CN114971761A - A product recommendation method based on machine learning - Google Patents

Abstract

An embodiment of the present specification provides a commodity recommendation method based on machine learning, including: judging whether a user is in an isolated state or not based on the consumption condition and the position information of the user in the first time; and determining recommended commodities pushed to the user through a recommendation model in response to the user being in the isolation state.

Description

A product recommendation method based on machine learning

technical field

This specification relates to the field of computer technology, in particular to a method for recommending products based on machine learning.

Background technique

With the continuous development of electronic technology and network technology, shopping platforms continue to collect user data and can recommend products to users based on product feature information and user feature information. However, in special periods (such as flu outbreaks), users tend to live at home for a longer time, which leads to changes in users' shopping preferences, so that the recommended products do not meet users' preferences. Therefore, how to identify the user's home status to make more reasonable product recommendation is an urgent problem to be solved.

SUMMARY OF THE INVENTION

One of the embodiments of this specification provides a product recommendation method based on machine learning. The product recommendation method includes: judging whether the user is in an isolated state based on the user's consumption situation and location information within a first time; in response to the user being in an isolated state, determining a recommendation pushed to the user through a recommendation model commodity.

One of the embodiments of the present specification provides a product recommendation system based on machine learning. The system includes: a judgment module for judging whether the user is in an isolated state based on the user's consumption situation and location information in the first time. and a recommendation module, configured to determine, through a recommendation model, a commodity recommended to the user in response to the user being in an isolated state.

One of the embodiments of this specification provides a device for recommending products based on machine learning, including a processor, where the processor is configured to execute the method for recommending products based on machine learning described in the embodiments of this specification. method.

One of the embodiments of this specification provides a computer-readable storage medium, where the storage medium stores computer instructions, and after the computer reads the computer instructions in the storage medium, the computer executes the machine learning-based product recommendation described in the embodiments of this specification method.

Description of drawings

The present specification will be further explained by way of exemplary embodiments, which will be described in detail by means of the accompanying drawings. These embodiments are not limiting, and in these embodiments, the same numbers refer to the same structures, wherein:

FIG. 1 is a schematic diagram of an application scenario of a product recommendation system according to some embodiments of this specification;

Fig. 2 is a block diagram of a product recommendation system based on machine learning according to some embodiments of the present specification;

3 is an exemplary flowchart of a method for recommending products based on machine learning according to some embodiments of the present specification;

Fig. 4 is an exemplary flowchart of a method for judging an isolation state according to some embodiments of the present specification;

FIG. 5 provides a schematic diagram of a method for determining a recommended product according to some embodiments of the present specification;

FIG. 6 provides a flowchart of another method of determining recommended items according to some embodiments of the present specification.

Detailed ways

In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the following briefly introduces the accompanying drawings that are required to be used in the description of the embodiments. Obviously, the accompanying drawings in the following description are only some examples or embodiments of the present specification. For those of ordinary skill in the art, without creative efforts, the present specification can also be applied to the present specification according to these drawings. other similar situations. Unless obvious from the locale or otherwise specified, the same reference numbers in the figures represent the same structure or operation.

It should be understood that "system", "device", "unit" and/or "module" as used herein is a method used to distinguish different components, elements, parts, sections or assemblies at different levels. However, other words may be replaced by other expressions if they serve the same purpose.

As shown in the specification and claims, unless the context clearly dictates otherwise, the words "a," "an," "an," and/or "the" are not intended to be specific in the singular and may include the plural. Generally speaking, the terms "comprising" and "comprising" only imply the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

Flowcharts are used in this specification to illustrate operations performed by a system according to an embodiment of this specification. It should be understood that the preceding or following operations are not necessarily performed in the exact order. Rather, the various steps may be processed in reverse order or simultaneously. At the same time, other actions can be added to these procedures, or a step or actions can be removed from these procedures.

FIG. 1 is a schematic diagram of an application scenario of a product recommendation system according to some embodiments of the present specification. As shown in Figure 1, the application scenario of the product recommendation system based on machine learning may include a server 110, a processor 120, a storage device 130, a user terminal 140, a network 150, and the like.

The product recommendation system can be used in the sales service platform. In some embodiments, the system includes a sales service platform for retail merchandise. For example, e-commerce platforms, vending machines, etc. The product recommendation system can recommend products to users through the product recommendation method disclosed in this specification.

In some embodiments, the sales service platform may provide product display services for offline merchants (also referred to as stores), and the user may access the sales service platform through the user terminal 140 and purchase offline merchants according to the push information.

The server 110 may communicate with the processor 120, the storage device 130, and the user terminal 140 through the network 150 to provide various functions of product recommendation. In some embodiments, the user terminal 140 may send the current user's consumption situation and location information in the first time to the server 110, and receive the product recommendation information sent by the server 110. The server 110 may acquire store location information and store epidemic prevention assessment information, and perform processing to determine recommended products to be sent to nearby user terminals 140 . The information transfer relationship between the above devices is only an example. In some cases, the information transfer between the devices may also include other forms, which can be determined according to the actual situation.

