WO2020253357A1 - 数据产品推荐方法、装置、计算机设备和存储介质 - Google Patents

数据产品推荐方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2020253357A1
WO2020253357A1 PCT/CN2020/086060 CN2020086060W WO2020253357A1 WO 2020253357 A1 WO2020253357 A1 WO 2020253357A1 CN 2020086060 W CN2020086060 W CN 2020086060W WO 2020253357 A1 WO2020253357 A1 WO 2020253357A1
Authority
WO
WIPO (PCT)
Prior art keywords
target
data product
historical
information
customer group
Prior art date
Application number
PCT/CN2020/086060
Other languages
English (en)
French (fr)
Inventor
马新俊
Original Assignee
深圳壹账通智能科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳壹账通智能科技有限公司 filed Critical 深圳壹账通智能科技有限公司
Publication of WO2020253357A1 publication Critical patent/WO2020253357A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201Market modelling; Market analysis; Collecting market data
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q30/00Commerce
    • G06Q30/06Buying, selling or leasing transactions
    • G06Q30/0601Electronic shopping [e-shopping]
    • G06Q30/0631Item recommendations

Definitions

  • This application relates to the field of big data technology, in particular to a data product recommendation method, device, computer equipment, and storage medium.
  • Data product is a form of product that can play the value of data to assist users in making better decisions (or even actions). For example: flow analysis products, sales analysis products, personalized recommendation products and so on.
  • a data product recommendation method includes:
  • the data product recommendation instruction carries the target identifier and the target customer group information corresponding to the target identifier;
  • the method before receiving the data product recommendation instruction, the data product recommendation instruction carries the target identifier and target customer group information, the method further includes:
  • a data product recommendation device which includes:
  • the instruction receiving module is used to receive the data product recommendation instruction, and the data product recommendation instruction carries the target identifier and the target customer group information corresponding to the target identifier;
  • the matrix building module is used to build a target feature matrix based on the target customer group information
  • the identification determination module is used to obtain the historical feature matrix corresponding to each historical target identification, calculate the similarity between the historical feature matrix and the target feature matrix, and determine the historical target identification corresponding to the target identification according to the similarity;
  • the recommendation module is used to obtain the data product information corresponding to the historical target identifier, and send the data product information to the target terminal corresponding to the target identifier.
  • a computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:
  • the data product recommendation instruction carries the target identifier and the target customer group information corresponding to the target identifier;
  • a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:
  • the data product recommendation instruction carries the target identifier and target customer group information corresponding to the target identifier;
  • the above data product recommendation method, device, computer equipment and storage medium calculate the similarity between the target feature matrix of the target identification customer group and the historical feature matrix of the historical target identification customer group to determine the most similar historical target identification of the target identification customer group Customer groups, then according to the most similar historical target identification customer group’s historical target identification has purchased data products to recommend, that is, the data products purchased with the same customer group’s historical target identification are recommended to the target identification, which improves the efficiency of data product recommendation And pertinence.
  • Fig. 1 is an application scenario diagram of a data product recommendation method in an embodiment
  • FIG. 2 is a schematic flowchart of a data product recommendation method in an embodiment
  • FIG. 3 is a schematic flowchart of the step of obtaining a historical target feature matrix in an embodiment
  • FIG. 4 is a schematic diagram of obtaining a feature matrix of historical cooperative institutions in a specific embodiment
  • Figure 5 is a schematic diagram of a process for obtaining a target feature matrix in an embodiment
  • Figure 6 is a block diagram of a data product recommendation device in an embodiment
  • Fig. 7 is an internal structure diagram of a computer device in an embodiment.
  • the data product recommendation method provided in this application can be applied to the application environment shown in FIG. 1.
  • the terminal 102 communicates with the server 104 through the network.
  • the server 104 receives the data product recommendation instruction sent by the target terminal 102.
  • the data product recommendation instruction carries the target identification and target customer group information corresponding to the target identification;
  • the server 104 builds a target feature matrix according to the target customer group information;
  • the server 104 obtains each historical target Identify the corresponding historical feature matrix, calculate the similarity between the historical feature matrix and the target feature matrix, and determine the historical target identifier corresponding to the target identifier according to the similarity;
  • the server 104 obtains the data product information corresponding to the historical target identifier, and sends the data product information to the target identifier
  • the corresponding target terminal 104 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.
  • the server 104 may be implemented as an independent server or a server cluster composed of multiple servers
  • a method for recommending data products is provided. Taking the method applied to the server in FIG. 1 as an example for description, the method includes the following steps:
  • S202 Receive a data product recommendation instruction, where the data product recommendation instruction carries the target identifier and target customer group information corresponding to the target identifier.
  • the target identifier refers to the enterprise identifier, which is used to uniquely identify the enterprise that wants to purchase the data product.
  • the target customer group information refers to the enterprise customer group information, which is the detailed information of the customer group served by the company that wants to purchase the data product.
  • the customer group information can include multiple attributes, such as customer group preference attributes and customer group age attributes And the geographical attributes of the customer group and so on.
  • the server receives the data product recommendation instruction sent by the enterprise terminal, and the data product recommendation instruction carries the enterprise identity and the enterprise customer group information corresponding to the enterprise identity.
  • the target feature matrix refers to the numerical feature matrix established based on the information of the enterprise customer group.
  • the server obtains the information of each attribute of the target customer group according to the target customer group information.
  • the numerical feature refers to a preset mathematical feature to be calculated, such as expectations.
  • the numerical feature vector of each attribute of the target customer group can be obtained, and the numerical feature matrix can be obtained according to the numerical feature vector of the attribute of the target customer group.
  • the customer group A has attribute 1, attribute 2, and attribute 3.
  • the numerical feature vector of attribute 1 is calculated according to the customer information of attribute 1 of customer group A (A11).
  • the obtained numerical characteristic matrix of the enterprise corresponding to the target identifier is [(A11),(A21),(A31)]
  • S206 Obtain the historical feature matrix corresponding to each historical target identifier, calculate the similarity between the historical feature matrix and the target feature matrix, and determine the historical target identifier corresponding to the target identifier according to the similarity.
  • the historical target identification refers to the corporate identification of the company that has historically purchased data products
  • the historical feature matrix refers to the historical numerical feature matrix established in advance based on the historical corporate identification corresponding to the information of each attribute of the customer group of each enterprise.
  • the row in the historical numerical characteristic matrix represents the enterprise corresponding to the historical enterprise identifier
  • the column represents the numerical characteristic corresponding to the attribute of the customer group of the historical enterprise.
  • the historical numerical characteristic matrix has M rows and N columns, that is, the historical numerical characteristic matrix obtained can be B represents the historical feature matrix.
  • B mn represents the numerical feature vector of the nth attribute of the mth company.
  • the server obtains the historical feature matrix corresponding to each historical enterprise identity, and calculates the similarity between the historical feature matrix and the target feature matrix respectively, that is, calculates the numerical feature vector of each customer group attribute in the historical feature matrix and the customer group in the target feature matrix
  • the similarity between the numerical feature vectors of the attributes is used to indicate the similarity between the customer groups of each historical company and the customer groups of the target company.
  • the similarity between each customer group in the calculated historical feature matrix and the customer group in the target feature matrix is sorted.
  • Obtain the historical customer group in the historical feature matrix corresponding to the largest similarity in the sorting result obtain the historical enterprise identity corresponding to the historical customer group, and obtain the historical enterprise identity with the most similar enterprise identity. Similar historical enterprises.
  • the cosine similarity algorithm is used to calculate the similarity between customer groups.
  • the characteristic vector of the customer group is obtained according to the historical enterprise characteristic matrix
  • the characteristic vector of the customer group is obtained according to the enterprise characteristic matrix
  • the similarity between the characteristic vectors of the customer group is calculated according to the cosine similarity calculation formula.
  • S208 Acquire data product information corresponding to the historical target identifier, and send the data product information to the target terminal corresponding to the target identifier.
  • the server obtains the data product information that the company corresponding to the saved historical company identifier has purchased.
  • the data product information that the historical company has purchased is sent to the corresponding enterprise terminal that wants to purchase the data product, and the enterprise terminal receives the historical company’s purchased data product information. Data product information and display it so that business users can choose to purchase the displayed data product.
