WO2020048065A1 - 产品智能推荐方法、装置、计算机设备及存储介质 - Google Patents
产品智能推荐方法、装置、计算机设备及存储介质 Download PDFInfo
- Publication number
- WO2020048065A1 WO2020048065A1 PCT/CN2018/125345 CN2018125345W WO2020048065A1 WO 2020048065 A1 WO2020048065 A1 WO 2020048065A1 CN 2018125345 W CN2018125345 W CN 2018125345W WO 2020048065 A1 WO2020048065 A1 WO 2020048065A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- product
- value
- customer
- unrated
- unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Recommending goods or services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Definitions
- the present application relates to the field of computer technology, and in particular, to a product intelligent recommendation method, device, computer device, and storage medium.
- the process of calculating a customer's rating of other products based on the customer's rating of a certain product and recommending related products to the customer based on the rating result it involves predicting and scoring an unrated product.
- the currently used prediction scoring method is matrix decomposition.
- a mapping matrix is established in combination with the user's existing ratings to achieve predictive ratings for other products.
- the predictive scoring method that establishes a mapping matrix to calculate ungraded products there are problems such as a large amount of calculation and a long time-consuming scoring, which results in a longer waiting time for customers. Therefore, the existing scoring methods have a poor prediction scoring efficiency. High problem.
- the embodiments of the present application provide a product intelligent recommendation method, device, computer equipment, and storage medium, which aim to solve the problem of low prediction scoring efficiency existing in the product recommendation methods in the prior art.
- an embodiment of the present application provides a product intelligent recommendation method, which includes: obtaining a pre-stored product information table, and constructing a score prediction model by using a preset model construction rule and product information in the product information table; Obtain a pre-stored historical rating data table, train the constructed rating prediction model according to the product's rating data and preset training parameters in the historical rating data table; obtain a pre-stored historical rating data table for a customer's rating Data, according to the customer ’s rating data of the rated product and the trained rating prediction model to score the customer ’s unrated product to obtain the predicted score of the unrated product; according to preset product screening rules and prediction of the unrated product The rating screens the customer's unrated products to get the target product recommended for the customer.
- an embodiment of the present application provides a product intelligent recommendation device, which includes: a scoring model construction unit, configured to obtain a pre-stored product information table, and use a preset model construction rule and the product information table.
- Product information constructs a rating prediction model;
- a rating model training unit is configured to obtain a pre-stored historical rating data table, and train the constructed rating prediction model according to product rating data and preset training parameters in the historical rating data table.
- Predictive scoring unit used to obtain pre-stored scoring data of a customer in the historical scoring data table, and scoring the customer's unrated products according to the customer's scoring data for the rated product and the trained scoring prediction model to obtain Predicted scores of unrated products
- target product screening unit configured to filter the unrated products of the customer according to preset product screening rules and predicted ratings of the unrated products to obtain target products recommended by the customer.
- an embodiment of the present application further provides a computer device including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer
- the program implements the product intelligent recommendation method described in the first aspect.
- an embodiment of the present application further provides a computer-readable storage medium, where the storage medium stores a computer program, and when the computer program is executed by a processor, causes the processor to execute the foregoing first aspect.
- FIG. 1 is a schematic flowchart of a product intelligent recommendation method according to an embodiment of the present application
- FIG. 2 is a schematic diagram of a score prediction model of a product intelligent recommendation method according to an embodiment of the present application
- FIG. 3 is a schematic flowchart of a sub-process of a product intelligent recommendation method according to an embodiment of the present application
- FIG. 4 is a schematic diagram of another sub-process of a product intelligent recommendation method according to an embodiment of the present application.
- FIG. 5 is a schematic diagram of another sub-process of a product intelligent recommendation method according to an embodiment of the present application.
- FIG. 6 is a schematic diagram of another sub-process of a product intelligent recommendation method according to an embodiment of the present application.
- FIG. 7 is a schematic block diagram of a product intelligent recommendation device according to an embodiment of the present application.
- FIG. 8 is a schematic block diagram of a subunit of a product intelligent recommendation device according to an embodiment of the present application.
- FIG. 9 is a schematic block diagram of another subunit of a product intelligent recommendation device according to an embodiment of the present application.
- FIG. 10 is a schematic block diagram of another subunit of a product intelligent recommendation device according to an embodiment of the present application.
- FIG. 11 is a schematic block diagram of another subunit of a product intelligent recommendation device according to an embodiment of the present application.
- FIG. 12 is a schematic block diagram of a computer device according to an embodiment of the present application.
- FIG. 1 is a schematic flowchart of a product intelligent recommendation method according to an embodiment of the present application.
- the product intelligent recommendation method is applied to a management server, and the method is executed by application software installed in the management server.
- the management server is an enterprise terminal for intelligently recommending a product.
- the method includes steps S101 to S104.
- a score prediction model including a fully connected hidden layer, an input node, and an output node is constructed.
- the model construction rule is the rule information used to construct the scoring prediction model.
- the scoring prediction model is constructed based on a neural network.
- the model construction rule includes a fully connected hidden layer and a unit that configures the number of feature units in the fully connected hidden layer. Quantity configuration information.
- the fully connected hidden layer is the middle layer used to connect the input node with the output node.
- the product information table contains information on all products provided by the enterprise to consumers.
- the product information in the product information table includes, but is not limited to, the number of products and the names of the products.
- FIG. 2 is a schematic diagram of a scoring prediction model of a product intelligent recommendation method according to an embodiment of the present application. As shown in FIG. 2, the scoring prediction model includes a fully connected hidden layer, multiple input nodes, and multiple output nodes.
- step S101 includes sub-steps S1011, S1012, S1013, S1014, and S1015.
- An input node and an output node of a scoring prediction model are constructed according to product information, and the input node and the output node correspond one-to-one.
- An input node and an output node of the scoring prediction model are constructed according to the product information, and the input nodes correspond to the output nodes one-to-one. Specifically, according to the number of products in the product information, multiple input nodes with the same number of products are constructed, and multiple output nodes with the same number of products are constructed, where one product corresponds to one input node and one output node, that is, inputs
- the node contains the product name corresponding to the output node
- the output node contains the product name corresponding to the output node
- the input node and output node of the same product correspond one-to-one.
- the number of feature units in the fully connected hidden layer in the model building rule is configured according to the product information and the unit number configuration information in the model building rule.
- the fully-connected hidden layer is an intermediate layer used to connect the input node and the output node.
- the fully-connected hidden layer contains several feature units, and each feature unit is associated with all input nodes and all output nodes.
- Features Units can be used to reflect the relationship between product characteristics and customer preferences.
- the number of feature units in a fully connected hidden layer can be configured based on the unit number configuration information and the number of products in the product information. For example, in the unit number configuration information, the number of characteristic units is preset to be 2/5 of the number of products, and the number of all products of the enterprise in the product information table is 100. Then, the number of characteristic units in the fully connected hidden layer is configured to 40.
- S1013 Construct a first formula group from the input node to the feature unit according to the feature unit of the fully connected hidden layer and the constructed input node, using the input node value as the input value and the feature unit value as the output value.