Server 110 may be used to manage historical consumption data of users and process data and/or information from at least one component of the system or external data sources (e.g., cloud data centers). In some embodiments, server 110 may be a single server or a group of servers. The server group may be centralized or distributed (e.g., server 110 may be a distributed system), dedicated or concurrently served by other devices or systems. In some embodiments, server 110 may be regional or remote. In some embodiments, server 110 may be implemented on a cloud platform, or provided virtually.

Processor 120 refers to a device or system having computing capabilities. In some embodiments, processor 120 may include one or more processing engines (eg, a single-core processing engine or a multi-core processor). For example only, the processing engine may include one or more combinations of a central processing unit, an application specific instruction set processor (ASIP), a digital signal processor (DSP), a reduced instruction set computer (RISC), a microprocessor, and the like. In some embodiments, the processor 120 may invoke data and/or instructions stored in the storage device 130 to implement the commodity recommendation method provided in this specification. In some embodiments, the processor 120 may determine whether the user is in an isolation state based on the user's consumption situation and location information in the first time. In some embodiments, the processor 120 may recommend items to the user based on the user's isolation status.

Storage device 130 may be used to store data and/or instructions. The storage device 130 may include one or more storage components, and each storage component may be an independent device or a part of other devices. In some embodiments, the storage device 130 may also store relevant information generated during the commodity sales process. For example, the relevant information may include the store and consumption information generated by the user. Exemplarily, the relevant information may include the consumption situation and location information of the user. As well as user-related information such as consumption habits; for another example, relevant information can also include store-related information such as store location information, store epidemic prevention assessment information, store inventory information, and product characteristics. In some embodiments, the storage device 130 may store the relevant data generated by the sales service platform, for example, the pushed commodities pushed by the platform to the user, the user's isolation status and the recommendation model, etc. The relevant data generated when the commodity recommendation method is executed. In some embodiments, storage device 130 may include random access memory (RAM), read only memory (ROM), mass storage, removable memory, volatile read-write memory, the like, or any combination thereof. In some embodiments, the storage device 130 may be implemented on a cloud platform.

The user terminal 140 refers to one or more terminal devices or software used by the user. The user terminal 140 may include at least a processing unit, a display unit, an input/output unit, a storage unit, and the like. In some embodiments, the user can purchase commodities through the user terminal, and upload information/data to the server 110 or the storage device 130 through the user terminal. In some embodiments, one or more users may use the user terminal 140, which may include users who directly use the service, and may also include other related users.

The network 150 may connect components of the system and/or connect portions of the system with external resources. The network 150 enables communication between the various components, as well as with other components outside the system. For example, the processor 120 may obtain the location information of the user from the user terminal 140 through the network 150 . For another example, the processor 120 may obtain user-related information (eg, the user's consumption situation, location information, and consumption habits, etc.) and store-related information (eg, store location information of the store, store epidemic prevention assessment, etc.) from the storage device 130 through the network 150 . information, store inventory information, product characteristics, etc.). In some embodiments, the processor 120 may also send the product recommendation information to the user terminal 140 via the network 150.

Fig. 2 is a block diagram of a product recommendation system based on machine learning according to some embodiments of the present specification.

In some embodiments, the product recommendation system module 200 based on machine learning may include a judgment module 210 and a recommendation module 220.

In some embodiments, the judging module 210 is configured to judge whether the user is in an isolation state based on the consumption situation and location information of the user in the first time. For more details on determining whether the user is in the isolation state, please refer to Fig. 3 and its related descriptions.

In some embodiments, the judging module 210 is further configured to determine the user's evaluation score based on the difference between the user's purchasing habits in the first time period and the user's purchasing habits in the second time period, where The second time period is a time period with the same length before the first time period; whether the user is in an isolation state is determined according to the user's evaluation score. More information on the differences in evaluation scores and purchasing habits can be found in Figure 4 and its associated description

In some embodiments, the recommendation module 220 is configured to determine, through a recommendation model, an item to recommend to the user in response to the user being in an isolated state. For more content about the recommended products to the user, please refer to Fig. 3 and related descriptions.

In some embodiments, the recommendation module 220 is further configured to determine candidate recommended commodities based on store location information and store epidemic prevention assessment information; and determine the recommended commodities from the candidate recommended commodities through a recommendation model. For more information about store related information, please refer to Figure 5 and its related description.

In some embodiments, the recommending module 220 is further used for the recommending module to determine, based on the location information of the user, a potential isolated user in a nearby area; determine the consumption situation of the potential isolated user, and based on the location information of the user The consumption situation of the potential isolated user determines a reference commodity; based on the reference commodity, the recommended commodity is determined through a recommendation model. See Figure 6 and its associated description for more on potential segregated users and reference items.