  • the data product recommendation instruction by receiving the data product recommendation instruction, the data product recommendation instruction carries the target identifier and the target customer group information corresponding to the target identifier, and builds the target feature matrix based on the target customer group information, and obtains the corresponding historical target identifiers Calculate the similarity between the historical feature matrix and the target feature matrix, determine the historical target identifier corresponding to the target identifier according to the similarity, obtain the data product information corresponding to the historical target identifier, and send the data product information to the target terminal corresponding to the target identifier .
  • the efficiency and pertinence of data product recommendations are improved.
  • the data product recommendation instruction before step S202, that is, before receiving the data product recommendation instruction, carries the target identifier and target customer group information, and further includes the following steps:
  • S302 Obtain historical target customer group information corresponding to each historical target identifier, and preprocess the historical target customer group information to obtain each attribute value of the historical target customer group.
  • preprocessing refers to data cleaning, data conversion, and data standardization in historical target customer group information.
  • the server obtains historical enterprise customer group information corresponding to each historical enterprise identifier, preprocesses the historical enterprise customer group information, and obtains each attribute value of the historical customer group. For example, data conversion is performed on the gender attributes of customers in the historical customer group, and the gender is converted into a value of 1 for male and a value of 0 for gender. and many more.
  • S304 Calculate the numerical characteristics of each attribute of the historical target customer group, and generate a numerical characteristic matrix.
  • the server calculates the numerical characteristics of each attribute of the historical target customer group, including the expected value Ex, the entropy En, and the hyper-entropy He, and generates a numerical characteristic matrix according to the calculated numerical characteristics.
  • the numerical characteristic matrix is normalized, and the numerical value in the numerical characteristic matrix is linearly transformed, and the result value is mapped between 0 and 1. That is, the historical target feature matrix corresponding to each historical target identifier is obtained.
  • the normalization can be calculated using min-max standardization or Z-score standardization methods.
  • M historical cooperative organization enterprises correspond to M cooperative organization enterprise customer groups, and each customer group has N attributes.
  • the numerical characteristics of each attribute include expectations, Entropy and super entropy, get M*3N order matrix.
  • M represents that there are M customer groups, and 3N represents the expectations, entropy and super entropy of N attributes. Then normalize the obtained matrix to obtain the characteristic matrix of historical cooperative institutions.
  • the historical target customer group information is preprocessed to obtain each attribute value of the historical target customer group, and the numerical characteristics of each attribute of the historical target customer group are calculated to generate the value
  • the feature matrix is to normalize the numerical feature matrix to obtain the historical target feature matrix corresponding to each historical target identifier.
  • the historical target feature matrix is obtained by pre-calculation, which can be directly used when calculating the similarity of the customer group, which is convenient and quick.
  • step S204 namely, establishing a target feature matrix according to the target customer group information, includes the steps:
  • S502 Preprocess the target customer group information to obtain various attribute values of the target customer group.
  • the server preprocesses the enterprise customer group information, including data cleaning, data conversion, and data standardization, etc., to obtain various attribute values of the enterprise customer group information.
  • S506 Calculate the numerical characteristics of each attribute of the target customer group, and generate a numerical characteristic matrix.
  • the age attribute of a customer group the age of customers in the customer group can be 18, 24, 26, 20, and so on.
  • the numerical characteristics of the age attribute namely expectation, entropy and super entropy, are calculated to represent the numerical characteristics of the age attribute.
  • S507 Normalize the numerical feature matrix to obtain a target feature matrix corresponding to the target identifier.
  • the server performs normalization processing on the generated numerical feature matrix, that is, linearly transforms the values in the numerical feature matrix, and maps the result value to a value between 0 and 1, to obtain the corporate feature matrix corresponding to the corporate identity.
  • each attribute value of the target customer group is obtained, the numerical characteristics of each attribute of the target customer group are calculated, the numerical characteristic matrix is generated, and the numerical characteristic matrix is normalized to obtain the target identification corresponding
  • the target feature matrix can be calculated to obtain the feature matrix of the customer group, which is convenient for subsequent use.
  • step S502 which is to preprocess the target customer group information to obtain the attribute values of the target customer group, includes the steps:
  • the attribute information in the enterprise customer group information acquired by the server is non-numeric attribute information, such as gender attributes, preference attributes, education attributes, and occupation attributes of the enterprise customer groups.
  • the preset rule refers to a preset mapping relationship between a non-numeric attribute and a value, such as mapping male to a value of 1, and female to a value of 0 in the gender attribute. That is, each attribute in the target customer group is converted into an attribute value, and each attribute value of the target customer group is obtained, which is convenient for subsequent calculation of numerical characteristics and improves calculation efficiency.
  • step S208 that is, obtaining the data product information corresponding to the historical target identifier, and sending the data product information to the target terminal corresponding to the target identifier, includes the steps:
  • the purchased data product information corresponding to the target identifier is acquired, and when the data product information corresponding to the historical target identifier does not contain the purchased data product information, the data product information is sent to the target terminal corresponding to the target identifier.
  • the server obtains the data product information that the company corresponding to the company ID has purchased, and the data product information may be a product name, a product ID, and so on. It is judged whether the data product information that the company has purchased is in the recommended data product information, that is, it can be judged according to the product name, and it can be judged whether the purchased product name is in the recommended product name list. When there is no data product information that the company has purchased in the recommended data product information, the recommended data product information is sent to the company terminal corresponding to the company identifier for display.
  • step S208 acquiring the data product information corresponding to the historical target identifier, and sending the data product information to the target terminal corresponding to the target identifier, includes the steps:
  • the corresponding evaluation information is acquired according to the data product information, the evaluation score is calculated according to the evaluation information, and the data product information with the evaluation score greater than the preset threshold is sent to the target terminal corresponding to the target identifier.
  • the server obtains the evaluation information of each data product according to the data product information to be recommended, and calculates the evaluation score according to the evaluation information.
  • the keyword in the evaluation information can be extracted through the keyword extraction algorithm, and the pre-set keywords The corresponding relationship with the score is calculated to obtain the evaluation score of the evaluation information.
  • the server sends the data product information with the evaluation score greater than the preset threshold to the enterprise terminal corresponding to the enterprise identification for display.
  • calculating the evaluation score according to the evaluation information includes the steps:
  • the corresponding evaluation vector is established according to the evaluation information, and the evaluation vector is input into the trained data product evaluation model to obtain the output result vector of the data product evaluation model, and the evaluation score is obtained according to the output result vector of the data product evaluation model.
  • the trained data product evaluation model is obtained by training using historical data product evaluation information and historical evaluation scores using linear regression algorithms.
  • the server establishes a corresponding evaluation vector according to the evaluation information, and generates a corresponding evaluation vector according to the mapping relationship according to the evaluation level and the evaluation keyword in the evaluation information. For example, if the evaluation level includes three levels of poor, good, and excellent, the evaluation If the level is good, the corresponding evaluation vector is (0,1,0). Input the obtained evaluation vector into the trained data product evaluation model to obtain the output result vector of the data product evaluation model, and obtain the evaluation score according to the output result vector of the data product evaluation model. Using the data product evaluation model to obtain the evaluation score can improve the accuracy and accuracy of the evaluation branch.
  • a data product recommendation device 600 including: an instruction receiving module 602, a matrix establishing module 604, an identification determining module 606, and a recommendation module 608, wherein:
  • the instruction receiving module 602 is configured to receive a data product recommendation instruction, and the data product recommendation instruction carries target identification and target customer group information corresponding to the target identification;
  • the matrix building module 604 is used to build a target feature matrix according to the target customer group information
  • the identification determination module 606 is configured to obtain the historical feature matrix corresponding to each historical target identification, calculate the similarity between the historical feature matrix and the target feature matrix, and determine the historical target identification corresponding to the target identification according to the similarity;
