CN113570437A - Product recommendation method and device - Google Patents
Product recommendation method and device Download PDFInfo
- Publication number
- CN113570437A CN113570437A CN202110869207.2A CN202110869207A CN113570437A CN 113570437 A CN113570437 A CN 113570437A CN 202110869207 A CN202110869207 A CN 202110869207A CN 113570437 A CN113570437 A CN 113570437A
- Authority
- CN
- China
- Prior art keywords
- customer
- rule
- data
- frequent item
- determining
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
- G06F16/24564—Applying rules; Deductive queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
Abstract
The invention provides a product recommendation method and a product recommendation device, which relate to big data technology and comprise the following steps: acquiring historical purchase data of a customer for preprocessing, and determining a customer data set; establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation; determining a frequent item set according to a client data set, support degree and confidence coefficient parameters; determining a rule set according to the frequent item set; and determining a recommended product list according to the rule set and the customer data set. The recommendation technology based on the association rule realizes personalized recommendation aiming at different users and accurate recommendation.
Description
Technical Field
The invention relates to the technical field of computer data processing, in particular to a product recommendation method and device.
Background
This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.
From the technical point of view, the existing recommendation technologies are simply divided into two categories, namely a method based on machine learning and a method based on deep learning, the method based on machine learning is not beneficial to processing heterogeneous data, the deep learning can process a large amount of heterogeneous data, but needs a large amount of training data sets, and the result is mostly undecipherable.
The recommendation of the client by the existing recommendation technology may depend on the misoperation of the client, such as panning, when the client clicks a certain product carelessly, the following recommended product is still updated according to the current misoperation, and obviously, the recommendation technology is a poor client experience. The invention uses the actual ordering data of the client, such as the purchase record of the financial products of the client, and does not depend on the click data of the client.
Conventional recommendation techniques may be inaccurate or even useless to the user because the training sample set may contain a lot of mishandling data or unreal data, such as a user browsing a financial product but not intending to buy, which may be of little reference to most users, but may weigh the feature heavily due to the training of the model itself. This may result in ineffective recommendations and a poor experience for the user.
Therefore, how to provide a new solution, which can solve the above technical problems, is a technical problem to be solved in the art.
Disclosure of Invention
The embodiment of the invention provides a product recommendation method, which is based on a recommendation technology of association rules, realizes personalized recommendation aiming at different users and realizes accurate recommendation, and comprises the following steps:
acquiring historical purchase data of a customer for preprocessing, and determining a customer data set;
establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation;
determining a frequent item set according to a client data set, support degree and confidence coefficient parameters;
determining a rule set according to the frequent item set;
and determining a recommended product list according to the rule set and the customer data set.
An embodiment of the present invention further provides a product recommendation device, including:
the preprocessing module is used for acquiring historical purchase data of a customer for preprocessing and determining a customer data set;
the support degree and confidence coefficient parameter determining module is used for establishing an evaluation function and determining support degree and confidence coefficient parameters through iterative calculation;
the frequent item set determining module is used for determining a frequent item set according to the client data set, the support degree and the confidence coefficient parameters;
the rule set determining module is used for determining a rule set according to the frequent item set;
and the product recommendation module is used for determining a recommended product list according to the rule set and the customer data set.
The embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the product recommendation method is implemented.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program for executing the above product recommendation method is stored in the computer-readable storage medium.
The embodiment of the invention provides a product recommendation method and device, which comprise the following steps: firstly, acquiring historical purchase data of a customer for preprocessing, and determining a customer data set; then establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation; then, determining a frequent item set according to the client data set, the support degree and the confidence coefficient parameters; determining a rule set according to the frequent item set; and finally, determining a recommended product list according to the rule set and the customer data set. The embodiment of the invention realizes the recommendation technology based on the association rule, has the core idea of mining the rule set of the strong association rule, is an interpretable recommendation method, solves the defects of most of the existing recommendation methods, and realizes the personalized recommendation aiming at different users and the accurate recommendation. The union rule mining technology is a method in machine learning, but the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users. According to the invention, the actual purchase data of the user is utilized, so that the influence of useless data on the recommendation result is reduced, and the experience of the user is enhanced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts. In the drawings:
fig. 1 is a schematic diagram of a product recommendation method according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of a data preprocessing process of a product recommendation method according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of a calculation process of support and confidence parameters of a product recommendation method according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of a rule set determination process of a product recommendation method according to an embodiment of the present invention.
Fig. 5 is a schematic diagram of a recommended product list determination process of a product recommendation method according to an embodiment of the present invention.
FIG. 6 is a block diagram of a modular example interaction flow of a product recommendation method in accordance with an embodiment of the present invention.
FIG. 7 is a schematic diagram of a computer device for executing a product recommendation method according to the present invention.
Fig. 8 is a schematic diagram of a product recommendation device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. The exemplary embodiments and descriptions of the present invention are provided to explain the present invention, but not to limit the present invention.