It should be noted that the above description of the product recommendation system and its modules is only for the convenience of description, and does not limit the present specification to the scope of the illustrated embodiments. It can be understood that for those skilled in the art, after understanding the principle of the system, various modules may be combined arbitrarily, or a subsystem may be formed to connect with other modules without departing from the principle. In some embodiments, the judgment module 210 and the recommendation module 220 disclosed in FIG. 2 may be different modules in a system, or may be a module that implements the functions of the above-mentioned two or more modules. For example, each module may share one storage module, and each module may also have its own storage module. Such deformations are all within the protection scope of this specification.

Fig. 3 is an exemplary flowchart of a method for recommending products based on machine learning according to some embodiments of the present specification. In some embodiments, process 300 may be performed by a processing device (eg, processor 120). For example, the process 300 may be stored in a storage device in the form of a program or an instruction, and the process 300 may be implemented when the server or the module shown in FIG. 2 executes the program or the instruction. As shown in Figure 3, the process 300 may include the following steps:

Step 310, based on the user's consumption situation and location information in the first time, determine whether the user is in an isolated state. In some embodiments, step 310 may be performed by the determination module 210 .

A user may refer to a consumer of the sales service platform. Users can view and/or purchase products from offline stores on the sales service platform. In some embodiments, the user can be identified by a user identification (User Identification, UID), wherein the user identification can be a unique and unchangeable account label generated by the sales service platform when the user registers an account with the sales service platform. For example, after receiving the user, the sales service platform can determine the specific user according to the UID in the user order.

The first time may refer to a period of time before the current time. In some embodiments, the first time may be determined according to user behavior. For example, the first time may be a period of time since the user's location is changed. The time can be from 10 days ago to the present. In some embodiments, the first time may be determined according to the user's purchase behavior, for example, the first time may be a time period from the date when the user first placed an order during the user's quarantine or home time.

The user's consumption situation may refer to the user's commodity consumption situation, for example, the consumption situation may be a list of commodities successfully traded by the user, wherein the commodity list may include the name, quantity, price, time and other related information of the commodities purchased by the user. In some embodiments, the consumption situation of the user may include the commodity consumption situation of the user in the sales service platform and the commodity consumption situation of the user in the store. For example, when a user buys a product in a store, he can pay the bill for the purchased product through the cash register system of the sales service platform.

The location information may be the current geographic information of the user when purchasing the commodity. For example, when the user purchases the commodity through the sales service platform, the location information may be the location information corresponding to the positioning information of the user terminal. For another example, when a user purchases a commodity in a store, the location information may be the location information of the store.

The user's consumption situation and location information in the first time can be determined according to the user's consumption order. For example, when the user purchases goods through the sales service platform, a consumption order can be generated according to the user's consumption situation, that is, the user's consumption order in the first time can be obtained. The consumption situation and location information can be realized by acquiring and analyzing the user's consumption orders in the first time. For another example, when a user buys a product in a store, the cashier system of the store can communicate with the sales service platform. When the user pays the bill, the sales service platform can generate a corresponding consumption order according to the user's consumption situation.

The isolation state can refer to the state in which the user mainly moves indoors for a period of time and goes out less or not. For example, the isolation state may refer to the state in which patients with infectious diseases or suspected patients live in a specific place (such as an isolation point) and are prohibited from going out. For another example, the quarantine status can also refer to the fact that during the outbreak of influenza, the user actively

In some embodiments, whether the user is in quarantine can be determined based on quarantined commodities and non-quarantined commodities, wherein non-quarantined commodities may refer to commodities that users cannot purchase during quarantine, such as fresh meat commodities, large electrical appliances, and outdoor sports commodities Wait. Quarantine commodities can refer to commodities that users are more willing to buy during the quarantine period, such as masks, instant noodles, milk, and sealed meat products.

In some embodiments, the user's possibility of being quarantined can be reflected based on the consumption of quarantined commodities and non-quarantined commodities by the user, and it can be determined that the user may be in the quarantined state as the user is in the quarantined state. For example, quarantine products can reflect the user's willingness to stay at home, and when the user buys a lot of quarantine products, it can be judged that the user may be in a state of quarantine. Exemplarily, when the commodity placed by the user within the first time period is only a quarantine commodity, it is considered that the user may be in a quarantine state or is about to enter a quarantine state.

In some embodiments, if the type and/or quantity of commodities during the quarantine period in the commodity data is greater than the first threshold, it is considered that the aforementioned user may be in a quarantine state. The first threshold may be determined according to the actual situation. For example, the first threshold may be a preset value. Exemplarily, the first threshold may be 2 or 10, that is, the number of commodities in the quarantine period in the commodity data. If the type is greater than 2 and the number is greater than 10, it can be judged that the user is in isolation. In some embodiments, when the number or quantity of non-quarantine commodities in the user's consumption situation and location information is greater than a threshold, it is considered that the user is not in a quarantine state.

In some embodiments, the user's evaluation score may be determined according to the user's consumption situation, and then whether the user is in an isolation state may be determined based on the user's evaluation score. For the specific content of determining whether the user is in the isolation state based on the evaluation score, reference may be made to Figure 4 and its related descriptions.