  • the recommendation module 608 is configured to obtain the data product information corresponding to the historical target identifier, and send the data product information to the target terminal corresponding to the target identifier.
  • the data product recommendation device 600 further includes:
  • the historical information obtaining module is used to obtain historical target customer group information corresponding to each historical target identifier, preprocess the historical target customer group information, and obtain each attribute value of the historical target customer group;
  • the historical calculation module is used to calculate the numerical characteristics of each attribute of the historical target customer group and generate a numerical characteristic matrix
  • the historical matrix obtaining module is used to normalize the numerical feature matrix to obtain the historical target feature matrix corresponding to each historical target identifier.
  • the matrix building module 604 includes:
  • the information obtaining module preprocesses the target customer group information to obtain the value of each attribute of the target customer group
  • the calculation module is used to calculate the numerical characteristics of each attribute of the target customer group and generate a numerical characteristic matrix
  • the matrix obtaining module is used to normalize the numerical feature matrix to obtain the target feature matrix corresponding to the target identifier.
  • the information obtaining module includes:
  • the information conversion module is used to obtain the non-numerical attribute information in the target customer group information, convert the non-numerical attribute information into numerical attribute information according to preset rules, and obtain the attribute values of the target customer group.
  • the recommendation module 608 includes:
  • the information judgment module is used to obtain the purchased data product information corresponding to the target identifier.
  • the data product information corresponding to the historical target identifier does not contain the purchased data product information, the data product information is sent to the target terminal corresponding to the target identifier.
  • the recommendation module 608 includes:
  • the evaluation score calculation module is used to obtain corresponding evaluation information according to the data product information, calculate the evaluation score according to the evaluation information, and send the data product information with the evaluation score greater than a preset threshold to the target terminal corresponding to the target identifier.
  • the evaluation score calculation module includes:
  • the module calculation module is used to establish the corresponding evaluation vector according to the evaluation information, and input the evaluation vector into the trained data product evaluation model to obtain the output result vector of the data product evaluation model, and get the evaluation according to the output result vector of the data product evaluation model Points
  • Each module in the above-mentioned data product recommendation device can be implemented in whole or in part by software, hardware, and a combination thereof.
  • the foregoing modules may be embedded in the form of hardware or independent of the processor in the computer device, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the foregoing modules.
  • a computer device is provided.
  • the computer device may be a server, and its internal structure diagram may be as shown in FIG. 7.
  • the computer equipment includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, a computer program, and a database.
  • the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
  • the database of the computer equipment is used to store data product data.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer program is executed by the processor to realize a data product recommendation method.
  • FIG. 7 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied.
  • the specific computer equipment may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.
  • the processor further implements the following steps when executing the computer program: receiving a data product recommendation instruction, the data product recommendation instruction carrying the target identifier and target customer group information corresponding to the target identifier; establishing target characteristics based on the target customer group information Matrix; obtain the historical feature matrix corresponding to each historical target identifier, calculate the similarity between the historical feature matrix and the target feature matrix, and determine the historical target identifier corresponding to the target identifier according to the similarity; obtain the data product information corresponding to the historical target identifier, and compare the data product The target terminal corresponding to the information sending target identifier.
  • the processor further implements the following steps when executing the computer program: acquiring historical target customer group information corresponding to each historical target identifier, preprocessing the historical target customer group information to obtain each attribute value of the historical target customer group; calculating history The numerical characteristics of each attribute of the target customer group are generated to generate a numerical characteristic matrix; the numerical characteristic matrix is normalized to obtain the historical target characteristic matrix corresponding to each historical target identifier.
  • the processor further implements the following steps when executing the computer program: preprocess the target customer group information to obtain the value of each attribute of the target customer group; calculate the numerical characteristics of each attribute of the target customer group to generate a numerical characteristic matrix; The feature matrix is normalized to obtain the target feature matrix corresponding to the target identifier.
  • the processor further implements the following steps when executing the computer program: acquiring the non-numerical attribute information in the target customer group information, and converting the non-numerical attribute information into numerical attribute information according to preset rules to obtain each attribute of the target customer group value.
  • the processor further implements the following steps when executing the computer program: acquiring the purchased data product information corresponding to the target identifier, and when the data product information corresponding to the historical target identifier does not contain the purchased data product information, the data product The target terminal corresponding to the information sending target identifier.
  • the processor further implements the following steps when executing the computer program: obtaining corresponding evaluation information according to the data product information, calculating evaluation scores based on the evaluation information, and sending data product information with evaluation scores greater than a preset threshold to the target identifier The corresponding target terminal.
  • the processor further implements the following steps when executing the computer program: establishing a corresponding evaluation vector according to the evaluation information, inputting the evaluation vector into the trained data product evaluation model, and obtaining the output result vector of the data product evaluation model, The evaluation score is obtained according to the output result vector of the data product evaluation model.
  • a computer-readable storage medium is provided.
  • the storage medium is a volatile storage medium or a non-volatile storage medium, and a computer program is stored thereon.
  • the computer program is executed by a processor, the following steps are implemented : Receive the data product recommendation instruction, the data product recommendation instruction carries the target identification and the target customer group information corresponding to the target identification; establish the target feature matrix according to the target customer group information; obtain the historical feature matrix corresponding to each historical target identification, and calculate the historical feature Based on the similarity between the matrix and the target feature matrix, the historical target identifier corresponding to the target identifier is determined according to the similarity; the data product information corresponding to the historical target identifier is obtained, and the data product information is sent to the target terminal corresponding to the target identifier.
  • the following steps are also implemented: obtaining historical target customer group information corresponding to each historical target identifier, preprocessing the historical target customer group information to obtain each attribute value of the historical target customer group; The numerical characteristics of each attribute of the historical target customer group are generated to generate a numerical characteristic matrix; the numerical characteristic matrix is normalized to obtain the historical target characteristic matrix corresponding to each historical target identifier.
  • the following steps are further implemented: preprocessing the target customer group information to obtain the value of each attribute of the target customer group; calculating the numerical characteristics of each attribute of the target customer group to generate a numerical characteristic matrix;
  • the following steps are also implemented: obtaining non-numerical attribute information in the target customer group information, and converting the non-numerical attribute information into numerical attribute information according to preset rules to obtain each target customer group. Attribute value.
  • the following steps are also implemented: acquiring the purchased data product information corresponding to the target identifier, and when the data product information corresponding to the historical target identifier does not contain the purchased data product information, the data The target terminal corresponding to the product information transmission target identifier.
  • the following steps are also implemented: obtaining corresponding evaluation information according to data product information, calculating evaluation scores based on the evaluation information, and sending data product information with evaluation scores greater than a preset threshold to the target Identify the corresponding target terminal.
  • the following steps are also implemented: establish a corresponding evaluation vector based on the evaluation information, and input the evaluation vector into the trained data product evaluation model to obtain the output result vector of the data product evaluation model , According to the output result vector of the data product evaluation model, the evaluation score is obtained.
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Business, Economics & Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Engineering & Computer Science (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种数据产品推荐方法、装置、计算机设备和存储介质,涉及大数据技术。所述方法包括:接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息(S202);根据目标客户群体信息建立目标特征矩阵(S204);获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识(S206);获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端(S208)。能够提高数据产品推荐效率及针对性。