The invention relates to big data technology. Fig. 1 is a schematic diagram of a product recommendation method according to an embodiment of the present invention, and as shown in fig. 1, an embodiment of the present invention provides a product recommendation method, which implements personalized recommendation for different users based on a recommendation technology of association rules, and implements accurate recommendation, and the method includes:
step 101: acquiring historical purchase data of a customer for preprocessing, and determining a customer data set;
step 102: establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation;
step 103: determining a frequent item set according to a client data set, support degree and confidence coefficient parameters;
step 104: determining a rule set according to the frequent item set;
step 105: and determining a recommended product list according to the rule set and the customer data set.
The product recommendation method provided by the embodiment of the invention comprises the following steps: firstly, acquiring historical purchase data of a customer for preprocessing, and determining a customer data set; then establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation; then, determining a frequent item set according to the client data set, the support degree and the confidence coefficient parameters; determining a rule set according to the frequent item set; and finally, determining a recommended product list according to the rule set and the customer data set. The embodiment of the invention realizes the recommendation technology based on the association rule, has the core idea of mining the rule set of the strong association rule, is an interpretable recommendation method, solves the defects of most of the existing recommendation methods, and realizes the personalized recommendation aiming at different users and the accurate recommendation. The union rule mining technology is a method in machine learning, but the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users. According to the invention, the actual purchase data of the user is utilized, so that the influence of useless data on the recommendation result is reduced, and the experience of the user is enhanced.
In the embodiments of the present invention, the terms referred to are explained as follows:
the support degree is as follows: indicating that the transactions containing both a and B are a proportion of all transactions.
Confidence coefficient: represents the proportion of transactions containing A and B at the same time, i.e. the proportion of transactions containing A and B at the same time is the proportion of transactions containing A.
When the product recommendation method provided by the embodiment of the present invention is implemented specifically, in an embodiment, the method may include:
acquiring historical purchase data of a customer for preprocessing, and determining a customer data set;
establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation;
determining a frequent item set according to a client data set, support degree and confidence coefficient parameters;
determining a rule set according to the frequent item set;
and determining a recommended product list according to the rule set and the customer data set.
The invention provides a product recommendation method based on association rules, which is an interpretable recommendation method, overcomes the defects of most of the existing recommendation methods, realizes personalized recommendation for different users, and realizes accurate recommendation. The embodiment of the invention takes the situation that users buy financial products as an example, and realizes that different investment financial products are recommended to different users.
Marking each operation or action of the user (such as purchasing a financial product) as biFrom this, a historical purchase record B ═ B of each user can be obtained1,b4,...bi,..bnSuppose the user purchases multiple products, product 1, product 4, etc. In a database D of purchase records of a plurality of users, a set of 1, 2,. k items of the users (a set of k items refers to a sequence of k purchase records) is mined, indicating that the set of items is frequent if the set of k items satisfies a predefined minimum degree of support, while strong rules can be generated from the set of items. The product recommendation method of the embodiment of the invention is a recommendation method based on the strong rule.
Fig. 2 is a schematic diagram of a data preprocessing process of a product recommendation method according to an embodiment of the present invention, and as shown in fig. 2, when a product recommendation method according to an embodiment of the present invention is implemented specifically, in an embodiment, the obtaining of historical purchase data of a customer for preprocessing and determining a customer data set includes:
step 201: connecting a customer transaction database to obtain historical purchase data of a customer;
step 202: analyzing historical purchase data of a customer to obtain a purchase record of the customer, marking a product purchased by the customer, and generating customer data;
step 203: and traversing the customer transaction database, and summarizing and counting all the generated customer data into a customer data set.
In an embodiment, the main process of data preprocessing includes: firstly, connecting a customer transaction database to obtain historical purchase data of a customer; then analyzing the historical purchase data of the customer to obtain a purchase record of the customer, marking the product purchased by the customer and generating a piece of customer data; and finally, traversing a customer transaction database, and summarizing and counting all the generated customer data into a customer data set.
In the embodiment, the data preprocessing is to acquire the historical purchase data of the customer by traversing the database; marking the products purchased by the customer according to the purchase record of the customer to generate a piece of customer data, namely: b isi＝{b1,b5,...bj,..bnAnd (i represents i users, and the users purchase a plurality of products such as product 1, product 5 and the like), and after one-time traversal, obtaining data sets of all customers.
The embodiment of the invention uses the actual ordering and purchasing data of the user to generate a strong association rule through the purchasing record of the user in the database, and the rule is determined by the mined frequent item set, such as the frequent 4 item set L4＝{b1,b5,b8,b11}(biRepresenting the purchase of a product i) strong rules can be generated: { b1,b5,b8}→{b11The rule can be explained as follows: a user who purchased product 1, product 5, and product 8, may purchase product 11 at a high probability that the minimum confidence is met. Because the frequent item set is satisfied with the minimum support, the resulting rule is also satisfied with the minimum confidence.
When the product recommendation method provided by the embodiment of the present invention is implemented specifically, in an embodiment, the connecting with the customer transaction database to obtain the historical purchase data of the customer includes:
connecting a customer transaction database, and extracting historical purchase data of each customer within a preset time length; when historical purchase data is extracted, old data is discarded in real time, and new data is added continuously.