- a first formula group of input node to feature unit is constructed with the input node value as the input value and the feature unit value as the output value.
- the first formula group includes formulas from all input nodes to all feature units.
- the input node is the node used to input the score of a product in the score prediction model.
- the specific value of the input node is the input node value.
- the characteristic unit value is the calculated value of the characteristic unit in the fully connected hidden layer.
- a second formula group of the feature unit to the output node is constructed with the feature unit value as an input value and the output node value as an output value.
- the second formula group includes formulas from all characteristic units to all output nodes.
- the output node is the node used to output the predicted score of a product in the rating prediction model.
- the specific value of the output node is the output node value.
- the output node value is the score value used to predict a product.
- the characteristic unit value is the calculated value of the characteristic unit in the fully connected hidden layer.
- Randomly configuring the parameters of the formulas in the first formula group and randomly configuring the parameters of the formulas in the second formula group completes the process of constructing the score prediction model.
- the score prediction model constructed is trained according to the score data of multiple customers on the product and preset training parameters. Because the company has a large number of products, customers use some of the company's products to rate the corresponding products, but because the company's other products are not used by the customer, the customer only made some of the products that the company has used score. By obtaining multiple customers 'rating data for products, you can fully cover the ratings of all products of the enterprise, and train the constructed rating prediction model through multiple customers' rating data for all products.
- the training parameters are preset parameter information for training the score prediction model.
- the training parameters include a mean square error function, a mean square error threshold, and a parameter adjustment rule.
- step S102 includes sub-steps S1021, S1022, S1023, S1024, and S1025.
- a characteristic unit in the calculation result of the corresponding formula in the first formula group includes multiple calculation values, an average value of the multiple calculation values is taken as the characteristic unit value of the characteristic unit; if the calculation of the corresponding formula in the first formula group A characteristic unit in the result contains only one calculated value, and the calculated value is taken as the characteristic unit value of the characteristic unit.
- the output node value of the corresponding output node is calculated according to the corresponding formula and characteristic unit value in the second formula group. Since the input node and output node of the same product correspond one-to-one, the corresponding formula and characteristic unit value in the second formula group can be used.
- the output node value of the same product output node corresponding to the input node is calculated, and the output node value is the score value obtained by predicting the product through the score prediction model. Because the parameters of the formula in the score prediction model have not been trained The adjustment and update are performed, so the output node value currently calculated can only be used to train the scoring prediction model, and cannot be used as a prediction scoring result to recommend related products to customers.
- the mean square error function (MSE function) is It is a calculation function used to reflect the degree of difference between the estimated quantity and the estimated quantity.
- the error value between the customer's rating of the product and the product rating calculated by the rating prediction model can be calculated by the mean square error function. The larger the square error value, the greater the error between the customer's rating and the score calculated by the rating prediction model, and the smaller the mean square error value, the smaller the error between the customer's rating and the score calculated by the rating prediction model. .
- z n is the score of the n-th rated product of the customer calculated by the rating prediction model, that is, the output node value;
- x n is the score of the n-th rated product of the customer, that is, the output node value .
- Whether the mean square error value is greater than the mean square error threshold is determined according to the mean square error threshold. The judgment result is divided into two cases, that is, the mean square error value is not greater than the mean square error threshold, and the mean square error value is greater than the mean square error threshold.
- the parameters of the formula in the score prediction model are adjusted according to the parameter adjustment rule. Specifically, if the judgment result is that the mean square error value is less than or equal to the mean square error threshold, the parameters of the formula in the score prediction model are not adjusted, and the next customer's score data for the product is obtained to iteratively train the score prediction model; If the judgment result is that the mean square error is greater than the mean square error threshold, the parameters of the corresponding formula in the first formula group and the parameters of the corresponding formula in the second formula group are adjusted and updated according to the parameter adjustment rule.
- Parameter adjustment rules include parameter adjustment direction and parameter adjustment range.
- the parameter adjustment direction is the direction information used to adjust or reduce the parameters of the formula.
- the parameter adjustment range is the amplitude information used to adjust the parameters of the formula.
- the parameter adjustment direction is enlarged and the parameter adjustment range is 2%.
- the customer's rating data for the product can be input multiple times to adjust the parameters of the formula in the rating prediction model according to parameter adjustment rules. Perform multiple adjustments and calculate multiple mean square error adjustment values. From the obtained multiple mean square error adjustment values, select a set of parameters corresponding to the smallest mean square error adjustment value to the corresponding formula in the first formula group. And update the parameters of the corresponding formula in the second formula group. After updating the parameters of the formula in the score prediction model, the next customer's score data for the product is obtained to iteratively train the score prediction model.
- the customer's rating data in the pre-stored historical rating data table, and score the customer's unrated product according to the customer's rating data for the product and the trained rating prediction model to obtain the predicted rating result, where the predicted rating
- the results include predicted scores for all unrated products for the customer, and the output node includes the product name.
- a customer scores the corresponding product after using some of the company's products, but because the customer's other products are not used by the customer, the customer only evaluates some of the products that the company has used. If you get a rating, you get the customer ’s rating data for the rated product. Other products that the customer has not rated are unrated products.
- step S103 includes sub-steps S1031, S1032 and S1033.
- a characteristic unit in the calculation result of the corresponding formula in the first formula group includes multiple calculation values, an average value of the multiple calculation values is taken as the characteristic unit value of the characteristic unit; if the calculation of the corresponding formula in the first formula group A characteristic unit in the result contains only one calculated value, and the calculated value is taken as the characteristic unit value of the characteristic unit.
- the output node values of the corresponding output nodes are calculated according to the corresponding formulas and feature unit values in the second formula group. Since the parameters of all formulas in the second formula group in the score prediction model have been adjusted and updated after training, the The output node values of all output nodes can be calculated by all formulas and characteristic unit values in the second formula group. Because the parameters of the formula in the scoring prediction model have been adjusted and updated through training, the output node value currently calculated can be used to predict the scoring product of the customer.
- the product name of the output node is matched with the unscored product, and the output node value of the output node that matches the unscored product is used as the predicted score of the unscored product.
- the predicted scores are obtained by obtaining the predicted scores of all unscored products. Since a product corresponds to an input node and an output node, the output node value of the corresponding output node can be used as the predicted score of the unrated product according to the product name of the output node and the unrated product.
- the unrated products of the customer are filtered, and the target product recommended for the customer can be obtained. Specifically, the predicted product with a higher rated rating among the unrated products of the customer can be selected. The product is recommended by the customer as the target product.
- step S104 includes sub-steps S1041 and S1042.
- the unrated products can be filtered according to the settings in the product recommendation rules, such as The product recommendation rule may be to select the top three unrated products in the prediction score as target products to recommend to the customer.
- a rating prediction model is obtained through product information and model construction rules, and the rating prediction model is trained through the historical rating data table to calculate the predicted rating of the customer's unrated product through the rating prediction model and the rating data of the customer's rated product.