Description

数据产品推荐方法、装置、计算机设备和存储介质
本申请要求于2019年6月17日提交中国专利局、申请号为201910522016.1,发明名称为“数据产品推荐方法、装置、计算机设备和存储介质”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。
技术领域
本申请涉及大数据技术领域,特别是涉及一种数据产品推荐方法、装置、计算机设备和存储介质。
背景技术
数据产品是可以发挥数据价值去辅助用户更优的做决策(甚至行动)的一种产品形式。比如:流量分析产品,销售分析产品,个性化推荐产品等等。在互联网高速发展的今天,每个企业都有属于自己的客户群体,企业通常需要对自己的客户群体进行分析从而更好的服务客户。目前,企业会根据自己的需求购买已有的数据产品进行客户数据分析,从而节省开发成本。然而发明人意识到市场的数据产品数量众多,如何高效快速的找到适合自己客户群体的数据产品是一个企业在购买数据产品时遇到的难题。
发明内容
基于此,有必要针对上述技术问题,提供一种能够提高数据产品推荐效率和针对性的数据产品推荐方法、装置、计算机设备和存储介质。
一种数据产品推荐方法,所述方法包括:
接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;
根据目标客户群体信息建立目标特征矩阵;
获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;
获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对 应的目标终端。
在其中一个实施例中,在接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标客户群体信息之前,还包括:
获取各个历史目标标识对应的历史目标客户群体信息,将历史目标客户群体信息预处理,得到历史目标客户群体各个属性值;
计算历史目标客户群体各个属性的数值特征,生成数值特征矩阵;
对数值特征矩阵归一化,得到各个历史目标标识对应的历史目标特征矩阵。
一种数据产品推荐装置,该装置包括:
指令接收模块,用于接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;
矩阵建立模块,用于根据目标客户群体信息建立目标特征矩阵;
标识确定模块,用于获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;
推荐模块,用于获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
一种计算机设备,包括存储器和处理器,所述存储器存储有计算机程序,所述处理器执行所述计算机程序时实现以下步骤:
接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;
根据目标客户群体信息建立目标特征矩阵;
获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;
获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
一种计算机可读存储介质,其上存储有计算机程序,所述计算机程序被处理器执行时实现以下步骤:
接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识 对应的目标客户群体信息;
根据目标客户群体信息建立目标特征矩阵;
获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;
获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
上述数据产品推荐方法、装置、计算机设备和存储介质,通过计算目标标识客户群体的目标特征矩阵与历史目标标识客户群体的历史特征矩阵的相似度,从而确定目标标识客户群体最相似的历史目标标识客户群体,然后根据最相似的历史目标标识客户群体的历史目标标识已购买的数据产品进行推荐,即将具有相同客户群体的历史目标标识购买的数据产品推荐给目标标识,提高了数据产品推荐的效率和针对性。
附图说明
图1为一个实施例中数据产品推荐方法的应用场景图;
图2为一个实施例中数据产品推荐方法的流程示意图;
图3为一个实施例中得到历史目标特征矩阵步骤的流程示意图;
图4为一个具体的实施例中得到历史合作机构企业特征矩阵的示意图;
图5为一个实施例中得到目标特征矩阵的流程示意图
图6为一个实施例中数据产品推荐装置的结构框图;
图7为一个实施例中计算机设备的内部结构图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
本申请提供的数据产品推荐方法,可以应用于如图1所示的应用环境中。其中,终端102通过网络与服务器104进行通信。服务器104接收目标终端102 发送的数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;服务器104根据目标客户群体信息建立目标特征矩阵;服务器104获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;服务器104获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端104。其中,终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备,服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
在一个实施例中,如图2所示,提供了一种数据产品推荐方法,以该方法应用于图1中的服务器为例进行说明,包括以下步骤:
S202,接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息。
其中,目标标识是指企业标识,用于唯一标识要购买数据产品的企业。目标客户群体信息是指企业客户群体信息,是该要购买数据产品的企业所服务的客户群体的详细信息,该客户群体信息中可以包括多个属性,比如,客户群体偏好属性、客户群体年龄属性和客户群体地域属性等等。
具体地,服务器接收到企业终端发送的数据产品推荐指令,该数据产品推荐指令中携带有企业标识和企业标识对应的企业客户群体信息。
S204,根据目标客户群体信息建立目标特征矩阵。
其中,目标特征矩阵是指根据企业客户群体信息建立的数值特征矩阵。
具体地,服务器根据目标客户群体信息得到该目标客户群体的各个属性的信息。根据目标客户群体中各个属性的信息计算各个属性对应的数值特征向量,该数值特征是指预先设置好的要计算的数学特征,比如期望等等。此时,可以得到目标客户群体各个属性的数值特征向量,根据目标客户群体属性的数值特征向量得到数值特征矩阵。比如目标标识的企业对应的客户群体A,该客户群体A有属性1、属性2和属性3,根据客户群体A属性1的客户信息计算属性1的数值特征向量(A11)。根据客户群体A属性2的客户信息计算属性2的数值特征向量(A21)。根据客户群体A属性3的客户信息计算属性2的数值特征向 量(A31)。则得到的目标标识对应的企业的数值特征矩阵为[(A11),(A21),(A31)]
S206,获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识。
其中,历史目标标识是指历史购买过数据产品企业的企业标识,历史特征矩阵是指预先根据历史企业标识对应各个企业的客户群体的各个属性的信息建立的历史数值特征矩阵。该历史数值特征矩阵中行表示历史企业标识对应的企业,列表示该历史企业的客户群体的属性对应的数值特征。当有M个历史企业标识对应的企业,且每个企业对应的客户群体有N个属性时,该历史数值特征矩阵有M行N列,即得到的历史数值特征矩阵可以是
Figure PCTCN2020086060-appb-000001
B表示该历史特征矩阵。B mn表示第m企业的第n个属性的数值特征向量。
具体地,服务器获取到各个历史企业标识对应的历史特征矩阵,分别计算历史特征矩阵和目标特征矩阵的相似度,即计算历史特征矩阵中各个客户群体属性的数值特征向量与目标特征矩阵中客户群体属性的数值特征向量之间的相似性。该相似度用于表示各个历史企业的客户群体和目标企业的客户群体之间的相似度,将计算得到的历史特征矩阵中各个客户群体与目标特征矩阵中客户群体之间的相似度进行排序,获取排序结果中最大相似度对应的历史特征矩阵中的历史客户群体,获取到该历史客户群体对应的历史企业标识,得到企业标识最相似的历史企业标识,即得到了要购买数据产品的企业最相似的历史企业。其中,使用余弦相似性算法来计算客户群体之间的相似度。根据历史企业特征矩阵得到客群的特征向量,根据企业特征矩阵得到客群的特征向量,根据余弦相似度计算公式计算客群特征向量之间的相似度。