Step 320, in response to the user being in the isolation state, determine the recommended product to be pushed to the user through the recommendation model. In some embodiments, step 320 may be performed by recommendation module 220 .

Recommended products are products that are recommended to users, and can also be referred to as recommended products. Recommended products can be determined from products other than non-isolated products and are related to the user’s isolation status and user preferences. For example, the recommended products can be products that are not affected by the user’s shopping preferences or shopping needs, such as skin care. products, tobacco and alcohol products, etc. For another example, the recommended products may be isolated products such as mineral water and instant noodles. In some embodiments, the recommended items may also be items purchased by other users in a quarantined state.

In some embodiments, products that need to be recommended to the user (ie, recommended products) may be determined through the first recommendation model. The input of the first recommendation model may include user-related information in an isolated state, and the output of the first recommendation model may be recommended products. The user-related information may include the user's basic information and/or the user's consumption situation. For example, the user-related information may include the user's location information, gender, age, and other information. For another example, the user-related information may include information such as commodity type, quantity, price, purchase time, and purchase location in each transaction order successfully traded by the user in the sales service platform.

In some embodiments, the first recommendation model may be a neural network such as Convolutional Neural Networks (CNN), Deep Neural Networks (DNN), and Long Short-Term Memory (LSTM) models. Model.

In some embodiments, the first recommendation model may also be other models, for example, the recommendation model may be a support vector machine model, a logistic regression model, a naive Bayesian classification model, a Gaussian distribution Bayesian classification model, a decision tree Machine learning algorithms or models such as models, random forest models, and KNN classification models

The first recommendation model can be obtained by training based on historical data. That is, the labeled training samples can be input into the initial first recommended model, and the parameters of the initial first recommended model can be trained and updated through optimization methods such as stochastic gradient descent. When the trained model meets the preset conditions, the training ends and the trained model is obtained. recommended model.

The training samples may be historical user-related data in the historical data. For example, the historical user-related data may include the basic information of the user and the user's historical consumption. Exemplarily, the historical consumption may include the user's transactions in the sales service platform. Information such as commodity type, quantity, price, purchase time, and purchase location in each successful transaction order. In some embodiments, historical users may include users in a quarantined state in the historical data. The training labels are the items purchased by the user in the history.

In some embodiments, the input of the first recommendation model may further include store-related data, wherein the store-related data may include related information such as commodity inventory, commodity type, commodity price, store promotion, etc., and may also include epidemic prevention evaluation information of the store, etc. . In some embodiments, the store may be a store selected by the user on the sales service platform and/or a nearby store determined according to the user's location information.

Fig. 4 is an exemplary flowchart of a method for judging an isolation state according to some embodiments of the present specification. In some embodiments, process 400 may be performed by a processing device (e.g., processor 120) or an item recommendation system module (e.g., determination module 210). For example, the process 400 may be stored in a storage device in the form of a program or an instruction, and the process 400 may be implemented when the server or the module shown in FIG. 2 executes the program or the instruction. As shown in Figure 4, the process 400 may include the following steps:

Step 410: Determine the user's evaluation score based on the difference between the user's consumption habits in the first time and the user's consumption habits in the second time. In some embodiments, step 410 may be performed by determination module 210.

The second time may be a period of time before the first time, and the length of the second time may be the same as the first time. In some embodiments, the second time may be a time when the user is in a non-isolated state. In some embodiments, in order to avoid that the user during the second time is also in the isolation state, the time distance between the second time and the first time may be greater than a preset threshold, for example, the last day of the second time must be at least the first time 14 days before the first day of time.

Consumption habits can be used to describe users' shopping needs and preferences. Among them, shopping needs may refer to the commodities that the user needs to purchase, and shopping preferences may refer to the shopping tendency of the user to the characteristics of the commodities when purchasing commodities. For example, the user's shopping preferences may include the tendency to purchase commodities with smaller packaging specifications.

In some embodiments, consumption habits can be described by a user's consumption feature vector. In some embodiments, the consumption feature vector may include the consumption situation of the user under each commodity category. For example, the consumption feature vector may be the purchase amount of the user in each commodity category. Exemplarily, the consumption feature vector may be (a, b , c, d, e), where the value at a-e can represent the quantity of a certain commodity purchased, a can represent the purchase quantity of drinking water in this time period, and b can represent the purchase quantity of convenience food in this time period , c can represent the purchase amount of canned food in this time period, d can represent the purchase amount of daily necessities in this time period, e can represent the purchase amount of tobacco and alcohol in this time period, etc. For example, the consumption feature vector of a user can be (3, 5, 5, 1), which means that the user purchases 3 drinking water products, 5 convenience food products, 5 canned products, and tobacco and alcohol products. 1 item. In some embodiments, elements in the consumption feature vector may also include other commodity types, which may be determined according to actual conditions.

In some embodiments, the consumption characteristic vector of the user may be determined according to the consumption situation of the user in the first time to represent the consumption habit of the user in the first time. In some embodiments, the consumption characteristic vector of the user may be determined according to the consumption situation of the user in the second time to represent the consumption habit of the user in the second time.