- the preset product screening rules select target products with higher prediction scores from unscored products and recommend the customer, which can greatly improve the efficiency of prediction scores and the accuracy of prediction scores, and reduce the time consuming of prediction scores. , It is convenient for enterprises to select target products and recommend to customers.
- the embodiment of the present application further provides a product intelligent recommendation device, which is used to execute any one of the foregoing embodiments of the product intelligent recommendation method.
- a product intelligent recommendation device which is used to execute any one of the foregoing embodiments of the product intelligent recommendation method.
- FIG. 7 is a schematic block diagram of a product intelligent recommendation device provided by an embodiment of the present application.
- the product intelligent recommendation device 100 may be configured in a management server.
- the product intelligent recommendation device 100 includes a scoring model construction unit 101, a scoring model training unit 102, a prediction scoring unit 103, and a target product screening unit 104.
- the scoring model construction unit 101 is configured to obtain a pre-stored product information table, and construct a scoring prediction model by using a preset model construction rule and product information in the product information table.
- the scoring model construction unit 101 includes sub-units: a node construction unit 1011, a feature unit number configuration unit 1012, a first formula group construction unit 1013, and a second formula group construction unit 1014. And parameter configuration unit 1015.
- the node construction unit 1011 is configured to construct an input node and an output node of the score prediction model according to the product information, and the input node corresponds to the output node one-to-one.
- the feature unit number configuration unit 1012 is configured to configure the number of feature units in the fully connected hidden layer in the model construction rule according to the product information and the unit number configuration information in the model construction rule.
- a first formula group constructing unit 1013 is configured to construct a first input node to feature unit according to the feature unit of the fully connected hidden layer and the constructed input node, using the input node value as the input value and the feature unit value as the output value.
- a second formula group constructing unit 1014 is configured to construct a second feature unit to an output node according to the feature unit of the fully connected hidden layer and the constructed output node, using the feature unit value as the input value and the output node value as the output value.
- a parameter configuration unit 1015 is configured to randomly configure parameters of formulas in the constructed first formula group and the second formula group.
- the scoring model training unit 102 is configured to obtain a pre-stored historical scoring data table, and train the constructed scoring prediction model according to product scoring data and preset training parameters in the historical scoring data table.
- the scoring model training unit 102 includes sub-units: a first feature unit value calculation unit 1021, a first output node value calculation unit 1022, a mean square error value calculation unit 1023, and a mean The square error determination unit 1024 and the parameter adjustment unit 1025.
- the first characteristic unit value calculation unit 1021 is configured to obtain a customer's product rating data in the historical rating data table as an input node value of a corresponding formula in the first formula group, and calculate according to the corresponding formula and input node value in the first formula group. Get the feature unit values of all feature units in the fully connected hidden layer.
- the first output node value calculation unit 1022 is configured to calculate an output node value of a corresponding output node according to a corresponding formula and a characteristic unit value in the second formula group.
- a mean square error value calculation unit 1023 is configured to obtain an input node value of an input node and an output node value of an output node corresponding to the input node, and obtain the input node value and the output node value according to the mean square error function. Calculations are performed to obtain the mean square error value for the customer.
- the mean square error determining unit 1024 is configured to determine whether the mean square error value is greater than the mean square error threshold according to the mean square error threshold.
- a parameter adjustment unit 1025 is configured to adjust parameters of a formula in the score prediction model according to the parameter adjustment rule if the mean square error value is greater than a mean square error threshold.
- the prediction and scoring unit 103 is configured to obtain the scoring data of a customer in a pre-stored historical scoring data table, and score the unrated product of the customer according to the customer's scoring data of the rated product and the trained scoring prediction model to obtain Forecast score for unrated products.
- the prediction scoring unit 103 includes sub-units: a second feature unit value calculation unit 1031, a second output node value calculation unit 1032, and a product name matching unit 1033.
- the second characteristic unit value calculation unit 1031 is configured to obtain the customer's rating data of the rated product as the input node value of the corresponding formula in the first formula group, and obtain a full connection according to the corresponding formula and input node value in the first formula group.
- the feature unit values of all feature units in the hidden layer are configured to obtain the customer's rating data of the rated product as the input node value of the corresponding formula in the first formula group, and obtain a full connection according to the corresponding formula and input node value in the first formula group.
- the second output node value calculation unit 1032 is configured to calculate output node values of all output nodes according to corresponding formulas and characteristic unit values in the second formula group.
- the product name matching unit 1033 is configured to match the product name of the output node with the unrated product, and use the output node value that matches the unrated product as the predicted score of the customer's unrated product.
- the target product screening unit 104 is configured to screen an unrated product of the customer according to a preset product screening rule and a predicted score of the unrated product to obtain a target product recommended by the customer.
- the target product screening unit 104 includes sub-units: a product sorting unit 1041 and a product screening unit 1042.
- the product sorting unit 1041 is configured to sort the predicted scores of the unrated products of the customer to obtain a ranking result of the unrated products.
- the product screening unit 1042 is configured to filter the unrated products according to a preset product recommendation rule and a ranking result of the unrated products to obtain a target product recommended by the customer.
- the above product intelligent recommendation device can be implemented in the form of a computer program, which can be run on a computer device as shown in FIG. 12.
- FIG. 12 is a schematic block diagram of a computer device according to an embodiment of the present application.
- the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501.
- the memory may include a non-volatile storage medium 503 and an internal memory 504.
- the non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032.
- the processor 502 can execute the product intelligent recommendation method.
- the processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500.
- the internal memory 504 provides an environment for running the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute the product intelligent recommendation method.
- the network interface 505 is used for network communication, such as providing transmission of data information.
- the structure shown in FIG. 12 is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer equipment 500 to which the scheme of the present application is applied.
- the specific computer equipment 500 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.
- the processor 502 is configured to run a computer program 5032 stored in a memory to implement the product intelligent recommendation method of this embodiment.
- the embodiment of the computer device shown in FIG. 12 does not constitute a limitation on the specific configuration of the computer device.
- the computer device may include more or fewer components than shown in the figure. Either some parts are combined or different parts are arranged.
- the computer device may include only a memory and a processor. In such an embodiment, the structures and functions of the memory and the processor are consistent with the embodiment shown in FIG. 12, and details are not described herein again.
- the processor 502 may be a central processing unit (CPU), and the processor 502 may also be another general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), Application-specific integrated circuits (Application Specific Integrated Circuits, ASICs), ready-made programmable gate arrays (Field-Programmable Gate Arrays, FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
- the general-purpose processor may be a microprocessor, or the processor may be any conventional processor.
- a storage medium is provided in another embodiment of the present application.
- the storage medium may be a non-transitory computer-readable storage medium.
- the storage medium stores a computer program, wherein the computer program implements the product intelligent recommendation method in the embodiment of the present application when the computer program is executed by a processor.
- the storage medium may be an internal storage unit of the foregoing device, such as a hard disk or a memory of the device.
- the storage medium may also be an external storage device of the device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, and a flash memory card provided on the device. (Flash Card), etc.
- the storage medium may include both an internal storage unit of the device and an external storage device.