S208,获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
具体地,服务器获取到已保存的历史企业标识对应的企业已购买的数据产 品信息,历史企业已购买的数据产品信息发送到对应的要购买数据产品的企业终端,企业终端接收到历史企业已购买的数据产品信息并进行显示,以使企业用户对显示的数据产品进行选择购买。
在上述数据产品推荐方法中,通过接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息,根据目标客户群体信息建立目标特征矩阵,获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识,获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。通过根据企业的客户群体信息的相似度来找到相似度的客户群体,进而获取到相似客户群体的企业购买的数据产品进行推荐,提高了数据产品推荐的效率和针对性。
在一个实施例中,如图3所示,在步骤S202之前,即在接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标客户群体信息之前,还包括步骤:
S302,获取各个历史目标标识对应的历史目标客户群体信息,将历史目标客户群体信息预处理,得到历史目标客户群体各个属性值。
其中,预处理是指将历史目标客户群体信息中的数据清洗、数据转换和数据标准化等等。
具体地,服务器获取各个历史企业标识对应的历史企业客户群体信息,将历史企业客户群体信息预处理,得到历史客户群体各个属性值。比如,讲历史客户群体中客户的性别属性进行数据转换,转换为性别男为数值1,性别女为数值0。等等。
S304,计算历史目标客户群体各个属性的数值特征,生成数值特征矩阵。
具体地,服务器计算历史目标客户群体各个属性的数值特征,包括期望值Ex、熵En、超熵He,根据该计算得到的数字特征生成数值特征矩阵。
S306,对数值特征矩阵归一化,得到各个历史目标标识对应的历史目标特征矩阵。
具体地,将数值特征矩阵归一化处理,将数值特征矩阵中的数值进行线性 变换,将结果值映射到0到1之间。即得到各个历史目标标识对应的历史目标特征矩阵。其中,归一化可以使用min-max标准化或者Z-score标准化方法来进行计算。在一个具体的实施例中,如图4所示,M个历史合作机构企业对应有M个合作机构企业客户群体,每个客户群体都有N个属性,计算每个属性的数值特征包括期望、熵和超熵,得到M*3N阶矩阵。其中,M表示有M个客群,3N表示N个属性的期望、熵和超熵。然后将得到的矩阵进行归一化处理,得到历史合作机构企业特征矩阵。
上述实施例中,通过获取各个历史目标标识对应的历史目标客户群体信息,将历史目标客户群体信息预处理,得到历史目标客户群体各个属性值,计算历史目标客户群体各个属性的数值特征,生成数值特征矩阵,对数值特征矩阵归一化,得到各个历史目标标识对应的历史目标特征矩阵,通过预先计算得到历史目标特征矩阵,在进行客群相似度计算时可以直接进行使用,方便快捷。
在一个实施例中,如图5所示,步骤S204,即根据目标客户群体信息建立目标特征矩阵,包括步骤:
S502,将目标客户群体信息预处理,得到目标客户群体各个属性值。
具体地,服务器将企业客户群体信息预处理,包括数据清洗、数据转换和数据标准化等等,得到企业客户群体信息各个属性数值。
S506,计算目标客户群体各个属性的数值特征,生成数值特征矩阵。
具体地,服务器计算企业客户群体各个属性是数值特征包括期望、熵和超熵,生成该企业客户群体的数值特征矩阵。例如,该企业客户群体有N个属性,则生成的特征矩阵为T=[(Ex 11,En 11,He 11)(Ex 11,En 11,He 11)...(Ex 1n,En 1n,He 1n)]。比如,客户群体的年龄属性,该客户群体的中客户的年龄可以是18、24、26、20等等。计算出该年龄属性的数值特征即期望、熵和超熵,来表征该年龄属性的数值特征。
S507,对数值特征矩阵归一化,得到目标标识对应的目标特征矩阵。
具体地,服务器对生成的数值特征矩阵进行归一化处理,即将数值特征矩阵中的数值进行线性变换,将结果值映射到0到1之间,得到企业标识对应的企业特征矩阵。
在上述实施例中,通过将目标客户群体信息预处理,得到目标客户群体各个属性值,计算目标客户群体各个属性的数值特征,生成数值特征矩阵,对数值特征矩阵归一化,得到目标标识对应的目标特征矩阵,能够计算得到客户群体的特征矩阵,方便后续使用。
在一个实施例中,步骤S502,即将目标客户群体信息预处理,得到目标客户群体各个属性值,包括步骤:
获取目标客户群体信息中的非数值属性信息,按照预设规则将非数值属性信息转换为数值属性信息,得到目标客户群体各个属性值。
具体地,服务器获取到企业客户群体信息中的属性信息是非数值的属性信息,比如,企业客户群体的性别属性、偏好属性、教育属性和职业属性等等。预设规则是指预先设置好的非数值属性和数值的映射关系,比如将性别属性中的男映射为数值1,女映射为数值0等等。即将目标客户群体中各个属性都转换为属性数值,得到目标客户群体各个属性值,便于后续计算数值特征,提高计算效率。
在一个实施例中,步骤S208,即获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端,包括步骤:
获取目标标识对应的已购数据产品信息,当历史目标标识对应的数据产品信息中未存在已购数据产品信息,则将数据产品信息发送目标标识对应的目标终端。
具体地,服务器获取到企业标识对应的企业已经购买的数据产品信息,该数据产品信息可以是产品名称和产品标识等等。判断该企业已经购买的数据产品信息是否在推荐的数据产品信息中,即可以根据产品名称进行判断,判断已购买的产品名称是否在推荐的产品名称列表中。当推荐的数据产品信息中没有企业已经购买的数据产品信息时,将推荐的数据产品信息发送该企业标识对应的企业终端中进行显示。当推荐的数据产品信息中有企业已经购买的数据产品信息时,将除企业已经购买的数据产品信息以外的推荐的数据产品信息发送该企业标识对应的企业终端中进行显示,提高推荐效率,能够防止推荐重复,便于企业选择。
在一个实施例中,步骤S208,获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端,包括步骤:
根据数据产品信息获取对应的评价信息,根据评价信息计算评价分值,将评价分值大于预设阈值的数据产品信息发送目标标识对应的目标终端。
具体地,服务器根据要推荐的数据产品信息获取各个数据产品的评价信息,根据评价信息计算评价分值,比如,可以通过关键字提取算法提取评价信息中的关键字,通过预先设置好的关键字与分值的对应关系计算得到评价信息的评价分值。然后,服务器根据计算得到的评价分值,将评价分值大于预设阈值的数据产品信息发送企业标识对应的企业终端中进行显示。
在一个实施例中,根据评价信息计算评价分值,包括步骤:
根据评价信息建立对应的评价向量,将评价向量输入到已训练的数据产品评价模型中,得到数据产品评价模型的输出结果向量,根据数据产品评价模型的输出结果向量得到评价分值。
其中,已训练的数据产品评价模型是指是使用历史数据产品评价信息和历史评价分值使用线性回归算法进行训练得到的。
具体地,服务器根据评价信息建立对应的评价向量,根据评价信息中的评价等级和评价关键字等根据映射关系生成对应的评价向量,比如,若评价等级包括差,良和优三个等级,该评价等级为良,则对应的评价向量为(0,1,0)。将得到的评价向量输入到已训练的数据产品评价模型中,得到数据产品评价模型的输出结果向量,根据数据产品评价模型的输出结果向量得到评价分值。使用数据产品评价模型得到评价分值,能够提高评价分支的精确性和准确性。
应该理解的是,虽然图2-3和5的流程图中的各个步骤按照箭头的指示依次显示,但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明,这些步骤的执行并没有严格的顺序限制,这些步骤可以以其它的顺序执行。而且,图2-3和5中的至少一部分步骤可以包括多个子步骤或者多个阶段,这些子步骤或者阶段并不必然是在同一时刻执行完成,而是可以在不同的时刻执行,这些子步骤或者阶段的执行顺序也不必然是依次进行,而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执 行。