In an embodiment, the main process of obtaining the historical purchase data of the customer in the data preprocessing comprises the following steps: connecting a customer transaction database, and extracting historical purchase data of each customer within a preset time length; when historical purchase data is extracted, old data is discarded in real time, and new data is added continuously.
The invention mainly uses the purchase data of the customer, and the actual meaning of possible reference of too early purchase is not great when the historical purchase data of the customer is obtained, so that the value of the invention determines and extracts the purchase record of each customer in nearly three years, and the data preprocessing module needs to discard the old data in real time and continuously add new data, thereby bringing the defect of greatly improving the calculation load.
Fig. 3 is a schematic diagram of a calculation process of a support degree and a confidence coefficient parameter of a product recommendation method according to an embodiment of the present invention, and as shown in fig. 3, when a product recommendation method according to an embodiment of the present invention is implemented specifically, in an embodiment, the establishing of an evaluation function and the determining of the support degree and the confidence coefficient parameter by iterative computation include:
step 301: weighting according to the calculation complexity, the time complexity and the number of frequent itemsets to establish an evaluation function;
step 302: setting two parameter initial values according to the empirical value; wherein, two parameters include: support and confidence;
step 303: importing the initial values of the two parameters into an evaluation function for iterative computation, and updating the two parameters through iterative computation;
step 304: and when the value of the evaluation function reaches the optimal value, the threshold value of the iteration times is reached, and two parameter values of the evaluation function when the optimal value is selected are output as the support degree and the confidence coefficient parameters.
In an embodiment, the main process of iteratively calculating the support degree and the confidence degree parameters includes: firstly, weighting according to the calculation complexity, the time complexity and the number of frequent itemsets to establish an evaluation function; then setting two parameter initial values according to the empirical value; wherein, two parameters include: support and confidence; secondly, importing the initial values of the two parameters into an evaluation function for iterative computation, and updating the two parameters through iterative computation; and finally, when the value of the evaluation function reaches the optimal value, the iteration time threshold is reached, and two parameter values of the evaluation function when the optimal value is selected are output as the support degree and the confidence coefficient parameters.
The most important point of the embodiment of the invention is that the evaluation function of the parameters is determined and then iterative computation is carried out to obtain the parameters of the support degree and the confidence degree; the difficulty in the development process of the invention lies in the determination of the support degree and the confidence coefficient parameters, if the support degree and the confidence coefficient parameters are too small, more rules can be obtained, and if the support degree and the confidence coefficient parameters are too large, fewer rules can be obtained, so that in order to obtain a proper rule set, proper support degree and confidence coefficient parameters must be determined.
The evaluation function can be obtained by weighting according to factors such as calculation complexity, time complexity, frequent item set number and the like, two initial values can be given by referring to an empirical value, the iteration times are determined, an evaluation function value is calculated in each iteration, and finally a parameter value when the evaluation function takes the optimal value is selected;
the support degree and confidence coefficient parameters influence the recommendation result of the invention, and the finding of the most suitable recommended product is the maximum purpose of the invention. Therefore, in the experimental stage, conditions such as different iteration times, initial values, threshold values and the like need to be set for multiple experiments, and the optimal values of the support degree and the confidence coefficient parameters are found.
When the product recommendation method provided in an embodiment of the present invention is implemented specifically, in an embodiment, the determining a frequent item set according to a customer data set, a support degree, and a confidence coefficient parameter includes:
selecting a frequent item set mining algorithm according to the data volume and the system performance;
and (4) performing frequent item set mining on the client data set by using a frequent item set mining algorithm and utilizing the support degree and the confidence coefficient parameters to determine a frequent item set.
In the embodiment, the process of routing a frequent itemset mainly comprises the following steps: firstly, selecting a frequent item set mining algorithm according to the data volume and the system performance; and then, performing frequent item set mining on the client data set by using a frequent item set mining algorithm and using the support degree and the confidence coefficient parameters to determine a frequent item set.
During frequent item set mining, which mining methods can be selected according to factors such as data volume and system performance, and the method with the best final effect is selected through comparison, for example, Apriori algorithm and other improved algorithms can be adopted, and frequent item set mining is performed on a client data set by using support degree and confidence coefficient parameters, so that the maximum frequent item set is generated.
Fig. 4 is a schematic diagram of a rule set determining process of a product recommendation method according to an embodiment of the present invention, and as shown in fig. 4, when a product recommendation method provided in an embodiment of the present invention is implemented specifically, in an embodiment, the determining a rule set according to a frequent item set includes:
step 401: mining the frequent item sets according to a rule mining algorithm to generate a maximum frequent item set rule;
step 402: acquiring all subsets of the frequent item set;
step 403: mining each subset of the frequent item set according to a rule mining algorithm to generate a subset rule;
step 404: and determining a rule set according to the maximum frequent item set rule and the subset rule.
In an embodiment, the main process of mining the rule set includes: firstly, mining a frequent item set according to a rule mining algorithm to generate a maximum frequent item set rule; then acquiring all subsets of the frequent item set; next, mining each subset of the frequent item set according to a rule mining algorithm to generate a subset rule; and finally, determining a rule set according to the maximum frequent item set rule and the subset rule.