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
产品智能推荐方法、装置、计算机设备及存储介质。所述方法包括:获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型(S101);获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练(S102);获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分(S103);根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品(S104)。
Description
本申请要求于2018年9月5日提交中国专利局、申请号为201811029043.7、申请名称为“产品智能推荐方法、装置、计算机设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本申请涉及计算机技术领域,尤其涉及一种产品智能推荐方法、装置、计算机设备及存储介质。
在根据客户对某部分产品的评分计算客户对其他产品进行评分并根据评分结果为客户推荐相关产品的过程中,涉及对未评分的产品进行预测评分,目前所采用的预测评分方法是通过矩阵分解结合用户已有评分建立映射矩阵,以实现对其他产品进行预测评分。然而采用建立映射矩阵的预测评分方法对未评分产品进行计算时,存在计算量较大、评分耗时长的问题,导致客户所需等待的时间较长,因而现有的评分方法存在预测评分效率不高的问题。
发明内容
本申请实施例提供了一种产品智能推荐方法、装置、计算机设备及存储介质,旨在解决现有技术中产品推荐方法所存在的预测评分效率不高的问题。
第一方面,本申请实施例提供了一种产品智能推荐方法,其包括:获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型;获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练;获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分;根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
第二方面,本申请实施例提供了一种产品智能推荐装置,其包括:评分模 型构建单元,用于获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型;评分模型训练单元,用于获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练;预测评分单元,用于获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分;目标产品筛选单元,用于根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
第三方面,本申请实施例又提供了一种计算机设备,其包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序,所述处理器执行所述计算机程序时实现上述第一方面所述的产品智能推荐方法。
第四方面,本申请实施例还提供了一种计算机可读存储介质,其中所述存储介质存储有计算机程序,所述计算机程序当被处理器执行时使所述处理器执行上述第一方面所述的产品智能推荐方法。
为了更清楚地说明本申请实施例技术方案,下面将对实施例描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的产品智能推荐方法的流程示意图;
图2为本申请实施例提供的产品智能推荐方法的评分预测模型示意图;
图3为本申请实施例提供的产品智能推荐方法的子流程示意图;
图4为本申请实施例提供的产品智能推荐方法的另一子流程示意图;
图5为本申请实施例提供的产品智能推荐方法的另一子流程示意图;
图6为本申请实施例提供的产品智能推荐方法的另一子流程示意图;
图7为本申请实施例提供的产品智能推荐装置的示意性框图;
图8为本申请实施例提供的产品智能推荐装置的子单元示意性框图;
图9为本申请实施例提供的产品智能推荐装置的另一子单元示意性框图;
图10为本申请实施例提供的产品智能推荐装置的另一子单元示意性框图;
图11为本申请实施例提供的产品智能推荐装置的另一子单元示意性框图;
图12为本申请实施例提供的计算机设备的示意性框图。
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
请参阅图1,图1是本申请实施例提供的产品智能推荐方法的流程示意图。该产品智能推荐方法应用于管理服务器中,该方法通过安装于管理服务器中的应用软件进行执行,其中,管理服务器即是用于对产品进行智能推荐的企业终端。
如图1所示,该方法包括步骤S101~S104。
S101、获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型。
通过预设的模型构建规则及所述产品信息表中的产品信息,构建包含全连接隐层、输入节点和输出节点的评分预测模型。模型构建规则即是用于对评分预测模型进行构建的规则信息,评分预测模型基于神经网络进行构建,模型构建规则中包括全连接隐层以及对全连接隐层中特征单元的数量进行配置的单元数量配置信息,全连接隐层即是用于对输入节点与输出节点进行联系的中间层。产品信息表中包含企业为消费者所提供的所有产品的信息,产品信息表中的产品信息包括但不限于产品的数量、产品的名称。图2为本申请实施例提供的产品智能推荐方法的评分预测模型示意图,如图2所示,评分预测模型中包括全连接隐层、多个输入节点和多个输出节点。
在一实施例中,如图3所示,步骤S101包括子步骤S1011、S1012、S1013、S1014和S1015。
S1011、根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与输出节点一一对应。
根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与 输出节点一一对应。具体的,根据产品信息中产品的数量构建与产品数量相同的多个输入节点,并构建与产品数量相同的多个输出节点,其中,一个产品对应一个输入节点和一个输出节点,也即是输入节点包含与该输出节点相对应的产品名称,输出节点包含与该输出节点相对应的产品名称,同一个产品的输入节点与输出节点一一对应。
S1012、根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置。
根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置。其中,全连接隐层即是用于对输入节点与输出节点进行联系的中间层,全连接隐层中包含若干个特征单元,每一个特征单元均与所有输入节点和所有输出节点进行关联,特征单元可用于反映产品特性与客户偏好之间的关系,全连接隐层中特征单元的数量可根据单元数量配置信息及产品信息中产品的数量进行配置。例如,单元数量配置信息中预先设定特征单元的数量为产品数量的2/5,产品信息表中企业所有产品的数量为100个,则全连接隐层中特征单元的数量配置为40个。
S1013、根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组。
根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组。其中,第一公式组包含所有输入节点至所有特征单元的公式。输入节点即是评分预测模型中用于对某一个产品的评分进行输入的节点,输入节点的具体数值即为输入节点值,特征单元值即是全连接隐层中的特征单元的计算值。
例如,某一输入节点的输入节点值为x
1,某一特征单元的特征单元值为y
1,则该输入节点至该特征单元的公式为y
1=a×x
1+b;其中,a和b为该公式中的参数。
S1014、根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组。
根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组。其中,第二公式组包含所有特征单元至所有输出节点的公式。输出节点即是评分预测 模型中用于对某一个产品的预测评分进行输出的节点,输出节点的具体数值即是输出节点值,输出节点值即是用于对某一产品进行预测的评分值,特征单元值即是全连接隐层中的特征单元的计算值。
例如,某一特征单元的特征单元值为y
1,某一输出节点的输出节点值为z
1,则该特征单元至该输出节点的公式为z
1=c×y
1+d;其中,c和d为该公式中的参数。
S1015、对所构建的第一公式组和第二公式组中公式的参数进行随机配置。
对第一公式组中公式的参数进行随机配置,对第二公式组中公式的参数进行随机配置,即完成了评分预测模型的构建过程。
S102、获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练。
根据多个客户对产品的评分数据及预设的训练参数,对所构建的评分预测模型进行训练。由于企业所包含的产品数量较多,客户在使用企业的部分产品以后对相应产品进行评分,而由于企业的其他产品该客户未使用,因此该客户只对企业中部分已使用过的产品进行了评分。获取多个客户对产品的评分数据,即可对企业所有产品的评分进行全覆盖,通过多个客户对所有产品的评分数据,对所构建的评分预测模型进行训练。其中,训练参数即是预设的用于对评分预测模型进行训练的参数信息,训练参数中包括均方误差函数、均方误差阈值及参数调整规则。
在一实施例中,如图4所示,步骤S102包括子步骤S1021、S1022、S1023、S1024和S1025。
S1021、获取历史评分数据表中一个客户对产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值。
获取一个客户对产品的评分数据,作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值。由于每一个特征单元均与所有输入节点和所有输出节点进行关联,也即是通过第一公式组中相应公式及输入节点值即可计算得到全连接隐层中所有特征单元的特征单元值。若第一公式组中相应公式的计算结果中某一特征单元包含多个计算值,则取多个计算值的平均值作为该特征单元的特征 单元值;若第一公式组中相应公式的计算结果中某一特征单元仅包含一个计算值,则取该计算值作为该特征单元的特征单元值。