在一个实施例中,如图6所示,提供了一种数据产品推荐装置600,包括:指令接收模块602、矩阵建立模块604、标识确定模块606和推荐模块608,其中:
指令接收模块602,用于接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;
矩阵建立模块604,用于根据目标客户群体信息建立目标特征矩阵;
标识确定模块606,用于获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;
推荐模块608,用于获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
在一个实施例中,数据产品推荐装置600,还包括:
历史信息得到模块,用于获取各个历史目标标识对应的历史目标客户群体信息,将历史目标客户群体信息预处理,得到历史目标客户群体各个属性值;
历史计算模块,用于计算历史目标客户群体各个属性的数值特征,生成数值特征矩阵;
历史矩阵得到模块,用于对数值特征矩阵归一化,得到各个历史目标标识对应的历史目标特征矩阵。
在一个实施例中,矩阵建立模块604,包括:
信息得到模块,将目标客户群体信息预处理,得到目标客户群体各个属性值;
计算模块,用于计算目标客户群体各个属性的数值特征,生成数值特征矩阵;
矩阵得到模块,用于对数值特征矩阵归一化,得到目标标识对应的目标特征矩阵。
在一个实施例中,信息得到模块,包括:
信息转换模块,用于获取目标客户群体信息中的非数值属性信息,按照预 设规则将非数值属性信息转换为数值属性信息,得到目标客户群体各个属性值。
在一个实施例中,推荐模块608,包括:
信息判断模块,用于获取目标标识对应的已购数据产品信息,当历史目标标识对应的数据产品信息中未存在已购数据产品信息,则将数据产品信息发送目标标识对应的目标终端。
在一个实施例中,推荐模块608,包括:
评价分值计算模块,用于根据数据产品信息获取对应的评价信息,根据评价信息计算评价分值,将评价分值大于预设阈值的数据产品信息发送目标标识对应的目标终端。
在一个实施例中,评价分值计算模块,包括:
模块计算模块,用于根据评价信息建立对应的评价向量,将评价向量输入到已训练的数据产品评价模型中,得到数据产品评价模型的输出结果向量,根据数据产品评价模型的输出结果向量得到评价分值
关于数据产品推荐装置的具体限定可以参见上文中对于数据产品推荐方法的限定,在此不再赘述。上述数据产品推荐装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中,也可以以软件形式存储于计算机设备中的存储器中,以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中,提供了一种计算机设备,该计算机设备可以是服务器,其内部结构图可以如图7所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中,该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的数据库用于存储数据产品数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种数据产品推荐方法。
本领域技术人员可以理解,图7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的限定, 具体的计算机设备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。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;根据目标客户群体信息建立目标特征矩阵;获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:获取各个历史目标标识对应的历史目标客户群体信息,将历史目标客户群体信息预处理,得到历史目标客户群体各个属性值;计算历史目标客户群体各个属性的数值特征,生成数值特征矩阵;对数值特征矩阵归一化,得到各个历史目标标识对应的历史目标特征矩阵。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:将目标客户群体信息预处理,得到目标客户群体各个属性值;计算目标客户群体各个属性的数值特征,生成数值特征矩阵;对数值特征矩阵归一化,得到目标标识对应的目标特征矩阵。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:获取目标客户群体信息中的非数值属性信息,按照预设规则将非数值属性信息转换为数值属性信息,得到目标客户群体各个属性值。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:获取目标标识对应的已购数据产品信息,当历史目标标识对应的数据产品信息中未存在已购数据产品信息,则将数据产品信息发送目标标识对应的目标终端。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:根据数据产品信息获取对应的评价信息,根据评价信息计算评价分值,将评价分值大于预设阈值的数据产品信息发送目标标识对应的目标终端。
在一个实施例中,处理器执行计算机程序时还实现以下步骤:根据评价信息建立对应的评价向量,将评价向量输入到已训练的数据产品评价模型中,得 到数据产品评价模型的输出结果向量,根据数据产品评价模型的输出结果向量得到评价分值。
在一个实施例中,提供了一种计算机可读存储介质,该存储介质为易失性存储介质或非易失性存储介质,其上存储有计算机程序,计算机程序被处理器执行时实现以下步骤:接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;根据目标客户群体信息建立目标特征矩阵;获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
在一个实施例中,计算机程序被处理器执行时还实现以下步骤:获取各个历史目标标识对应的历史目标客户群体信息,将历史目标客户群体信息预处理,得到历史目标客户群体各个属性值;计算历史目标客户群体各个属性的数值特征,生成数值特征矩阵;对数值特征矩阵归一化,得到各个历史目标标识对应的历史目标特征矩阵。
在一个实施例中,计算机程序被处理器执行时还实现以下步骤:将目标客户群体信息预处理,得到目标客户群体各个属性值;计算目标客户群体各个属性的数值特征,生成数值特征矩阵;
对数值特征矩阵归一化,得到目标标识对应的目标特征矩阵。
在一个实施例中,计算机程序被处理器执行时还实现以下步骤:获取目标客户群体信息中的非数值属性信息,按照预设规则将非数值属性信息转换为数值属性信息,得到目标客户群体各个属性值。
在一个实施例中,计算机程序被处理器执行时还实现以下步骤:获取目标标识对应的已购数据产品信息,当历史目标标识对应的数据产品信息中未存在已购数据产品信息,则将数据产品信息发送目标标识对应的目标终端。
在一个实施例中,计算机程序被处理器执行时还实现以下步骤:根据数据产品信息获取对应的评价信息,根据评价信息计算评价分值,将评价分值大于预设阈值的数据产品信息发送目标标识对应的目标终端。
在一个实施例中,计算机程序被处理器执行时还实现以下步骤:根据评价信息建立对应的评价向量,将评价向量输入到已训练的数据产品评价模型中,得到数据产品评价模型的输出结果向量,根据数据产品评价模型的输出结果向量得到评价分值。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机程序来指令相关的硬件来完成,所述的计算机程序可存储于一非易失性计算机可读取存储介质中,该计算机程序在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术特征所有可能的组合都进行描述,然而,只要这些技术特征的组合不存在矛盾,都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式,其描述较为具体和详细,但并不能因此而理解为对发明专利范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本申请构思的前提下,还可以做出若干变形和改进,这些都属于本申请的保护范围。因此,本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种数据产品推荐方法,其中,所述方法包括:
    接收数据产品推荐指令,所述数据产品推荐指令中携带有目标标识和所述目标标识对应的目标客户群体信息;
    根据所述目标客户群体信息建立目标特征矩阵;