In some embodiments, the second time may also be a time period group sampled multiple times, that is, multiple time periods with the same duration as the first time may be selected as the second time, and the consumption habits of the second time may be each time The average consumption habits of the segment.

In some embodiments, the difference in consumption habits may be determined according to the difference between the first time consumption feature vector and the second time consumption characteristic vector. For example, by calculating the difference between the two, calculating the vector distance between the two, etc.

An assessment score can be used to describe the likelihood that a user is in quarantine. In some embodiments, the evaluation score reflects the difference in the consumption habits of the user at the first time and the second time. That is, the evaluation scores are related to differences in consumption habits. In some embodiments, the specific mapping relationship between the evaluation score and the difference in consumption habits can be determined according to actual needs. For example, the evaluation score is positively correlated with the difference in consumption habits, that is, when the difference in consumption habits is large, the evaluation score is larger, which in turn represents the user more likely to be in isolation. Otherwise, the evaluation score will be smaller. In some embodiments, differences in consumption habits may be positively correlated nonlinearly with assessment scores.

In some embodiments, the difference in consumption habits may be determined by the vector distance between the first time consumption feature vector and the second time consumption feature vector. When calculating the evaluation score, it can be determined according to the vector distance between the first time consumption feature vector and the second time consumption feature vector, wherein the vector distance can be determined according to related algorithms such as Euclidean distance, Manhattan distance, Chebyshev distance, Mahalanobis distance, etc. Determining the evaluation score based on the vector distance may be determined by a nonlinear function, such as an exponential function, a power function, and the like. In some embodiments, the evaluation score may range from 0-100.

In some embodiments, the user's evaluation score may be adjusted based on a weighted number of potentially segregated users within the user's vicinity.

In some embodiments, the weighted number of potential quarantined users may be determined based on the evaluation scores of the potential quarantined users. Among them, the evaluation score of the potential isolated user can be determined according to the difference between the feature vector of the potential isolated user's first time consumption and its second time consumption feature vector, and the difference can be determined by methods such as vector difference and distance between vectors. When determining the evaluation score of a potential quarantined user, only the difference between their own consumption habits in the first time period and the second time period is considered, and the potential quarantined users that may exist in the vicinity of the potential quarantined user are not considered. For example, the user's neighborhood includes 3 potential isolated users, and their evaluation scores are 90, 80, and 70, respectively. According to the evaluation scores of these three potential isolated users, the corresponding weighted numbers can be determined to be 0.9, 0.8, and 0.7. There are (0.9+0.8+0.7) = 2.4 potential isolated users in the adjacent area.

In some embodiments, adjusting the user's evaluation score based on the weighted number of potential quarantined users may include computationally adjusting the user's evaluation score based on the weighted number of potential quarantined users. For example, the adjustment coefficient is determined according to the weighted number of potential isolated users and the total number of users in the user's adjacent area, and the evaluation score is adjusted according to the adjustment coefficient. The weighted number of potential isolated users in an area is 2.4, and the total number of users in its adjacent area is 20, then the adjustment coefficient can be 2.4/20=0.12, and the adjusted evaluation score is 80*(1+1.12)=89.6. For another example, the balance value of the evaluation score is determined according to the weighted number of potential isolated users, and the evaluation score is adjusted based on the balance value. Exemplarily, regardless of the potential isolation user in the adjacent area, the score of User A is 70, and the weight of the potential isolated user in the adjacent area is 70. If the number is 2.4, the adjusted assessment score is 70+2.4=72.4 points.

In some embodiments, the user's evaluation score may be adjusted according to the weighted number of potential quarantined users in other ways, which may be determined according to the actual situation.

Step 420: Determine whether the user is in an isolation state according to the user's evaluation score. In some embodiments, step 420 may be performed by determination module 210.

In some embodiments, when the user's evaluation score is higher than a threshold, it can be determined that the user is in an isolated state. For example, the evaluation score can range from 0 to 100, and the threshold can be 60. That is, a user with an evaluation score higher than 60 is determined to be in a state of isolation. isolation state. The threshold value can also be other score values, which can be determined according to the actual situation.

FIG. 5 provides a schematic diagram of a method for determining a recommended product according to some embodiments of the present specification. In some embodiments, process 500 may be performed by processor 120 . As shown in Figure 5, process 500 includes the following steps:

Step 510: Determine the user in the isolation state and the corresponding user-related information.

For the specific content of the users in the isolation state and the user-related information, reference may be made to the relevant descriptions of steps 310 and 320 in FIG. 3 , which will not be repeated here.

Step 520: Determine candidate recommended products based on store location information and store epidemic prevention assessment information.

The store location information may include geographic location information of the store, for example, the latitude and longitude information of the store, or other information that can be used to determine the relative location of the store and the user. In some embodiments, the store in step 520 may be a store selected by the user for shopping. For example, the user may select a store for shopping on the sales service platform, and after selecting a store, product recommendation may be made based on the inventory of the store. . In some embodiments, the store in step 520 may also be a store near the user, wherein when the distance between the store and the user is less than the distance threshold, the store is determined to be a store near the user.