S1022、根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值。
根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值,由于同一个产品的输入节点与输出节点一一对应,通过第二公式组中相应公式及特征单元值即可计算得到与输入节点相对应的同一产品输出节点的输出节点值,该输出节点值即是通过评分预测模型对该产品进行预测得到的评分值,由于评分预测模型中的公式的参数还未经过训练进行调整更新,因此当前计算得到的输出节点值仅能用于对评分预测模型进行训练,而无法作为预测评分结果对客户进行相关产品的推荐。
S1023、获取输入节点的输入节点值以及与输入节点一一对应的输出节点的输出节点值,根据所述均方误差函数对所得到的输入节点值及输出节点值进行计算以得到该客户的均方误差值;
获取输入节点的输入节点值,以及与该输入节点一一对应的输出节点的输出节点值,也即是获取评分预测模型中同一产品的输入节点值及输出节点值作为一个产品的数值组,获取该客户已评分的多个产品的数值组,并根据所述均方误差函数对输入节点值及输出节点值进行计算,以得到该客户的均方误差值,均方误差函数(MSE函数)即是用于反映估计量与被估计量之间差异程度的一种计算函数,通过均方误差函数即可计算得到客户对产品的评分与评分预测模型计算得到的产品评分之间的误差值,均方误差值越大则表明客户的评分与评分预测模型计算得到的评分之间的误差越大,均方误差值越小则表明客户的评分与评分预测模型计算得到的评分之间的误差越小。
具体的,均方误差函数为:均方误差值X
MSE=((z
1-x
1)
2+(z
2-x
2)
2+……+(z
n-x
n)
2)/n;其中,z
n为评分预测模型计算得到的第n个该客户已评分产品的评分,也即是输出节点值;x
n为该客户的第n个已评分产品的评分,也即是输出节点值。
例如,第一个输入节点值为70,与第一个输入节点相对应的第一个输出节点值为55;第二个输入节点值为82,与第二个输入节点相对应的第二个输出节点值为60。则计算得到均方误差值X
MSE=((55-70)
2+(60-82)
2)/2=354.5。
S1024、根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断;
根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断。则判断结果分为两种情况,即均方误差值不大于所述均方误差阈值,以及均方误差值大于所述均方误差阈值。
S1025、若均方误差值大于均方误差阈值,根据所述参数调整规则对评分预测模型中公式的参数进行调整。
若均方误差值大于均方误差阈值,根据所述参数调整规则对评分预测模型中公式的参数进行调整。具体的,若判断结果为均方误差值小于或等于所述均方误差阈值,不对评分预测模型中公式的参数进行调整,获取下一个客户对产品的评分数据以对评分预测模型进行迭代训练;若判断结果为均方误差至大于所述均方误差阈值,根据参数调整规则对第一公式组中相应公式的参数及第二公式组中相应公式的参数进行调整及更新。参数调整规则包括参数调整方向及参数调整幅度,参数调整方向即是用于对公式的参数进行方法调整或缩小调整的方向信息,参数调整幅度即是用于对公式的参数进行调整的幅度信息。
例如,某参数调整规则中的参数调整方向放大调整、参数调整幅度为2%。
为达到良好的训练效果,在获取一个客户对产品的评分数据对评分预测模型进行训练的过程中,可多次输入该客户对产品的评分数据以根据参数调整规则对评分预测模型中公式的参数进行多次调整并计算得到多个均方误差调整值,从所得到的多个均方误差调整值中,选择与最小均方误差调整值相对应的一组参数对第一公式组中相应公式的参数及第二公式组中相应公式的参数进行更新,在对评分预测模型中公式的参数进行更新后,获取下一个客户对产品的评分数据以对评分预测模型进行迭代训练。
S103、获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分。
获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对产品的评分数据以及已训练的评分预测模型对该客户的未评分产品进行评分,以得到预测评分结果,其中,预测评分结果包括该客户所有的未评分产品的预测评分,所述输出节点包括产品名称。
由于企业所包含的产品数量较多,一个客户在使用企业的部分产品以后对相应产品进行评分,而由于企业的其他产品该客户未使用,因此该客户只对企业中部分已使用过的产品进行了评分,所得到的即是该客户对已评分产品的评分数据,其他该客户未进行评分的产品也即是未评分产品。
在一实施例中,如图5所示,步骤S103包括子步骤S1031、S1032和S1033。
S1031、获取该客户对已评分产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值。
获取一个客户对产品的评分数据,作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值。由于每一个特征单元均与所有输入节点和所有输出节点进行关联,也即是通过第一公式组中相应公式及输入节点值即可计算得到全连接隐层中所有特征单元的特征单元值。若第一公式组中相应公式的计算结果中某一特征单元包含多个计算值,则取多个计算值的平均值作为该特征单元的特征单元值;若第一公式组中相应公式的计算结果中某一特征单元仅包含一个计算值,则取该计算值作为该特征单元的特征单元值。
S1032、根据第二公式组中相应公式及特征单元值计算得到所有输出节点的输出节点值。
根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值,由于进过训练之后评分预测模型中第二公式组中所有公式的参数均已进行过调整更新,因此通过第二公式组中所有公式及特征单元值即可计算得到所有输出节点的输出节点值。由于评分预测模型中的公式的参数已通过训练进行了调整及更新,因此当前计算得到的输出节点值能够用于对该客户的未评分产品进行预测评分。
S1033、根据输出节点的产品名称与未评分产品进行匹配,将与未评分产品相匹配的输出节点值作为该客户未评分产品的预测评分。
根据输出节点的产品名称与未评分产品进行匹配,将输出节点中与未评分产品相匹配的输出节点值作为未评分产品的预测评分,获取所有未评分产品的预测评分即可得到预测评分结果。由于一个产品对应一个输入节点和一个输出节点,根据根据输出节点的产品名称与未评分产品进行匹配,即可将相应输出 节点的输出节点值作为未评分产品的预测评分。
S104、根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选,即可得到对该客户进行推荐的目标产品,具体的可选择该客户未评分产品中预测评分较高的产品作为目标产品相该客户进行推荐。
在一实施例中,如图6所示,步骤S104包括子步骤S1041和S1042。
S1041、对该客户的未评分产品的预测评分进行排序以得到未评分产品的排序结果。
对该客户的未评分产品的预测评分进行排序,以得到未评分产品的排序结果。
S1042、根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选以得到对该客户进行推荐的目标产品。
根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选以得到对该客户进行推荐的目标产品,具体的,可根据产品推荐规则中的设置对未评分产品进行筛选,例如,产品推荐规则可以是选择预测评分排名前三的未评分产品作为目标产品向该客户进行推荐。
通过产品信息及模型构建规则构建得到评分预测模型,并通过历史评分数据表对评分预测模型进行训练,以通过评分预测模型及客户已评分产品的评分数据计算得到该客户未评分产品的预测评分,并预设的产品筛选规则从未评分产品中筛选出预测评分较高的目标产品相该客户进行推荐,能够大幅提高对产品进行预测评分的效率以及预测评分的准确性,减少了预测评分耗时,方便了企业筛选出目标产品对客户进行推荐。
本申请实施例还提供一种产品智能推荐装置,该产品智能推荐装置用于执行前述产品智能推荐方法的任一实施例。具体地,请参阅图7,图7是本申请实施例提供的产品智能推荐装置的示意性框图。该产品智能推荐装置100可以配置于管理服务器中。
如图7所示,产品智能推荐装置100包括评分模型构建单元101、评分模型训练单元102、预测评分单元103、目标产品筛选单元104。
评分模型构建单元101,用于获取预先存储的产品信息表,通过预设的模型 构建规则及所述产品信息表中的产品信息构建评分预测模型。
其他申请实施例中,如图8所示,所述评分模型构建单元101包括子单元:节点构建单元1011、特征单元数量配置单元1012、第一公式组构建单元1013、第二公式组构建单元1014和参数配置单元1015。
节点构建单元1011,用于根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与输出节点一一对应。
特征单元数量配置单元1012,用于根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置。
第一公式组构建单元1013,用于根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组。
第二公式组构建单元1014,用于根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组。
参数配置单元1015,用于对所构建的第一公式组和第二公式组中公式的参数进行随机配置。
评分模型训练单元102,用于获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练。
其他申请实施例中,如图9所示,所述评分模型训练单元102包括子单元:第一特征单元值计算单元1021、第一输出节点值计算单元1022、均方误差值计算单元1023、均方误差判断单元1024和参数调整单元1025。