Mining the frequent item sets through a rule mining algorithm to generate a maximum frequent item set rule; the maximum frequent itemset rule belongs to a strong rule; for example, the largest frequent item set is the largest k item set, because all subsets of the largest k item set are frequent, all subsets of the largest frequent item set can generate rules; and mining each subset of the frequent item set according to a rule mining algorithm to generate a subset rule, wherein the subset rule also belongs to a strong rule. Finally, according to the most frequent item set rule and the subset rule, the determined rule set also belongs to the strong rule.
The recommendation method based on the strong rules has the core idea of mining the strong association rules. Although the association rule mining technology is a method in machine learning, the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users.
Fig. 5 is a schematic diagram of a recommended product list determining process of a product recommendation method according to an embodiment of the present invention, and as shown in fig. 5, when a product recommendation method according to an embodiment of the present invention is implemented specifically, in an embodiment, the determining a recommended product list according to a rule set and a customer data set includes:
step 501: carrying out sequence matching on the client data set and the rule set to determine the product recommendation rule;
step 502: dividing the product recommendation rule into a first half section and a second half section;
step 503: matching the first half section of the product recommendation rule with a customer purchase record in a customer data set to obtain a maximum matching value;
step 504: and outputting the products corresponding to the latter half section of the product recommendation rule as a recommended product list according to the maximum matching value.
In an embodiment, the main process of generating the recommended product list includes: firstly, carrying out sequence matching on a client data set and a rule set to determine a product recommendation rule; then dividing the product recommendation rule into a first half section and a second half section; next, matching the first half section of the product recommendation rule with a customer purchase record in a customer data set to obtain a maximum matching value; and finally, outputting the products corresponding to the latter half section of the product recommendation rule as a recommended product list according to the maximum matching value.
In embodiments where the customer data set is sequence matched to a rule set, we expect one rule to be: { a, b,. g, n } → { e }, we need to match the first half of the rule with the historical purchase records of the customer, and after the maximum matching value is obtained, the product in the second half of the rule is recommended to the customer.
When a user newly purchases a product, other products which are most purchased can be selected and recommended to the user according to the obtained rule set, and a specific numerical value is given, for example, 90% of users who purchase the product and purchase the product. If no suitable recommended product can be found, no recommendation is made.
The invention utilizes the actual purchasing data of the user, reduces the influence of useless data on the recommendation result, and can provide the accurate associated value of the purchased product and the recommended product in the historical purchasing record, such as how large group of users purchasing the current product buy the recommended product. If no recommended product meeting the requirements exists, no recommendation is made, and the user experience of the client is improved. When the recommendation is given to the user, a specific numerical value is given for the user to refer to, and the recommendation has strong persuasion.
When the product recommendation method of the embodiment of the invention is implemented, a technical developer firstly performs uniform preprocessing on purchase data of a user, namely, the output of a data preprocessing module is a data set suitable for association rule mining, then determines suitable support degree and confidence coefficient parameters, the parameters can be determined according to priori knowledge or can be set by self according to the condition of the data set, frequent item set mining is performed after the parameters are determined, a mature frequent item set mining technology can be used to obtain frequent item sets and then generate rule sets, and finally, recommended products are selected according to the matching degree of the rule sets. The difficulty in the development process lies in the determination of parameters, and if the parameters are too small, more rules are obtained, and if the parameters are too large, fewer rules are obtained, so that in order to obtain a proper rule set, proper parameters must be determined.
Each operation or action of the user (e.g. purchasing a financial product) is marked as biFrom this, a historical purchase record B ═ B of each user can be obtained1,b4,...bi,..bnSuppose the user purchases multiple products, product 1, product 4, etc. In a database D of purchase records of a plurality of users, a set of 1, 2,. k items of the users (a set of k items refers to a sequence of k purchase records) is mined, indicating that the set of items is frequent if the set of k items satisfies a predefined minimum degree of support, while strong rules can be generated from the set of items. The recommendation method of the invention is based on the strong rule.
The recommendation method is based on strong rules, and the core idea is to mine strong association rules. Although the association rule mining technology is a method in machine learning, the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users.
The invention uses the actual ordering data of the user to generate a strong association rule through the purchase record of the user in the database, and the rule is determined by the mined frequent item set, such as a frequent 4-item set L4＝{b1,b5,b8,b11}(biRepresenting the purchase of a product i) strong rules can be generated: { b1,b5,b8}→{b11The rule can be explained as follows: a user who purchased product 1, product 5, and product 8, may purchase product 11 at a high probability that the minimum confidence is met. Because the frequent item set is satisfied with the minimum support, the resulting rule is also satisfied with the minimum confidence.