In some embodiments, the store location information may also include surrounding traffic conditions, store scale, and travel distance between the store and the user, and other information that affects the user's selection of a store. In some embodiments, whether the user will choose the store can be assisted according to the traffic situation between the store and the user. The road is often congested, etc.), the store may not be judged as the user's neighboring store. In some embodiments, the recommendation module 220 may determine candidate recommended items in the user's nearby stores. For example, all or some of the products in nearby stores are selected as candidate recommended products.

The store's epidemic prevention assessment information can be the response of the store to the epidemic prevention policy, and can be used to judge whether the store's epidemic prevention measures are in place. For example, whether the store salesperson regularly conducts nucleic acid testing, whether the store is regularly disinfected, etc. The epidemic prevention assessment information of the store can be determined according to the epidemic prevention rules in special periods. For example, the epidemic prevention rules stipulate that the outer packaging of the frozen products in the store needs to be tested for nucleic acid every week. The epidemic prevention assessment information of the store can include the nucleic acid detection of the outer packaging of the frozen products in the store. In some embodiments, the recommendation module 220 may determine candidate recommended items in stores with epidemic prevention measures in place.

In some embodiments, the recommendation module 220 may determine candidate products based on store location information and store epidemic prevention assessment information. The candidate recommended products may be products that can be recommended to the user. For example, the recommendation module 220 determines at least one store near the user according to the store location information, determines the epidemic prevention situation of each store according to the epidemic prevention evaluation information of each store in the at least one store, and determines the merchants whose epidemic prevention measures are in place are selected as candidate merchants, and then the candidate merchants are selected. A candidate recommended product is determined among the merchants. For another example, the recommendation module determines at least one store that has implemented epidemic prevention measures according to the store's epidemic prevention assessment information, determines the stores near the user as candidate merchants according to the location information of each store in the at least one store, and then determines candidate recommended products among the candidate merchants. Among them, the candidate recommended products can be the products in the store's inventory that meet the epidemic prevention rules. In some embodiments, it can be judged whether the goods in stock in the store comply with the epidemic prevention rules according to the epidemic prevention evaluation information of the store, and the commodities that meet the epidemic prevention rules are used as candidate recommended commodities, wherein the epidemic prevention rules can be determined according to relevant local regulations. For example, the epidemic prevention rules may include that the outer packaging of frozen products in the store needs to undergo nucleic acid testing every week. When determining candidate recommended products, it can be determined whether the outer packaging of the frozen products in the store has undergone nucleic acid testing every week. Products can be used as candidate recommended products. For another example, the epidemic prevention rules may include that meat products must report the source. When the source of meat products in the store is uncertain, meat products cannot be used as candidate recommended products.

Step 530, based on the user-related information and the candidate recommended commodities, determine the recommended commodities through the recommendation model.

In some embodiments, the recommended item may be determined from the candidate items by the second recommendation model. In some embodiments, the second recommendation model may be a neural network such as Convolutional Neural Networks (CNN), Deep Neural Networks (DNN), and Long Short-Term Memory (LSTM) models. network model. In some embodiments, the second recommendation model may also be other models, for example, support vector machine model, logistic regression model, naive Bayesian classification model, Gaussian distribution Bayesian classification model, decision tree model, random forest models and machine learning algorithms or models such as KNN classification models.

In some embodiments, the input of the second recommendation model may include user-related information in an isolated state and candidate recommended products, and the output of the second recommendation model may be recommended products.

In some embodiments, the second recommendation model may be trained based on historical data. That is, the labeled training samples can be input into the initial second recommended model, and the parameters of the initial second recommended model can be trained and updated by optimization methods such as stochastic gradient descent. When the trained model meets the preset conditions, the training ends and the trained model is obtained. recommended model. The training samples can be the historical user-related data in the historical data, and the labels are the commodities purchased by the user in the past.

For a detailed description of the user-related information and historical user-related information, please refer to the related content in step 320.

In some embodiments, the recommended product may also be determined based on the user's preference value for the candidate recommended product through a preset algorithm. The preference value may describe the user's tendency to purchase the product or the purchase probability. For example, the higher the preference value of the candidate recommended product, the more likely the user is to purchase the product.

In some embodiments, the recommendation module 220 may generate a user preference feature vector according to the user-related information, and generate a candidate recommended product feature vector according to the candidate recommended product. Among them, the user preference feature vector can be determined according to the types of commodities the user has purchased. For example, the user preference feature vector is (L1, L2, L3), and L1, L2, and L3 represent different commodity categories that the user has purchased; candidate recommended commodities The feature vector can be determined according to the category of the candidate product. For example, the feature vector of the candidate recommended product is (R1, R2, R3), where R1, R2, and R3 represent different product categories that the user has purchased. In some embodiments, the preference value of each candidate recommended product may be determined by the user preference feature vector and the candidate recommended product feature vector. For example, the vector inner product may be performed on the user preference feature vector and the candidate recommended product feature vector, and the inner product The result is used as the preference value of each candidate recommended product.