第一特征单元值计算单元1021,用于获取历史评分数据表中一个客户对产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值。
第一输出节点值计算单元1022,用于根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值。
均方误差值计算单元1023,用于获取输入节点的输入节点值以及与输入节点一一对应的输出节点的输出节点值,根据所述均方误差函数对所得到的输入节点值及输出节点值进行计算以得到该客户的均方误差值。
均方误差判断单元1024,用于根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断。
参数调整单元1025,用于若均方误差值大于均方误差阈值,根据所述参数调整规则对评分预测模型中公式的参数进行调整。
预测评分单元103,用于获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分。
其他申请实施例中,如图10所示,所述预测评分单元103包括子单元:第二特征单元值计算单元1031、第二输出节点值计算单元1032和产品名称匹配单元1033。
第二特征单元值计算单元1031,用于获取该客户对已评分产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值。
第二输出节点值计算单元1032,用于根据第二公式组中相应公式及特征单元值计算得到所有输出节点的输出节点值。
产品名称匹配单元1033,用于根据输出节点的产品名称与未评分产品进行匹配,将与未评分产品相匹配的输出节点值作为该客户未评分产品的预测评分。
目标产品筛选单元104,用于根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
其他申请实施例中,如图11所示,所述目标产品筛选单元104包括子单元:产品排序单元1041和产品筛选单元1042。
产品排序单元1041,用于对该客户的未评分产品的预测评分进行排序以得到未评分产品的排序结果。
产品筛选单元1042,用于根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选以得到对该客户进行推荐的目标产品。
上述产品智能推荐装置可以实现为计算机程序的形式,该计算机程序可以在如图12所示的计算机设备上运行。
请参阅图12,图12是本申请实施例提供的计算机设备的示意性框图。
参阅图12,该计算机设备500包括通过系统总线501连接的处理器502、存储器和网络接口505,其中,存储器可以包括非易失性存储介质503和内存储 器504。该非易失性存储介质503可存储操作系统5031和计算机程序5032。该计算机程序5032被执行时,可使得处理器502执行产品智能推荐方法。该处理器502用于提供计算和控制能力,支撑整个计算机设备500的运行。该内存储器504为非易失性存储介质503中的计算机程序5032的运行提供环境,该计算机程序5032被处理器502执行时,可使得处理器502执行产品智能推荐方法。该网络接口505用于进行网络通信,如提供数据信息的传输等。本领域技术人员可以理解,图12中示出的结构,仅仅是与本申请方案相关的部分结构的框图,并不构成对本申请方案所应用于其上的计算机设备500的限定,具体的计算机设备500可以包括比图中所示更多或更少的部件,或者组合某些部件,或者具有不同的部件布置。
其中,所述处理器502用于运行存储在存储器中的计算机程序5032,以实现本实施例的产品智能推荐方法。
本领域技术人员可以理解,图12中示出的计算机设备的实施例并不构成对计算机设备具体构成的限定,在其他实施例中,计算机设备可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件布置。例如,在一些实施例中,计算机设备可以仅包括存储器及处理器,在这样的实施例中,存储器及处理器的结构及功能与图12所示实施例一致,在此不再赘述。
应当理解,在本申请实施例中,处理器502可以是中央处理单元(Central Processing Unit,CPU),该处理器502还可以是其他通用处理器、数字信号处理器(Digital Signal Processor,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现成可编程门阵列(Field-Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。其中,通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
在本申请的另一实施例中提供存储介质。该存储介质可以为非易失性的计算机可读存储介质。该存储介质存储有计算机程序,其中计算机程序被处理器执行时实现本申请实施例的产品智能推荐方法。
所述存储介质可以是前述设备的内部存储单元,例如设备的硬盘或内存。所述存储介质也可以是所述设备的外部存储设备,例如所述设备上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。进一步地,所述存储介质还可以既包括所述设备 的内部存储单元也包括外部存储设备。
所属领域的技术人员可以清楚地了解到,为了描述的方便和简洁,上述描述的设备、装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
以上所述,仅为本申请的具体实施方式,但本申请的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本申请揭露的技术范围内,可轻易想到各种等效的修改或替换,这些修改或替换都应涵盖在本申请的保护范围之内。因此,本申请的保护范围应以权利要求的保护范围为准。
Claims (20)
- 一种产品智能推荐方法,包括:获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型;获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练;获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分;根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 根据权利要求1所述的方法,其中,所述通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型,包括:根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与输出节点一一对应;根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置;根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组;根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组;对所构建的第一公式组和第二公式组中公式的参数进行随机配置。
- 根据权利要求2所述的方法,其中,所述训练参数中包括均方误差函数、均方误差阈值及参数调整规则;所述根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练,包括:获取历史评分数据表中一个客户对产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值;获取输入节点的输入节点值以及与输入节点一一对应的输出节点的输出节点值,根据所述均方误差函数对所得到的输入节点值及输出节点值进行计算以得到该客户的均方误差值;根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断;若均方误差值大于均方误差阈值,根据所述参数调整规则对评分预测模型中公式的参数进行调整。
- 根据权利要求2所述的方法,其中,所述根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分,包括:获取该客户对已评分产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;根据第二公式组中相应公式及特征单元值计算得到所有输出节点的输出节点值;根据输出节点的产品名称与未评分产品进行匹配,将与未评分产品相匹配的输出节点值作为该客户未评分产品的预测评分。
- 根据权利要求1所述的方法,其中,所述根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品,包括:对该客户的未评分产品的预测评分进行排序以得到未评分产品的排序结果;根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 一种产品智能推荐装置,包括:评分模型构建单元,用于获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型;评分模型训练单元,用于获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练;预测评分单元,用于获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分;目标产品筛选单元,用于根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 根据权利要求6所述的装置,其中,所述评分模型构建单元,包括:节点构建单元,用于根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与输出节点一一对应;特征单元数量配置单元,用于根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置;第一公式组构建单元,用于根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组;第二公式组构建单元,用于根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组;参数配置单元,用于对所构建的第一公式组和第二公式组中公式的参数进行随机配置。