Fig. 6 is a flow chart illustrating interaction between a modularized example of a product recommendation method according to an embodiment of the present invention, and as shown in fig. 6, a modularized example of a product recommendation method according to an embodiment of the present invention further includes:
the system comprises a data preprocessing module, a parameter determining module, frequent item set mining, rule generation and prediction according to the generated rule. The functions of the modules are briefly described as follows:
1) number ofAccording to the preprocessing module: traversing the database, marking the products purchased by the customer according to the purchase records of the customer, and generating a piece of data, namely: b isi＝{b1,b5,...bj,..bnThe user purchases a plurality of products such as product 1 and product 5, and the data set of all customers is obtained after one-time traversal;
2) determining parameters: the method mainly comprises the steps of determining an evaluation function of parameters, wherein the evaluation function can be obtained by weighting according to factors such as calculation complexity, time complexity, frequent item set number and the like, two initial values can be given by referring to an empirical value, the iteration times are determined, the evaluation function value is calculated in each iteration, and finally, a parameter value of the evaluation function when the optimal value is obtained is selected;
3) frequent item set mining: generating a maximum frequent item set by using an existing frequent item set mining algorithm, such as an Apriori algorithm and other improved algorithms;
4) determining a rule: because all subsets of the maximum k-term set are frequent, all subsets of the maximum frequent term set can generate rules;
5) recommending products: when a user newly purchases a product, other products which are most purchased can be selected and recommended to the user according to the obtained rule set, and a specific numerical value is given, for example, 90% of users who purchase the product and purchase the product. If no suitable recommended product can be found, no recommendation is made.
The corresponding relationship and the action flow between each step and each module are briefly described as follows:
step 1: namely a data pre-processing module. The invention mainly uses the purchasing data of the customers, but the actual meaning of possible reference of too early purchasing is not great, so the value of the invention is determined and extracted the purchasing records of each customer in nearly three years, and the data preprocessing module needs to discard the old data in real time and continuously add new data, thereby bringing the defect of greatly improving the calculation load and continuously optimizing the data in the follow-up process.
Step 2: namely a parameter update module. The final result of the module influences the recommendation result of the invention, and finding the most suitable recommended product is the maximum purpose of the invention. Therefore, in the experimental stage, conditions such as different iteration times, initial values, threshold values and the like need to be set for multiple experiments, and a value with the best effect is found.
And step 3: namely a frequent item set mining module. The module mainly relates to a frequent item set mining algorithm, a plurality of mature algorithms exist at present, and further algorithms are provided on the algorithms, so that the methods can be comprehensively selected according to factors such as data volume, system performance and the like. And selecting the method set with the best final effect through comparison.
And 4, step 4: namely a rule generation module. The module also has a more sophisticated algorithm because the subset of the frequent k-term set is also frequent, so the final rule set is a strong rule generated by the frequent k-term set and its subset together.
And 5: i.e. a recommendation module. This module requires sequence matching and we want a rule to be: { a, b,. g, n } → { e }, we need to match the first half of the rule with the historical purchase records of the customer, and after the maximum matching value is obtained, the product in the second half of the rule is recommended to the customer.
The association rule mining method aims at knowledge discovery, and the recommendation technology is realized by using the idea, so that the method has strong interpretability. Secondly, the invention uses the actual purchase data of the user, thereby greatly reducing the recommendation error and enhancing the experience of the customer. Finally, the method and the device give specific numerical values while recommending the users, can be referred by the users, and have strong persuasion.
The above process relies on frequent pattern mining, so the data needs to be preprocessed into a dataset suitable for mining a frequent set of items, so if the data meets the requirements, no preprocessing step may be required, but if the data does not meet the requirements, preprocessing is required.
The invention utilizes the actual purchasing data of the user, reduces the influence of useless data on the recommendation result, and can provide the accurate associated value of the purchased product and the recommended product in the historical purchasing record, such as how large group of users purchasing the current product buy the recommended product. If no recommended product meeting the requirements exists, no recommendation is made, and the user experience of the client is improved.
The key points of the embodiment of the invention are as follows: in the parameter determining module, an evaluation function needs to be determined, a final parameter value is determined according to whether the evaluation function is optimal or not, the evaluation function can be obtained by weighting according to factors such as calculation complexity, time complexity, frequent item set number and the like, the optimal value of the evaluation function determines a final rule set, and meanwhile, the final recommendation result is influenced.
The protection points of the embodiment of the invention are as follows: the invention is expected to protect the real data of the client from leakage in a data level, and is expected to protect the idea and algorithm of a parameter determination module in a technical level.
The invention provides a recommendation technology based on association rules, which is an interpretable recommendation method, overcomes the defects of most of the existing recommendation methods, realizes personalized recommendation for different users, and realizes accurate recommendation. The invention takes the situation that users buy financial products as an example, and realizes the recommendation of different investment financial products for different users.
The recommendation method based on the strong rules has the core idea of mining the strong association rules. Although the association rule mining technology is a method in machine learning, the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users. The influence of useless data on the recommendation result is reduced by utilizing the actual purchase data of the user, and the accurate association value of the purchased product and the recommended product in the historical purchase record, such as how large group of users purchasing the current product buy the recommended product, can be given by the invention. If no recommended product meeting the requirements exists, no recommendation is made, and the user experience of the client is improved.