    获取各个历史目标标识对应的历史特征矩阵,计算所述历史特征矩阵和所述目标特征矩阵的相似度,根据所述相似度确定所述目标标识对应的历史目标标识;
    获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端。
  2. 根据权利要求1所述的方法,其中,在所述接收数据产品推荐指令,所述数据产品推荐指令中携带有目标标识和目标客户群体信息之前,还包括:
    获取各个历史目标标识对应的历史目标客户群体信息,将所述历史目标客户群体信息预处理,得到历史目标客户群体各个属性值;
    计算所述历史目标客户群体各个属性的数值特征,生成数值特征矩阵;
    对所述数值特征矩阵归一化,得到所述各个历史目标标识对应的历史目标特征矩阵。
  3. 根据权利要求1所述的方法,其中,所述根据所述目标客户群体信息建立目标特征矩阵,包括:
    将所述目标客户群体信息预处理,得到目标客户群体各个属性值;
    计算所述目标客户群体各个属性的数值特征,生成数值特征矩阵;
    对所述数值特征矩阵归一化,得到所述目标标识对应的目标特征矩阵。
  4. 根据权利要求3所述的方法,其中,所述将所述目标客户群体信息预处理,得到目标客户群体各个属性值,包括:
    获取所述目标客户群体信息中的非数值属性信息,按照预设规则将所述非数值属性信息转换为数值属性信息,得到所述目标客户群体各个属性值。
  5. 根据权利要求1所述的方法,其中,所述获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端,包括:
    获取所述目标标识对应的已购数据产品信息,当所述历史目标标识对应的数据产品信息中未存在所述已购数据产品信息,则将所述数据产品信息发送所述目标标识对应的目标终端。
  6. 根据权利要求1所述的方法,其中,所述获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端,包括:
    根据所述数据产品信息获取对应的评价信息,根据所述评价信息计算评价分值,将所述评价分值大于预设阈值的数据产品信息发送所述目标标识对应的目标终端。
  7. 根据权利要求6所述的方法,其中,所述根据所述评价信息计算评价分 值,包括:
    根据所述评价信息建立对应的评价向量,将所述评价向量输入到已训练的数据产品评价模型中,得到所述数据产品评价模型的输出结果向量,根据所述数据产品评价模型的输出结果向量得到评价分值。
  8. 一种数据产品推荐装置,其中,所述装置包括:
    指令接收模块,用于接收数据产品推荐指令,所述数据产品推荐指令中携带有目标标识和所述目标标识对应的目标客户群体信息;
    矩阵建立模块,用于根据所述目标客户群体信息建立目标特征矩阵;
    标识确定模块,用于获取各个历史目标标识对应的历史特征矩阵,计算所述历史特征矩阵和所述目标特征矩阵的相似度,根据所述相似度确定所述目标标识对应的历史目标标识;
    推荐模块,用于获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端。
  9. 一种计算机设备,包括存储器和处理器,所述存储器存储有计算机程序,其中,所述处理器执行所述计算机程序时实现以下步骤:接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;根据目标客户群体信息建立目标特征矩阵;获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
  10. 根据权利要求9所述的计算机设备,其中,在所述接收数据产品推荐指令,所述数据产品推荐指令中携带有目标标识和目标客户群体信息之前,还包括:
    获取各个历史目标标识对应的历史目标客户群体信息,将所述历史目标客户群体信息预处理,得到历史目标客户群体各个属性值;
    计算所述历史目标客户群体各个属性的数值特征,生成数值特征矩阵;
    对所述数值特征矩阵归一化,得到所述各个历史目标标识对应的历史目标特征矩阵。
  11. 根据权利要求9所述的计算机设备,其中,所述根据所述目标客户群体信息建立目标特征矩阵,包括:
    将所述目标客户群体信息预处理,得到目标客户群体各个属性值;
    计算所述目标客户群体各个属性的数值特征,生成数值特征矩阵;
    对所述数值特征矩阵归一化,得到所述目标标识对应的目标特征矩阵。
  12. 根据权利要求11所述的计算机设备,其中,所述将所述目标客户群体信息预处理,得到目标客户群体各个属性值,包括:
    获取所述目标客户群体信息中的非数值属性信息,按照预设规则将所述非数值属性信息转换为数值属性信息,得到所述目标客户群体各个属性值。
  13. 根据权利要求9所述的计算机设备,其中,所述获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端,包括:
    获取所述目标标识对应的已购数据产品信息,当所述历史目标标识对应的数据产品信息中未存在所述已购数据产品信息,则将所述数据产品信息发送所述目标标识对应的目标终端。
  14. 根据权利要求9所述的计算机设备,其中,所述获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端,包括:
    根据所述数据产品信息获取对应的评价信息,根据所述评价信息计算评价分值,将所述评价分值大于预设阈值的数据产品信息发送所述目标标识对应的目标终端。
  15. 根据权利要求14所述的计算机设备,其中,所述根据所述评价信息计算评价分值,包括:
    根据所述评价信息建立对应的评价向量,将所述评价向量输入到已训练的数据产品评价模型中,得到所述数据产品评价模型的输出结果向量,根据所述数据产品评价模型的输出结果向量得到评价分值。
  16. 一种计算机可读存储介质,其上存储有计算机程序,其中,所述计算机程序被处理器执行时实现以下步骤:
    接收数据产品推荐指令,数据产品推荐指令中携带有目标标识和目标标识对应的目标客户群体信息;根据目标客户群体信息建立目标特征矩阵;获取各个历史目标标识对应的历史特征矩阵,计算历史特征矩阵和目标特征矩阵的相似度,根据相似度确定目标标识对应的历史目标标识;获取历史目标标识对应的数据产品信息,将数据产品信息发送目标标识对应的目标终端。
  17. 根据权利要求16所述的计算机可读存储介质,其中,所述根据所述目标客户群体信息建立目标特征矩阵,包括:
    将所述目标客户群体信息预处理,得到目标客户群体各个属性值;
    计算所述目标客户群体各个属性的数值特征,生成数值特征矩阵;
    对所述数值特征矩阵归一化,得到所述目标标识对应的目标特征矩阵。
  18. 根据权利要求16所述的计算机可读存储介质,其中,所述将所述目标客户群体信息预处理,得到目标客户群体各个属性值,包括:
    获取所述目标客户群体信息中的非数值属性信息,按照预设规则将所述非数值属性信息转换为数值属性信息,得到所述目标客户群体各个属性值。
  19. 根据权利要求17所述的计算机可读存储介质,其中,所述获取所述历 史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端,包括:
    获取所述目标标识对应的已购数据产品信息,当所述历史目标标识对应的数据产品信息中未存在所述已购数据产品信息,则将所述数据产品信息发送所述目标标识对应的目标终端。
  20. 根据权利要求16所述的计算机可读存储介质,其中,所述获取所述历史目标标识对应的数据产品信息,将所述数据产品信息发送所述目标标识对应的目标终端,包括:
    根据所述数据产品信息获取对应的评价信息,根据所述评价信息计算评价分值,将所述评价分值大于预设阈值的数据产品信息发送所述目标标识对应的目标终端;
    其中,所述根据所述评价信息计算评价分值,包括:
    根据所述评价信息建立对应的评价向量,将所述评价向量输入到已训练的数据产品评价模型中,得到所述数据产品评价模型的输出结果向量,根据所述数据产品评价模型的输出结果向量得到评价分值。
PCT/CN2020/086060 2019-06-17 2020-04-22 数据产品推荐方法、装置、计算机设备和存储介质 WO2020253357A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910522016.1A CN110363569A (zh) 2019-06-17 2019-06-17 数据产品推荐方法、装置、计算机设备和存储介质
CN201910522016.1 2019-06-17