- 根据权利要求7所述的装置,其中,所述评分模型训练单元,包括:第一特征单元值计算单元,用于获取历史评分数据表中一个客户对产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;第一输出节点值计算单元,用于根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值;均方误差值计算单元,用于获取输入节点的输入节点值以及与输入节点一一对应的输出节点的输出节点值,根据所述均方误差函数对所得到的输入节点值及输出节点值进行计算以得到该客户的均方误差值;均方误差判断单元,用于根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断;参数调整单元,用于若均方误差值大于均方误差阈值,根据所述参数调整 规则对评分预测模型中公式的参数进行调整。
- 根据权利要求7所述的装置,其中,所述预测评分单元,包括:第二特征单元值计算单元,用于获取该客户对已评分产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;第二输出节点值计算单元,用于根据第二公式组中相应公式及特征单元值计算得到所有输出节点的输出节点值;产品名称匹配单元,用于根据输出节点的产品名称与未评分产品进行匹配,将与未评分产品相匹配的输出节点值作为该客户未评分产品的预测评分。
- 根据权利要求6所述的装置,其中,所述目标产品筛选单元,包括:产品排序单元,用于对该客户的未评分产品的预测评分进行排序以得到未评分产品的排序结果;产品筛选单元,用于根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 一种计算机设备,包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序,其中,所述处理器执行所述计算机程序时实现以下步骤:获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型;获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练;获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分;根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 根据权利要求11所述的计算机设备,其中,所述通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型,包括:根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与输出节点一一对应;根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置;根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组;根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组;对所构建的第一公式组和第二公式组中公式的参数进行随机配置。
- 根据权利要求12所述的计算机设备,其中,所述训练参数中包括均方误差函数、均方误差阈值及参数调整规则;所述根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练,包括:获取历史评分数据表中一个客户对产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值;获取输入节点的输入节点值以及与输入节点一一对应的输出节点的输出节点值,根据所述均方误差函数对所得到的输入节点值及输出节点值进行计算以得到该客户的均方误差值;根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断;若均方误差值大于均方误差阈值,根据所述参数调整规则对评分预测模型中公式的参数进行调整。
- 根据权利要求12所述的计算机设备,其中,所述根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分,包括:获取该客户对已评分产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;根据第二公式组中相应公式及特征单元值计算得到所有输出节点的输出节点值;根据输出节点的产品名称与未评分产品进行匹配,将与未评分产品相匹配的输出节点值作为该客户未评分产品的预测评分。
- 根据权利要求11所述的计算机设备,其中,所述根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品,包括:对该客户的未评分产品的预测评分进行排序以得到未评分产品的排序结果;根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 一种计算机可读存储介质,其中,所述存储介质存储有计算机程序,所述计算机程序当被处理器执行时使所述处理器执行以下操作:获取预先存储的产品信息表,通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型;获取预先存储的历史评分数据表,根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练;获取预先存储的历史评分数据表中一个客户的评分数据,根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分;根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品。
- 根据权利要求16所述的存储介质,其中,所述通过预设的模型构建规则及所述产品信息表中的产品信息构建评分预测模型,包括:根据产品信息构建评分预测模型的输入节点和输出节点,所述输入节点与输出节点一一对应;根据所述产品信息及模型构建规则中的单元数量配置信息对模型构建规则中的全连接隐层中特征单元的数量进行配置;根据所述全连接隐层的特征单元及所构建的输入节点,以输入节点值作为输入值、特征单元值作为输出值构建输入节点至特征单元的第一公式组;根据所述全连接隐层的特征单元及所构建的输出节点,以特征单元值作为输入值、输出节点值作为输出值构建特征单元至输出节点的第二公式组;对所构建的第一公式组和第二公式组中公式的参数进行随机配置。
- 根据权利要求17所述的存储介质,其中,所述训练参数中包括均方误差函数、均方误差阈值及参数调整规则;所述根据所述历史评分数据表中产品的评分数据及预设的训练参数对所构建的评分预测模型进行训练,包括:获取历史评分数据表中一个客户对产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;根据第二公式组中相应公式及特征单元值计算得到相应输出节点的输出节点值;获取输入节点的输入节点值以及与输入节点一一对应的输出节点的输出节点值,根据所述均方误差函数对所得到的输入节点值及输出节点值进行计算以得到该客户的均方误差值;根据所述均方误差阈值对均方误差值是否大于均方误差阈值进行判断;若均方误差值大于均方误差阈值,根据所述参数调整规则对评分预测模型中公式的参数进行调整。
- 根据权利要求17所述的存储介质,其中,所述根据该客户对已评分产品的评分数据及已训练的评分预测模型对该客户的未评分产品进行评分以得到未评分产品的预测评分,包括:获取该客户对已评分产品的评分数据作为第一公式组中相应公式的输入节点值,根据第一公式组中相应公式及输入节点值计算得到全连接隐层中所有特征单元的特征单元值;根据第二公式组中相应公式及特征单元值计算得到所有输出节点的输出节点值;根据输出节点的产品名称与未评分产品进行匹配,将与未评分产品相匹配的输出节点值作为该客户未评分产品的预测评分。
- 根据权利要求16所述的存储介质,其中,所述根据预设的产品筛选规则及未评分产品的预测评分对该客户的未评分产品进行筛选以得到对该客户进行推荐的目标产品,包括:对该客户的未评分产品的预测评分进行排序以得到未评分产品的排序结果;根据预设的产品推荐规则及未评分产品的排序结果对未评分产品进行筛选 以得到对该客户进行推荐的目标产品。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811029043.7A CN109146193B (zh) | 2018-09-05 | 2018-09-05 | 产品智能推荐方法、装置、计算机设备及存储介质 |
| CN201811029043.7 | 2018-09-05 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020048065A1 true WO2020048065A1 (zh) | 2020-03-12 |
Family
ID=64826825
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2018/125345 Ceased WO2020048065A1 (zh) | 2018-09-05 | 2018-12-29 | 产品智能推荐方法、装置、计算机设备及存储介质 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109146193B (zh) |
| WO (1) | WO2020048065A1 (zh) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113763017A (zh) * | 2021-01-19 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | 一种信息处理方法及装置、存储介质 |
| CN114418663A (zh) * | 2021-12-10 | 2022-04-29 | 珠海格力电器股份有限公司 | 商品信息处理方法、装置、计算机设备和存储介质 |
| CN115935533A (zh) * | 2022-11-02 | 2023-04-07 | 北京能科瑞元数字技术有限公司 | 基于参数的产品智能设计系统 |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458191B (zh) * | 2019-07-05 | 2024-04-12 | 中国平安财产保险股份有限公司 | 疲劳状态判断方法、装置、计算机设备及存储介质 |