The association rule mining method aims at knowledge discovery, and the recommendation technology is realized by using the idea, so that the method has strong interpretability. Secondly, the invention uses the actual purchase data of the user, thereby greatly reducing the recommendation error and enhancing the experience of the customer. Finally, the method and the device give specific numerical values while recommending the users, can be referred by the users, and have strong persuasion.
Fig. 7 is a schematic diagram of a computer device for executing a product recommendation method implemented by the present invention, and as shown in fig. 7, an embodiment of the present invention further provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor executes the computer program to implement the product recommendation method.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program for implementing the above product recommendation method is stored in the computer-readable storage medium.
The embodiment of the invention also provides a product recommending device, which is described in the following embodiment. Because the principle of the device for solving the problems is similar to a product recommendation method, the implementation of the device can refer to the implementation of the product recommendation method, and repeated parts are not repeated.
Fig. 8 is a schematic diagram of a product recommendation device according to an embodiment of the present invention, and as shown in fig. 8, the embodiment of the present invention further provides a product recommendation device, which may include:
the preprocessing module 801 is used for acquiring historical purchase data of a customer for preprocessing and determining a customer data set;
a support degree and confidence coefficient parameter determining module 802, configured to establish an evaluation function, and determine support degree and confidence coefficient parameters through iterative computation;
a frequent item set determining module 803, configured to determine a frequent item set according to the client data set, the support degree, and the confidence coefficient parameter;
a rule set determining module 804, configured to determine a rule set according to the frequent item set;
and a product recommendation module 805 configured to determine a recommended product list according to the rule set and the customer data set.
In an embodiment of the invention, when the product recommendation apparatus provided in the embodiment of the present invention is implemented specifically, the preprocessing module is specifically configured to:
connecting a customer transaction database to obtain historical purchase data of a customer;
analyzing historical purchase data of a customer to obtain a purchase record of the customer, marking a product purchased by the customer, and generating customer data;
and traversing the customer transaction database, and summarizing and counting all the generated customer data into a customer data set.
In an embodiment of the invention, when the product recommendation apparatus provided in the embodiment of the present invention is implemented specifically, the preprocessing module is further configured to:
connecting a customer transaction database, and extracting historical purchase data of each customer within a preset time length; when historical purchase data is extracted, old data is discarded in real time, and new data is added continuously.
In an embodiment of the invention, when the product recommendation apparatus provided in the embodiment of the present invention is implemented specifically, the support degree and confidence level parameter determining module is specifically configured to:
weighting according to the calculation complexity, the time complexity and the number of frequent itemsets to establish an evaluation function;
setting two parameter initial values according to the empirical value; wherein, two parameters include: support and confidence;
importing the initial values of the two parameters into an evaluation function for iterative computation, and updating the two parameters through iterative computation;
and when the value of the evaluation function reaches the optimal value, the threshold value of the iteration times is reached, and two parameter values of the evaluation function when the optimal value is selected are output as the support degree and the confidence coefficient parameters.
In an embodiment of the invention, when the product recommendation apparatus provided in the embodiment of the present invention is implemented specifically, the frequent item set determining module is specifically configured to:
selecting a frequent item set mining algorithm according to the data volume and the system performance;
and (4) performing frequent item set mining on the client data set by using a frequent item set mining algorithm and utilizing the support degree and the confidence coefficient parameters to determine a frequent item set.
In an embodiment of the invention, when the product recommendation apparatus provided in the embodiment of the present invention is implemented specifically, the rule set determining module is specifically configured to:
mining the frequent item sets according to a rule mining algorithm to generate a maximum frequent item set rule;
acquiring all subsets of the frequent item set;
mining each subset of the frequent item set according to a rule mining algorithm to generate a subset rule;
and determining a rule set according to the maximum frequent item set rule and the subset rule.
In an embodiment of the invention, when the product recommendation device provided in the embodiment of the present invention is implemented specifically, the product recommendation module is specifically configured to:
carrying out sequence matching on the client data set and the rule set to determine the product recommendation rule;
dividing the product recommendation rule into a first half section and a second half section;
matching the first half section of the product recommendation rule with a customer purchase record in a customer data set to obtain a maximum matching value;
and outputting the products corresponding to the latter half section of the product recommendation rule as a recommended product list according to the maximum matching value.
To sum up, a product recommendation method and apparatus provided by the embodiments of the present invention include: firstly, acquiring historical purchase data of a customer for preprocessing, and determining a customer data set; then establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation; then, determining a frequent item set according to the client data set, the support degree and the confidence coefficient parameters; determining a rule set according to the frequent item set; and finally, determining a recommended product list according to the rule set and the customer data set. The embodiment of the invention realizes the recommendation technology based on the association rule, has the core idea of mining the rule set of the strong association rule, is an interpretable recommendation method, solves the defects of most of the existing recommendation methods, and realizes the personalized recommendation aiming at different users and the accurate recommendation. The union rule mining technology is a method in machine learning, but the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users. According to the invention, the actual purchase data of the user is utilized, so that the influence of useless data on the recommendation result is reduced, and the experience of the user is enhanced.