In some embodiments, candidate recommended products whose preference value satisfies preset rules may be selected as recommended products. In some embodiments, the preset rule may include a preset threshold, that is, when the preference value of the candidate recommended commodity is greater than the threshold preset value, the candidate recommended commodity may be regarded as a recommended commodity and pushed to the user. In some embodiments, the preset rules may include sorting rules, that is, the top N products in the preference value ranking among the candidate recommended products may be selected as the recommended products and pushed to the user.

Fig. 6 provides a flowchart of another method for determining recommended items according to some embodiments of the present specification. In some embodiments, process 600 may be performed by processor 120 . As shown in Figure 6, the process 600 includes the following steps:

Step 610: Determine the consumption situation of the potential isolated user in the user's vicinity, and determine the reference commodity based on the consumption situation of the potential isolated user.

The adjacent area may refer to the surrounding area of the user in an isolated state, for example, the adjacent area may be an area within a preset distance (such as 1.2km, 2.5km, etc.) of the user.

In some embodiments, the adjacent area may be determined according to the user's adjacent stores, for example, the user's adjacent stores may be determined according to the user's location information, and the adjacent areas may be determined based on the radiation range of the adjacent stores. Among them, the radiation range of adjacent stores can refer to the area where the store can provide shopping services, and the radiation range can be related to the size, geographical location and traffic conditions of the store.

Potential quarantined users may refer to users who may be in quarantine, for example, potential quarantined users may include users with an evaluation score of not less than 60. For the specific method of judging the isolation state, reference may be made to step 310 and the related description of FIG. 4 , which will not be repeated here.

The reference commodity can be the commodity purchased by the potential quarantined user. For example, if the consumption situation of the potential quarantined user A includes the purchase record of ready-to-eat food, the ready-to-eat food can be used as the reference commodity.

In some embodiments the reference item may be determined based on the purchases of potential quarantined users. For example, the recommendation module 220 may obtain the historical purchase information of the potential isolated user from the server according to the determined potential isolated user, and determine the reference product according to the product information in the historical purchase information. Among them, all the commodities in the historically purchased commodities can be used as reference commodities, and the quarantined commodities can also be used as reference commodities, which can be determined according to the actual situation.

Step 620: Adjust the weight of the recommended product based on the reference product.

In some embodiments, the recommendation module 220 may determine recommended commodities through a recommendation model, and adjust the weights of the recommended commodities according to the reference commodities. In some embodiments, the recommendation module 220 may adjust the weight of the recommended product according to the degree of matching between the recommended product and the reference product, and the recommended product with high weight may be displayed or marked with priority, wherein the matching degree of the recommended product and the reference product may be determined by The distance between the two feature vectors is determined. For example, the recommendation module 220 determines that commodity A, commodity B, commodity C, and commodity D are recommended commodities, wherein, the reference commodities determined according to the consumption situation of the potential isolated user are commodity B and commodity D, and can be added when recommending commodities. The weight of Commodity B and Commodity D. For example, Commodity B and Commodity D are displayed first, or the words "other quarantined/home users have also purchased this commodity" are displayed on the commodity labels of Commodity B and Commodity D to remind users.

For details on determining the recommended products, please refer to FIG. 3 , FIG. 5 and their related descriptions.

It should be noted that the above descriptions about the processes 300-600 are only for example and illustration, and do not limit the scope of application of this specification. Various modifications and changes can be made to the processes 300-600 to those skilled in the art under the guidance of this specification. However, such corrections and changes are still within the scope of this specification.

The possible beneficial effects of some embodiments of this specification include, but are not limited to: (1) Judging whether the user is in an isolation state, and providing users with accurate product recommendations based on the user isolation state. (2) By comparing the consumption habits of users in different time periods, the accuracy of judging whether the user is in an isolated state is improved. (3) Recommend products for users based on the recommendation model, which can make the recommended products meet the user's preferences and improve the accuracy of product recommendation. (4) Considering the impact of potential isolated users in stores and surrounding areas, it is easier to form group purchases when recommending products, and it is convenient for stores to specify sales plans.

It should be noted that different embodiments may have different beneficial effects, and in different embodiments, the possible beneficial effects may be any one or a combination of the above, or any other possible beneficial effects.

The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is merely an example, and does not constitute a limitation of the present specification. Although not explicitly described herein, various modifications, improvements and corrections to this specification may occur to those skilled in the art. Such modifications, improvements, and corrections are suggested in this specification, so such modifications, improvements, and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

Meanwhile, the present specification uses specific words to describe the embodiments of the present specification. References such as "one embodiment," "an embodiment," and/or "some embodiments" mean a certain feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various places in this specification are not necessarily referring to the same embodiment . Furthermore, certain features, structures or characteristics of the one or more embodiments of this specification may be combined as appropriate.

