units described may also be set in a processor, for example, it may be described as: a processor includes a first extraction unit, an input unit, and a generation unit. The names of these units do not constitute a limitation on the units themselves in some cases, for example, the first extraction unit may also be described as a "unit for extracting feature information from product information of a target product".

[0111] As another aspect, the present application also provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by the device, the device: extracts feature information from the product information of the target product; inputs the feature information into a pre-trained parameter information prediction model to obtain parameter information of the life cycle curve of the target product; based on the obtained parameter information, determines the value of the parameter of the pre-established life cycle model to generate the life cycle curve of the target product.

[0112] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other.

Claims

1. A method for generating information, comprising: Extracting feature information from product information of a target product; Input the characteristic information into a pre-trained parameter information prediction model to obtain parameter information of the life cycle curve of the target product, wherein the parameter information prediction model is used to characterize the corresponding relationship between the characteristic information and the parameter information of the life cycle curve, the parameter information prediction model is trained by a machine learning method, and the model used by the parameter information prediction model is a gradient boosting regression tree; the parameter information is a vector composed of parameter values ​​of each parameter in the expression of the life cycle curve; Based on the obtained parameter information, the values ​​of the parameters of the pre-established life cycle model are determined to generate the life cycle curve of the target product, wherein the life cycle model is used to characterize the life cycle curve of the product.

2. The method for generating information according to claim 1, wherein: The parameter information prediction model is trained by the following steps: Extracting pre-generated training samples, wherein the training samples include product information of a plurality of specified products and parameter information of a life cycle curve; Extracting feature information from product information of each specified product; Using machine learning methods, the feature information extracted from the product information of each specified product is used as input, and the parameter information of the life cycle curve of the specified product is used as output to train a parameter information prediction model.

3. The method for generating information according to claim 2, wherein: The life cycle model is established through the following steps: Establishing a first function for characterizing the change in the value of a target attribute of a product over time; Establishing a second function for characterizing the relationship between the value of the target attribute of the product and the demand elasticity; A life cycle model is established based on the first function, the second function and the Bass diffusion model.

4. The method for generating information according to claim 3, wherein: The training samples are generated by the following steps: Extracting operation data of a plurality of products, wherein the operation data of each of the plurality of products includes a value of a target attribute of the product, a time corresponding to the value of the target attribute, and a sales value of the product; Selecting operation data whose target attribute value is within a target range from the extracted operation data, and performing tail-shrinking and smoothing processing on the sales value in the selected operation data; For each product involved in the selected operation data, the first function is fitted based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product; the operation data of the product is divided according to time intervals; the values ​​of the parameters of the fitted first function are substituted into the life cycle model, and the sales value of the product in each divided time interval is substituted into the life cycle model to fit the life cycle curve of the product; The fitted life cycle curve is verified, and the products that pass the verification are taken as designated products. The values ​​of the parameters of the life cycle curve of each designated product are summarized as the parameter information of the life cycle curve of the designated product. The parameter information of the life cycle curve of each designated product and the product information of each designated product are summarized as training samples.

5. The method for generating information according to claim 4, wherein: The verifying of the fitted life cycle curve includes: For each fitted life cycle curve, determine at least one target time point corresponding to the life cycle curve; determine the sales value corresponding to each target time point in the at least one target time point; determine the amount of operational data used to fit the life cycle curve; and verify the size relationship of the target time points, the size relationship of the sales value corresponding to the target time points, and the quantity, respectively.

6. The method for generating information according to claim 4 or 5, wherein: The verifying of the fitted life cycle curve includes: For each fitted life cycle curve, the goodness of fit and mean absolute error of the life cycle curve are verified.

7. A device for generating information, comprising: A first extraction unit configured to extract feature information from product information of a target product; An input unit is configured to input the feature information into a pre-trained parameter information prediction model to obtain parameter information of the life cycle curve of the target product, wherein the parameter information prediction model is used to characterize the corresponding relationship between the feature information and the parameter information of the life cycle curve, the parameter information prediction model is trained by a machine learning method, and the model used by the parameter information prediction model is a gradient boosting regression tree; the parameter information is a vector composed of parameter values ​​of each parameter in the expression of the life cycle curve; A generating unit is configured to determine the values ​​of the parameters of the pre-established life cycle model based on the obtained parameter information, and generate the life cycle curve of the target product, wherein the life cycle model is used to characterize the life cycle curve of the product.

8. The device for generating information according to claim 7, wherein: The device also includes: A second extraction unit is configured to extract pre-generated training samples, wherein the training samples include product information of a plurality of specified products and parameter information of a life cycle curve; a third extraction unit configured to extract feature information from product information of each designated product; The training unit is configured to use a machine learning method to take feature information extracted from product information of each designated product as input and parameter information of the life cycle curve of the designated product as output, and train to obtain a parameter information prediction model.

9. The device for generating information according to claim 8, wherein: The device also includes: A first establishing unit configured to establish a first function for characterizing a change in a value of a target attribute of a product over time; A second establishing unit configured to establish a second function for characterizing the relationship between the value of the target attribute of the product and the demand elasticity; The third establishing unit is configured to establish a life cycle model based on the first function, the second function and the Bass diffusion model.

10. The device for generating information according to claim 9, wherein: The device also includes: a fourth extraction unit configured to extract operation data of a plurality of products, wherein the operation data of each of the plurality of products includes a value of a target attribute of the product, a time corresponding to the value of the target attribute, and a sales value of the product; A selection unit configured to select operation data whose target attribute value is within a target range from the extracted operation data, and perform tail shrinkage processing and smoothing processing on the sales value in the selected operation data; A fitting unit is configured to fit the first function for each product involved in the selected operation data based on the value of the target attribute of the product and the time corresponding to the value of the target attribute of the product; divide the operation data of the product according to time intervals; substitute the values ​​of the parameters of the fitted first function into the life cycle model, substitute the sales value of the product in each divided time interval into the life cycle model, and fit the life cycle curve of the product; A summary unit is configured to verify the fitted life cycle curve, take the products that pass the verification as designated products, summarize the values ​​of the parameters of the life cycle curve of each designated product into the parameter information of the life cycle curve of the designated product, and summarize the parameter information of the life cycle curve of each designated product and the product information of each designated product as training samples.

11. The device for generating information according to claim 10, wherein: The aggregation unit is further configured to: For each fitted life cycle curve, determine at least one target time point corresponding to the life cycle curve; determine the sales value corresponding to each target time point in the at least one target time point; determine the amount of operational data used to fit the life cycle curve; and verify the size relationship of the target time points, the size relationship of the sales value corresponding to the target time points, and the quantity, respectively.

12. The device for generating information according to claim 10 or 11, wherein: The aggregation unit is further configured to: For each fitted life cycle curve, the goodness of fit and mean absolute error of the life cycle curve are verified.

13. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

14. A computer-readable storage medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.