The invention provides a recommendation technology based on association rules, which is an interpretable recommendation method, overcomes the defects of most of the existing recommendation methods, realizes personalized recommendation for different users, and realizes accurate recommendation. The invention takes the situation that users buy financial products as an example, and realizes the recommendation of different investment financial products for different users.
The recommendation method based on the strong rules has the core idea of mining the strong association rules. Although the association rule mining technology is a method in machine learning, the method of the invention is not limited by data types, because the invention is realized based on the behavior data of users.
The influence of useless data on the recommendation result is reduced by utilizing the actual purchase data of the user, and the accurate association value of the purchased product and the recommended product in the historical purchase record, such as how large group of users purchasing the current product buy the recommended product, can be given by the invention. If no recommended product meeting the requirements exists, no recommendation is made, and the user experience of the client is improved.
The association rule mining method aims at knowledge discovery, and the recommendation technology is realized by using the idea, so that the method has strong interpretability. Secondly, the invention uses the actual purchase data of the user, thereby greatly reducing the recommendation error and enhancing the experience of the customer. Finally, the method and the device give specific numerical values while recommending the users, can be referred by the users, and have strong persuasion.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (16)
1. A method for recommending products, comprising:
acquiring historical purchase data of a customer for preprocessing, and determining a customer data set;
establishing an evaluation function, and determining support degree and confidence coefficient parameters through iterative calculation;
determining a frequent item set according to a client data set, support degree and confidence coefficient parameters;
determining a rule set according to the frequent item set;
and determining a recommended product list according to the rule set and the customer data set.
2. The method of claim 1, wherein obtaining historical purchase data for a customer for preprocessing to determine a customer data set comprises:
connecting a customer transaction database to obtain historical purchase data of a customer;
analyzing historical purchase data of a customer to obtain a purchase record of the customer, marking a product purchased by the customer, and generating customer data;
and traversing the customer transaction database, and summarizing and counting all the generated customer data into a customer data set.
3. The method of claim 2, wherein connecting to a customer transaction database to obtain customer historical purchase data comprises:
connecting a customer transaction database, and extracting historical purchase data of each customer within a preset time length; when historical purchase data is extracted, old data is discarded in real time, and new data is added continuously.
4. The method of claim 1, wherein establishing an evaluation function and determining support and confidence parameters by iterative calculations comprises:
weighting according to the calculation complexity, the time complexity and the number of frequent itemsets to establish an evaluation function;
setting two parameter initial values according to the empirical value; wherein, two parameters include: support and confidence;
importing the initial values of the two parameters into an evaluation function for iterative computation, and updating the two parameters through iterative computation;
and when the value of the evaluation function reaches the optimal value, the threshold value of the iteration times is reached, and two parameter values of the evaluation function when the optimal value is selected are output as the support degree and the confidence coefficient parameters.
5. The method of claim 1, wherein determining a frequent item set based on the customer data set, the support, and the confidence parameter comprises:
selecting a frequent item set mining algorithm according to the data volume and the system performance;
and (4) performing frequent item set mining on the client data set by using a frequent item set mining algorithm and utilizing the support degree and the confidence coefficient parameters to determine a frequent item set.
6. The method of claim 1, wherein determining a rule set based on a frequent item set comprises:
mining the frequent item sets according to a rule mining algorithm to generate a maximum frequent item set rule;
acquiring all subsets of the frequent item set;
mining each subset of the frequent item set according to a rule mining algorithm to generate a subset rule;
and determining a rule set according to the maximum frequent item set rule and the subset rule.
7. The method of claim 2, wherein determining the list of recommended products based on the set of rules and the set of customer data comprises:
carrying out sequence matching on the client data set and the rule set to determine the product recommendation rule;
dividing the product recommendation rule into a first half section and a second half section;
matching the first half section of the product recommendation rule with a customer purchase record in a customer data set to obtain a maximum matching value;
and outputting the products corresponding to the latter half section of the product recommendation rule as a recommended product list according to the maximum matching value.
8. A product recommendation device, comprising:
the preprocessing module is used for acquiring historical purchase data of a customer for preprocessing and determining a customer data set;
the support degree and confidence coefficient parameter determining module is used for establishing an evaluation function and determining support degree and confidence coefficient parameters through iterative calculation;
the frequent item set determining module is used for determining a frequent item set according to the client data set, the support degree and the confidence coefficient parameters;
the rule set determining module is used for determining a rule set according to the frequent item set;
and the product recommendation module is used for determining a recommended product list according to the rule set and the customer data set.
9. The apparatus of claim 8, wherein the pre-processing module is specifically configured to:
connecting a customer transaction database to obtain historical purchase data of a customer;
analyzing historical purchase data of a customer to obtain a purchase record of the customer, marking a product purchased by the customer, and generating customer data;
and traversing the customer transaction database, and summarizing and counting all the generated customer data into a customer data set.
10. The apparatus of claim 9, wherein the pre-processing module is further to:
connecting a customer transaction database, and extracting historical purchase data of each customer within a preset time length; when historical purchase data is extracted, old data is discarded in real time, and new data is added continuously.