| CN110533453B (zh) * | 2019-07-22 | 2024-10-18 | 平安科技(深圳)有限公司 | 基于用户匹配的产品推荐方法、装置、计算机设备 |
| CN110633304B (zh) * | 2019-08-15 | 2023-08-22 | 中国平安人寿保险股份有限公司 | 组合特征筛选方法、装置、计算机设备及存储介质 |
| CN110674104B (zh) * | 2019-08-15 | 2023-07-25 | 中国平安人寿保险股份有限公司 | 特征组合筛选方法、装置、计算机设备及存储介质 |
| CN111612581A (zh) * | 2020-05-18 | 2020-09-01 | 深圳市分期乐网络科技有限公司 | 一种物品推荐的方法、装置、设备及存储介质 |
| CN112330411B (zh) * | 2020-11-17 | 2023-10-31 | 中国平安财产保险股份有限公司 | 团体产品推荐方法、装置、计算机设备及存储介质 |
| CN113656122B (zh) * | 2021-07-28 | 2023-05-16 | 上海纽盾科技股份有限公司 | 面向等保测评的信息筛选方法、装置及系统 |
| CN113705782A (zh) * | 2021-08-18 | 2021-11-26 | 上海明略人工智能(集团)有限公司 | 一种用于媒体数据推荐的模型训练方法及装置 |
| CN118445868A (zh) * | 2024-06-06 | 2024-08-06 | 北京轻而易举智能科技有限责任公司 | 根据户型尺寸自动设计定制柜的方法及装置 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103294812A (zh) * | 2013-06-06 | 2013-09-11 | 浙江大学 | 一种基于混合模型的商品推荐方法 |
| CN104424247A (zh) * | 2013-08-28 | 2015-03-18 | 北京闹米科技有限公司 | 一种产品信息过滤推荐方法和装置 |
| CN106127506A (zh) * | 2016-06-13 | 2016-11-16 | 浙江大学 | 一种基于主动学习解决商品冷启动问题的推荐方法 |
| CN108287902A (zh) * | 2018-01-24 | 2018-07-17 | 厦门大学 | 一种基于数据非随机缺失机制的推荐系统方法 |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160225073A1 (en) * | 2015-01-30 | 2016-08-04 | Wal-Mart Stores, Inc. | System, method, and non-transitory computer-readable storage media for predicting a customer's credit score |
| CN107563841B (zh) * | 2017-08-03 | 2021-02-05 | 电子科技大学 | 一种基于用户评分分解的推荐系统 |
-
2018
- 2018-09-05 CN CN201811029043.7A patent/CN109146193B/zh active Active
- 2018-12-29 WO PCT/CN2018/125345 patent/WO2020048065A1/zh not_active Ceased
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103294812A (zh) * | 2013-06-06 | 2013-09-11 | 浙江大学 | 一种基于混合模型的商品推荐方法 |
| CN104424247A (zh) * | 2013-08-28 | 2015-03-18 | 北京闹米科技有限公司 | 一种产品信息过滤推荐方法和装置 |
| CN106127506A (zh) * | 2016-06-13 | 2016-11-16 | 浙江大学 | 一种基于主动学习解决商品冷启动问题的推荐方法 |
| CN108287902A (zh) * | 2018-01-24 | 2018-07-17 | 厦门大学 | 一种基于数据非随机缺失机制的推荐系统方法 |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113763017A (zh) * | 2021-01-19 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | 一种信息处理方法及装置、存储介质 |
| CN114418663A (zh) * | 2021-12-10 | 2022-04-29 | 珠海格力电器股份有限公司 | 商品信息处理方法、装置、计算机设备和存储介质 |
| CN115935533A (zh) * | 2022-11-02 | 2023-04-07 | 北京能科瑞元数字技术有限公司 | 基于参数的产品智能设计系统 |
| CN115935533B (zh) * | 2022-11-02 | 2023-10-03 | 北京能科瑞元数字技术有限公司 | 基于参数的产品智能设计系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109146193A (zh) | 2019-01-04 |
| CN109146193B (zh) | 2023-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020048065A1 (zh) | 产品智能推荐方法、装置、计算机设备及存储介质 | |
| US11055453B2 (en) | Automated configuration and data collection during modeling of network devices | |
| US11288598B2 (en) | Third-party analytics service with virtual assistant interface | |
| US11861733B2 (en) | Expense report submission interface | |
| US20190005431A1 (en) | Cross-domain multi-attribute hashed and weighted dynamic process prioritization | |
| WO2018103595A1 (zh) | 一种授权策略推荐方法及装置、服务器、存储介质 | |
| CN110730101B (zh) | 资源分配方法、终端、装置及可读存储介质 | |
| CN111797320A (zh) | 数据处理方法、装置、设备及存储介质 | |
| US20240127144A1 (en) | Task workflow modeling interface | |
| US10373071B2 (en) | Automated intelligent data navigation and prediction tool | |
| US20210073921A1 (en) | Expense report reviewing interface | |
| US20220207222A1 (en) | Interactive system for guiding electronics design and assembly | |
| WO2017080171A1 (zh) | 大数据的筛选分析方法及系统 | |
| CN114757397A (zh) | 一种坏料预测方法、坏料预测装置及电子设备 | |
| CN114417980A (zh) | 一种业务模型的建立方法、装置、电子设备及存储介质 | |
| US20190171989A1 (en) | Dynamic Workflow Control Based on Database Logic | |
| CN107704364B (zh) | 后台应用程序管控方法、装置、存储介质及电子设备 | |
| US11551024B1 (en) | Hybrid clustered prediction computer modeling | |
| US12608536B2 (en) | Using data submitted for a field to populate a different, associated field | |
| CN115019922B (zh) | 一种录入与审核的反馈方法、装置、设备以及存储介质 | |
| US12541681B2 (en) | Discrete optimization of configuration attributes | |
| US20190158628A1 (en) | Universal self-learning softsensor and its built platform that applies machine learning and advanced analytics | |
| CN110413632A (zh) | 管理状态的方法、装置、计算机可读介质及电子设备 | |
| CN109345341A (zh) | 商品筛选方法、装置、介质及电子设备 | |
| JP7545675B2 (ja) | 情報処理装置、情報処理方法、およびプログラム |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18932352 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 18932352 Country of ref document: EP Kind code of ref document: A1 |