Furthermore, unless explicitly stated in the claims, the order of processing elements and sequences described in this specification, the use of alphanumerics, or the use of other names is not intended to limit the order of the processes and methods of this specification. While the foregoing disclosure discusses by way of various examples some embodiments of the invention that are presently believed to be useful, it is to be understood that such details are for purposes of illustration only and that the appended claims are not limited to the disclosed embodiments, but rather The requirements are intended to cover all modifications and equivalent combinations falling within the spirit and scope of the embodiments of this specification. For example, although the system components described above may be implemented by hardware devices, it may also be implemented by software-only solutions, such as installing the described systems on existing servers or mobile devices.

Similarly, it should be noted that, in order to simplify the expressions disclosed in this specification and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of this specification, various features may sometimes be combined into one embodiment, in the drawings or descriptions thereof. However, this method of disclosure does not imply that the subject matter of the description requires more features than are recited in the claims. In practice, there are fewer features of an embodiment than all of the features of a single embodiment disclosed above.

Some examples use numbers to describe quantities of components and attributes, it should be understood that such numbers used to describe the examples, in some examples, use the modifiers "about", "approximately" or "substantially" to retouch. Unless stated otherwise, "about", "approximately" or "substantially" means that a variation of ±20% is allowed for the stated number. Accordingly, in some embodiments, the numerical parameters set forth in the specification and claims are approximations that can vary depending upon the desired characteristics of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and use a general digit reservation method. Notwithstanding that the numerical fields and parameters used in some embodiments of this specification to confirm the breadth of their ranges are approximations, in specific embodiments, such numerical values are set as precisely as practicable.

For each patent, patent application, patent application publication, and other material, such as articles, books, specifications, publications, documents, etc., cited in this specification, the entire contents of which are hereby incorporated by reference into this specification are hereby incorporated by reference. Application history documents that are inconsistent with or conflict with the contents of this specification are excluded, as are documents (currently or hereafter appended to this specification) limiting the broadest scope of the claims of this specification. It should be noted that, if there is any inconsistency or conflict between the descriptions, definitions and/or use of terms in the accompanying materials of this specification and the contents of this specification, the descriptions, definitions and/or use of terms in this specification shall prevail .

Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations are also possible within the scope of this specification. Accordingly, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to those expressly introduced and described in this specification.

Claims (10)

1. A product recommendation method based on machine learning, comprising: Determine whether the user is in isolation based on the user's consumption situation and location information within the first time; In response to the user being in an isolated state, a recommended product to be pushed to the user is determined through a recommendation model. 2. The product recommendation method according to claim 1, wherein determining whether the user is in an isolated state based on the user's consumption situation and location information in the first time, comprising: An evaluation score of the user is determined based on the difference between the consumption habits of the user in the first time period and the consumption habits of the user in a second time period, where the second time period is the first time period a previous period of the same length; Whether the user is in quarantine is determined according to the evaluation score of the user. 3. The product recommendation method according to claim 1, wherein in response to the user being in an isolated state, determining the recommended product to be pushed to the user through a recommendation model, comprising: Based on store location information and store epidemic prevention assessment information, determine candidate recommended products; The recommended product is determined from the candidate recommended products through the recommendation model. 4. The product recommendation method according to claim 1, wherein in response to the user being in an isolated state, determining the recommended product to be pushed to the user through a recommendation model, comprising: based on the location information of the user, determining a potential quarantined user in the vicinity; determining the consumption situation of the potential quarantined user, and determining the reference commodity based on the consumption situation of the potential quarantined user; Based on the reference item, the recommended item is determined through a recommendation model. 5. A product recommendation system based on machine learning, comprising: a judgment module, used for judging whether the user is in an isolated state based on the user's consumption situation and location information within the first time; A recommendation module, configured to determine, through a recommendation model, a commodity recommended to the user in response to the user being in an isolated state. 6. The product recommendation system according to claim 5, wherein the judging module is further used for: Determine the user's evaluation score based on the difference between the user's purchasing habits in the first time period and the user's purchasing habits in a second time period, where the second time period is the first time period a previous period of the same length; Whether the user is in quarantine is determined according to the evaluation score of the user. 7. The product recommendation system according to claim 5, wherein the recommendation module is further used for: Based on store location information and store epidemic prevention assessment information, determine candidate recommended products; The recommended product is determined from the candidate recommended products through a recommendation model. 8. The product recommendation system according to claim 5, wherein the recommendation module is further used for: based on the location information of the user, determining a potential quarantined user in the vicinity; determining the consumption situation of the potential quarantined user, and determining the reference commodity based on the consumption situation of the potential quarantined user; Based on the reference item, the recommended item is determined through a recommendation model. 9 . An apparatus for recommending products based on machine learning, comprising a processor configured to execute the method for recommending products based on machine learning according to any one of claims 1 to 4 . 10 . 10. A computer-readable storage medium, the storage medium stores computer instructions, and after the computer reads the computer instructions in the storage medium, the computer executes the product recommendation based on machine learning according to any one of claims 1 to 4 method.

Images