Publications (1)

Publication Number Publication Date
WO2020253357A1 true WO2020253357A1 (zh) 2020-12-24

Family

ID=68216232

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/086060 WO2020253357A1 (zh) 2019-06-17 2020-04-22 数据产品推荐方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN110363569A (zh)
WO (1) WO2020253357A1 (zh)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110363569A (zh) * 2019-06-17 2019-10-22 深圳壹账通智能科技有限公司 数据产品推荐方法、装置、计算机设备和存储介质
CN112836115A (zh) * 2019-11-25 2021-05-25 浙江大搜车软件技术有限公司 信息推荐方法、装置、计算机设备和存储介质
CN111049894B (zh) * 2019-12-07 2023-08-29 深圳市万佳安物联科技股份有限公司 云端物联网数据处理方法、装置以及电子设备
CN111178949B (zh) * 2019-12-18 2023-07-28 中电金信软件有限公司 服务资源匹配参考数据确定方法、装置、设备和存储介质
CN111191129A (zh) * 2019-12-27 2020-05-22 广东德诚科教有限公司 专业推荐方法、装置、计算机设备和存储介质
CN114186024A (zh) * 2021-12-14 2022-03-15 中国建设银行股份有限公司 推荐方法及装置
CN114429384B (zh) * 2021-12-30 2022-12-09 杭州盟码科技有限公司 基于电商平台的产品智能推荐方法及系统
CN115809373B (zh) * 2023-02-06 2023-06-09 一智科技有限公司 一种智能推荐方法、系统和存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109102332A (zh) * 2018-08-06 2018-12-28 北京小度信息科技有限公司 一种数据处理的方法、装置和电子设备
CN109325883A (zh) * 2017-07-31 2019-02-12 深圳市珍爱网信息技术有限公司 一种双向匹配的推荐方法及装置
CN109859004A (zh) * 2019-01-10 2019-06-07 珠海金山网络游戏科技有限公司 一种基于历史数据的商品推荐方法及系统
CN110363569A (zh) * 2019-06-17 2019-10-22 深圳壹账通智能科技有限公司 数据产品推荐方法、装置、计算机设备和存储介质

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109408665B (zh) * 2018-12-29 2021-11-23 咪咕音乐有限公司 一种信息推荐方法及装置、存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109325883A (zh) * 2017-07-31 2019-02-12 深圳市珍爱网信息技术有限公司 一种双向匹配的推荐方法及装置
CN109102332A (zh) * 2018-08-06 2018-12-28 北京小度信息科技有限公司 一种数据处理的方法、装置和电子设备
CN109859004A (zh) * 2019-01-10 2019-06-07 珠海金山网络游戏科技有限公司 一种基于历史数据的商品推荐方法及系统
CN110363569A (zh) * 2019-06-17 2019-10-22 深圳壹账通智能科技有限公司 数据产品推荐方法、装置、计算机设备和存储介质

Also Published As

Publication number Publication date
CN110363569A (zh) 2019-10-22

Similar Documents

Publication Publication Date Title
WO2020253357A1 (zh) 数据产品推荐方法、装置、计算机设备和存储介质
WO2021082426A1 (zh) 人脸聚类方法、装置、计算机设备及存储介质
CN111325353A (zh) 训练数据集的贡献度计算方法、装置、设备及存储介质
US10565525B2 (en) Collaborative filtering method, apparatus, server and storage medium in combination with time factor
US20150193548A1 (en) Recommendation System With Metric Transformation
WO2020125463A1 (zh) 报表生成方法、装置、计算机设备及可读存储介质
TW201917601A (zh) 使用者意圖識別方法及裝置
WO2020053854A1 (en) Systems and methods for secure prediction using an encrypted query executed based on encrypted data
CN111178949B (zh) 服务资源匹配参考数据确定方法、装置、设备和存储介质
US20170140417A1 (en) Campaign Effectiveness Determination using Dimension Reduction
EP3857468A1 (en) Recommendation method and system and method and system for improving a machine learning system
CN107291774B (zh) 错误样本识别方法和装置
CN115082209A (zh) 业务数据风险预警方法、装置、计算机设备及存储介质
CN112712086A (zh) 数据处理方法、装置、计算机设备和存储介质
CN110457361B (zh) 特征数据获取方法、装置、计算机设备和存储介质
US20210357553A1 (en) Apparatus and method for option data object performance prediction and modeling
CN108876422B (zh) 用于信息推广的方法、装置、电子设备及计算机可读介质
US20190156529A1 (en) User interface and method based on sliding-scale cluster groups for precise look-alike modeling
CN111291795A (zh) 人群特征分析方法、装置、存储介质和计算机设备
CN116861107A (zh) 业务内容展示方法、装置、设备、介质和产品
CN115758271A (zh) 数据处理方法、装置、计算机设备和存储介质
CN114969550A (zh) 业务推荐方法、装置、计算机设备和存储介质
WO2020247960A1 (en) Method and apparatus for cosmetic product recommendation
CN111859191A (zh) Gis服务聚合方法、装置、计算机设备和存储介质
CN113450169A (zh) 车辆推荐信息的处理方法、系统、计算机设备和存储介质

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: 20827003

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: 20827003

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 31/03/2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20827003

Country of ref document: EP

Kind code of ref document: A1