11. The apparatus of claim 8, wherein the support and confidence parameter determination module is specifically configured to:
weighting according to the calculation complexity, the time complexity and the number of frequent itemsets to establish an evaluation function;
setting two parameter initial values according to the empirical value; wherein, two parameters include: support and confidence;
importing the initial values of the two parameters into an evaluation function for iterative computation, and updating the two parameters through iterative computation;
and when the value of the evaluation function reaches the optimal value, the threshold value of the iteration times is reached, and two parameter values of the evaluation function when the optimal value is selected are output as the support degree and the confidence coefficient parameters.
12. The apparatus of claim 8, wherein the frequent item set determination module is specifically configured to:
selecting a frequent item set mining algorithm according to the data volume and the system performance;
and (4) performing frequent item set mining on the client data set by using a frequent item set mining algorithm and utilizing the support degree and the confidence coefficient parameters to determine a frequent item set.
13. The apparatus of claim 8, wherein the rule set determination module is specifically configured to:
mining the frequent item sets according to a rule mining algorithm to generate a maximum frequent item set rule;
acquiring all subsets of the frequent item set;
mining each subset of the frequent item set according to a rule mining algorithm to generate a subset rule;
and determining a rule set according to the maximum frequent item set rule and the subset rule.
14. The apparatus of claim 9, wherein the product recommendation module is specifically configured to:
carrying out sequence matching on the client data set and the rule set to determine the product recommendation rule;
dividing the product recommendation rule into a first half section and a second half section;
matching the first half section of the product recommendation rule with a customer purchase record in a customer data set to obtain a maximum matching value;
and outputting the products corresponding to the latter half section of the product recommendation rule as a recommended product list according to the maximum matching value.
15. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of claims 1 to 7 when executing the computer program.
16. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program for executing a method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110869207.2A CN113570437A (en) | 2021-07-30 | 2021-07-30 | Product recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110869207.2A CN113570437A (en) | 2021-07-30 | 2021-07-30 | Product recommendation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113570437A true CN113570437A (en) | 2021-10-29 |
Family
ID=78169341
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110869207.2A Pending CN113570437A (en) | 2021-07-30 | 2021-07-30 | Product recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113570437A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117252636A (en) * | 2023-11-17 | 2023-12-19 | 国网山东省电力公司营销服务中心(计量中心) | Electricity fee package type optimization method and system based on user |
| CN117290609A (en) * | 2023-11-24 | 2023-12-26 | 中国科学技术大学 | Product data recommendation method and product data recommendation device |
-
2021
- 2021-07-30 CN CN202110869207.2A patent/CN113570437A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117252636A (en) * | 2023-11-17 | 2023-12-19 | 国网山东省电力公司营销服务中心(计量中心) | Electricity fee package type optimization method and system based on user |
| CN117290609A (en) * | 2023-11-24 | 2023-12-26 | 中国科学技术大学 | Product data recommendation method and product data recommendation device |
| CN117290609B (en) * | 2023-11-24 | 2024-03-29 | 中国科学技术大学 | Product data recommendation method and product data recommendation device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330445B (en) | User attribute prediction method and device | |
| CN111339427B (en) | Book information recommendation method, device and system and storage medium | |
| CN106033424B (en) | Data mining method and device | |
| WO2019200480A1 (en) | Method and system for model auto-selection using an ensemble of machine learning models | |
| US11366806B2 (en) | Automated feature generation for machine learning application | |
| CN113570437A (en) | Product recommendation method and device | |
| CN109118155B (en) | Method and device for generating operation model | |
| US20190080352A1 (en) | Segment Extension Based on Lookalike Selection | |
| WO2020034880A1 (en) | Logistics object information processing method, device and computer system | |
| CN107507028A (en) | User preference determines method, apparatus, equipment and storage medium | |
| CN113449753B (en) | Service risk prediction method, device and system | |
| CN110956278A (en) | Method and system for retraining machine learning models | |
| CN114997916A (en) | Prediction method, system, electronic device and storage medium of potential user | |
| Li et al. | Research on the application of multimedia entropy method in data mining of retail business | |
| JP2005222445A (en) | Information processing method and analysis device in data mining | |
| CN115167965A (en) | Transaction progress bar processing method and device | |
| CN114881521A (en) | Service evaluation method, device, electronic equipment and storage medium | |
| CN114138743A (en) | ETL task automatic configuration method and device based on machine learning | |
| Nagashima et al. | Data Imputation Method based on Programming by Example: APREP-S | |
| CN112328899A (en) | Information processing method, information processing apparatus, storage medium, and electronic device | |
| CN114547231A (en) | Data tracing method and system | |
| CN111949783A (en) | Question and answer result generation method and device in knowledge base | |
| US20230132618A1 (en) | Method for denoising click data, electronic device and storage medium | |
| US11928128B2 (en) | Construction of a meta-database from autonomously scanned disparate and heterogeneous sources | |
| US11822564B1 (en) | Graphical user interface enabling interactive visualizations using a meta-database constructed from autonomously scanned disparate and heterogeneous sources |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |