A method and system for recommending instant retail products, and storage medium

Abstract

The present invention discloses a method and system for recommending instant retail goods, and a storage medium, and belongs to the technical field of goods recommendation. The method includes: obtaining the ratio between the user's attention to logistics speed and quality to form a user portrait; allocating the weight of the recommendation index according to the user portrait; calculating the urgency according to the product name, the system time when the user enters the search keyword, and the geographical location; adjusting the weight of the recommendation index according to the urgency; scoring the merchants who sell the goods required by the user, and recommending the merchants to the user; and correcting the user portrait according to the user's post-purchase evaluation and the acceptance of the recommended goods. The present invention does not rely on artificial intelligence models and gets rid of the dependence on chips and large computing power. Despite the lack of artificial intelligence support, the present invention can still predict the user's needs more accurately, thereby recommending the most suitable merchants to the user.

Description

Technical Field

[0001] The present invention relates to the field of commodity recommendation technology, and in particular to an instant retail commodity recommendation method and system, and a storage medium. Background Art

[0002] Instant retail is a retail format that meets local immediate needs by placing orders online and fulfilling orders offline, relying on local retail supply. Instant retail fills the "vacuum zone" of online and offline integration. "Localization" is a prominent feature of instant retail, which realizes the online transaction process and facilitates fulfillment and delivery.

[0003] In order to improve local supply capabilities and expand consumer demand, each sales platform has deployed a product recommendation strategy, which can recommend similar products of various brands and users who sell such products to users based on the products entered by the users.

[0004] Existing recommendation methods generally make recommendations based on the distance or comprehensive rating of stores when users search for the desired products, or make recommendations based on artificial intelligence models and big data, such as JD.com's "Hourly Delivery" and Meituan Waimai.

[0005] In actual operation, it is found that although the brute force recommendation method has low requirements, the recommended items often do not meet the requirements of consumers. Although the products recommended by artificial intelligence models and big data are more accurate, as the user group increases and the data becomes larger, the calculation through artificial intelligence models will consume huge computing power and reduce the operation speed of the entire system.

[0006] For example, Chinese patent application CNCN202311444961.7 provides a product recommendation model training method and a product recommendation method, including: determining a target product corresponding to a user based on a user's click sequence, the click sequence including the product identifier of the product clicked by the user, and the target product and the product corresponding to the click sequence belong to products on different sales platforms; constructing training samples, one training sample including a click sequence corresponding to a user and the product identifier of the target product; training a pre-constructed product recommendation model based on each of the training samples until the product recommendation model meets the requirements, thereby obtaining the trained product recommendation model.

[0007] For another example, Chinese patent application CN201710833187.7 provides a method for displaying all product results after a user enters product information. The method comprises: entering product information, including product image information or a combination of product image information and text information; extracting image features from the product image information using a pre-trained neural network to obtain product attributes corresponding to the image features; and / or extracting keywords from the product text information to obtain product attributes corresponding to the keywords; matching the product attributes obtained from the product information with the product attributes in an established product knowledge graph to determine a target product having the product attributes.

[0008] The above existing technologies all require the construction of a large number of training samples in advance to train the pre-constructed neural network model, which means that they are dependent on artificial intelligence. The dependence on artificial intelligence is actually the dependence on computing power, more specifically, the dependence on chips. Therefore, how to get rid of the dependence on chips is an urgent problem to be solved. Summary of the invention

[0009] The purpose of the present invention is to provide an instant retail product recommendation method and system, and a storage medium, which partially solve or alleviate the above-mentioned deficiencies in the prior art, and can recommend the most suitable designated products to users as much as possible on the basis of reducing computing power consumption, thereby improving user experience.

[0010] In order to solve the above-mentioned technical problems, the present invention specifically adopts the following technical solutions: a method for recommending instant retail products, comprising:

[0011] Obtain the ratio of users’ concerns about logistics speed and quality to form a user profile;

[0012] Allocate weights of recommendation indicators according to user portraits, where the recommendation indicators include logistics speed, quality, after-sales service, and price;

[0013] Obtain the product name of the product required by the user according to the search keyword entered by the user, and collect the system time and geographical location when the user enters the search keyword;

[0014] Calculate the urgency according to the product name, the system time and the geographical location;

[0015] Adjusting the weight of the recommended indicator according to the urgency;

[0016] Scoring a merchant that sells the product required by the user according to the recommendation index, and recommending the product sold by the merchant to the user based on the merchant score;

[0017] The user portrait is modified based on the user's post-purchase evaluation and acceptance of the recommended products.

[0018] As an improvement, the step of obtaining the ratio between the user's attention to logistics speed and quality specifically includes:

[0019] When a user registers, a progress bar is displayed on the user registration page; one end of the progress bar is the logistics speed, and the other end is the quality; a slider that can slide along the progress bar is provided on the progress bar; the user changes the ratio between the logistics speed end and the quality end by dragging the slider, and the sum of the material speed end ratio and the quality end ratio is a constant value.

[0020] As an improvement, the steps of allocating weights of recommendation indicators according to user portraits specifically include:

[0021] Initialize the weights of the recommended indicators and assign an initial weight to each recommended indicator;

[0022] The logistics speed weight and quality weight are reallocated according to the ratio between users' concerns about logistics speed and quality.

[0023] As an improvement, the step of calculating the urgency according to the product name, the system time and the geographical location specifically includes:

[0024] A plurality of urgency scoring templates are preset to construct a template library, wherein the urgency scoring template performs urgency scoring according to the input commodity type, purchase time type, and delivery address type;

[0025] Classify the goods according to the product name to obtain the product type; classify the time of purchase of the goods according to the system time to obtain the purchase time type; classify the address of purchase of the goods according to the geographical location to obtain the delivery address type;

[0026] An urgency score is obtained by matching an urgency score template from a template library based on the commodity type, the purchase time type, and the delivery address type.

[0027] As an improvement, the commodity types include medicines, food, and daily necessities; the purchase time types include working hours, meal time, meal preparation time, after-meal time, and night time; the delivery address types include home address, work address, and out-of-town address.

[0028] As an improvement, methods for assigning weights to recommendation indicators according to urgency include:

[0029] The logistics speed weight is K1, the quality weight is K2, the after-sales service weight is K3, and the price weight is K4;

[0030] When the urgency score is greater than the first urgency threshold, K1=(K1+K2)*80%, K2=(K1+K2)*20%, K3=(K3+K4)*80%, K4=(K3+K4)*20%, K1+K2+K3+K4=1;

[0031] When the urgency score is between the second urgency threshold and the first urgency threshold, K1=(K1+K2)*60%, K2=(K1+K2)*40%, K3=(K3+K4)*60%, K4=(K3+K4)*40%, K1+K2+K3+K4=1;

[0032] When the urgency score is between the third urgency threshold and the second urgency threshold, K1=(K1+K2)*40%, K2=(K1+K2)*60%, K3=(K3+K4)*40%, K4=(K3+K4)*60%, K1+K2+K3+K4=1;

[0033] When the urgency score is less than the third urgency threshold, K1=(K1+K2)*20%, K2=(K1+K2)*80%, K3=(K3+K4)*20%, K4=(K3+K4)*80%, K1+K2+K3+K4=1.

[0034] As an improvement, the method of scoring merchants selling the products required by users according to the recommendation indicators includes:

[0035] Obtain user reviews of the merchant, including individual scores of each recommendation indicator in the user reviews;

[0036] Calculate the average score of each recommendation indicator for all users;

[0037] Multiply the average of the individual scores of each recommended indicator by the weight of the recommended indicator to obtain the adjusted score;

[0038] The adjusted scores of all recommendation indicators are summed up to obtain the merchant score.

[0039] As an improvement, the steps of recommending merchants to users based on merchant ratings specifically include:

[0040] Sort the merchants in descending order according to their ratings, and recommend to the user the products needed by the user sold by the merchants whose rankings are higher than the ranking threshold; or

[0041] The products required by the users sold by merchants whose merchant scores are higher than the score threshold are recommended to the users.

[0042] As an improvement, the steps to modify the user profile are as follows:

[0043] When the number of times that the user does not accept the recommended merchants exceeds the number threshold, the user profile is adjusted; the adjustment steps specifically include:

[0044] The average score A1 of the user for a certain recommendation indicator is compared with the average score A2 of all other users for the certain indicator. When A1<A2, the weight of the recommendation indicator is increased by a first percentage; when A1≥A2 and A1 is not a full score, the weight of the recommendation indicator is increased by a second percentage; when A1 is a full score, the weight of the recommendation indicator is reduced by a third percentage; wherein the first percentage>the second percentage.

[0045] The present invention also provides an instant retail commodity recommendation system, comprising:

[0046] User portrait building module, used to obtain the ratio of users' attention to logistics speed and quality, and form a user portrait;

[0047] A weight allocation module is used to allocate weights of recommendation indicators according to user portraits, and the recommendation indicators include logistics speed, quality, after-sales service, and price;

[0048] The urgency calculation module is used to obtain the product name of the product required by the user according to the search keyword input by the user, and collect the system time and geographical location when the user inputs the search keyword; calculate the urgency according to the product name, the system time and geographical location when the user inputs the search keyword;

[0049] A weight adjustment module is used to adjust the weight of the recommendation indicator according to the urgency;

[0050] The recommendation module is used to rate the merchants selling the goods required by the user according to the recommendation index, and recommend the goods sold by the merchants to the user based on the merchant ratings;

[0051] The user portrait correction module is used to correct the user portrait based on the user's post-purchase evaluation and acceptance of the recommended products.

[0052] The present invention also provides a computer-readable storage medium storing computer program instructions, which are used to implement the above-mentioned instant retail product recommendation method when the computer program instructions are executed by at least one processor.

[0053] The present invention is beneficial in that: the present invention forms a user portrait by obtaining the ratio between the user's attention to logistics speed and quality. The weight of the recommendation index is assigned as a keynote according to the user portrait. The urgency is calculated according to the search keywords input by the user, and the weight of the recommendation index is adjusted for this shopping according to the urgency. The merchant who sells the goods required by the user is rated according to the recommendation index, and the merchant is recommended to the user based on the merchant rating. Finally, the user portrait is modified according to the user's evaluation after purchase and the acceptance of the recommended goods, forming a closed-loop control. Different from regular online shopping (for example, platforms such as JD.com or Taobao), instant retail products usually have much higher logistics requirements than regular online shopping (for example, delivery the next day or within three days). The reason for the high logistics requirements is that the urgency of the demand for goods is much higher than that of regular online shopping. Therefore, when the user enters the required goods, the corresponding recommendation indicator weights are dynamically adjusted according to the corresponding urgency and scored. That is, the merchants are ranked from the perspective of user needs. Compared with the static total score obtained by simply evaluating the goods and services provided by the merchants (that is, based on the consumer's evaluation results of all goods, logistics, services, etc.), it is more in line with the instant retail sales model and can recommend more suitable goods to users.

[0054] The present invention does not rely on artificial intelligence models, and is free from reliance on chips and large computing power. Despite the lack of artificial intelligence support, the present invention can still accurately predict the needs of users, thereby recommending the most suitable merchants for users. The present invention can also self-correct according to user feedback, thereby gradually improving itself and making it more in line with the user's true appearance.

[0055] In addition, the present invention uses the modes of progress bar and slider to obtain the user's attention to logistics speed and quality, which is more easily accepted by users. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale. Obviously, the drawings described below are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without paying creative labor.

[0057] Figure 1 A flowchart of a method for recommending instant retail products according to an embodiment of the present invention;

[0058] Figure 2The present invention is a schematic diagram of the structure of an instant retail product recommendation system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0060] Herein, suffixes such as "module", "component" or "unit" used to represent elements are only used to facilitate the description of the present invention, and have no specific meanings by themselves. Therefore, "module", "component" or "unit" can be used mixedly.

[0061] In this document, the terms "upper", "lower", "inner", "outer", "front", "back", "one end", "the other end" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0062] In this document, unless otherwise clearly specified and limited, the terms "installed", "provided with", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0063] Herein "and / or" includes any and all combinations of one or more of the associated listed items.

[0064] Herein, "plurality" means two or more than two, ie, it includes two, three, four, five, etc.

[0065] Embodiment 1

[0066] Prior art Chinese patent application CNCN202311444961.7 provides a product recommendation model training method and a product recommendation method, including: determining a target product corresponding to a user according to a user's click sequence, the click sequence including a product identifier of a product clicked by the user, and the target product and the product corresponding to the click sequence belong to products on different sales platforms; constructing training samples, one training sample including a click sequence corresponding to a user and a product identifier of the target product; training a pre-constructed product recommendation model based on each of the training samples until the product recommendation model meets the requirements, thereby obtaining the trained product recommendation model.

[0067] In the above-mentioned prior art, it is necessary to construct training samples and train the pre-constructed product recommendation model based on each of the training samples. After the product recommendation model meets the requirements, it is applied to the actual product recommendation. And as the number of users increases, more parameters need to be considered, and the computing power consumed in the application becomes greater.

[0068] Another example is the prior art Chinese patent application CN201710833187.7, which provides a method for displaying all product results after a user enters product information. The method includes: entering product information, including product image information or a combination of product image information and text information; extracting image features from the product image information using a pre-trained neural network to obtain product attributes corresponding to the image features; and / or extracting keywords from the product text information to obtain product attributes corresponding to the keywords; matching the product attributes obtained from the product information with the product attributes in the established product knowledge graph to determine the target product with the product attributes. In the above prior art, a large number of training samples still need to be pre-built to train the pre-built neural network model for image recognition.

[0069] like Figure 1 As shown, the present invention provides an instant retail commodity recommendation method that can be run without an artificial intelligence model. The recommendation method in this embodiment recommends based on the user's input of the desired commodity, and its specific steps include:

[0070] S101 obtains the ratio of users’ concerns about logistics speed and quality to form a user profile.

[0071] Instant retail refers to a retail model that meets consumer needs in a fast and convenient way by integrating online and offline channels. Its main characteristics include: 1. Real-time: Instant retail emphasizes responding to consumer needs in the shortest possible time. By using technological means, it can achieve fast order processing, goods delivery and customer service. 2. Seamless connection: Instant retail emphasizes the seamless connection between online and offline, and through the integration of multiple channels, it eliminates the time and space restrictions of shopping and provides a consistent shopping experience.

[0072] Compared with general online sales, instant retail sales focus more on fast-moving consumer goods such as food, medicine and daily necessities. In terms of logistics, instant retail emphasizes fast delivery, and its logistics time ranges from tens of minutes to several hours.

[0073] Therefore, for instant retail consumers, the main factors that play a leading role in their consumption are logistics speed and product quality. Based on this, in order to more accurately recommend products to users, this embodiment first needs to obtain the user's attention to logistics speed and quality. In order to better express the difference between the attention to logistics speed and quality, this embodiment uses a ratio to quantify the attention to the two.

[0074] In actual applications, there are many ways to obtain the proportion of users' attention to logistics speed and quality, but the existing methods either require filling in data or multiple clicks, which are cumbersome. In today's fast-paced life, a large number of users cannot patiently input as required.

[0075] In order to solve this problem, in this implementation, when a user registers, a progress bar is displayed on the user registration page; one end of the progress bar is the logistics speed, and the other end is the quality; a slider that can slide along the progress bar is provided on the progress bar; the user changes the ratio between the logistics speed end and the quality end by dragging the slider, and the sum of the material speed end ratio and the quality end ratio is fixed.

[0076] When the user pulls the slider, the changes in the ratio are displayed at both ends. For example, when the slider is initially in the middle, the material speed end ratio and the mass end ratio are both 50%. When the user pulls the slider toward the material speed end, the material speed end becomes shorter and the ratio decreases; while the mass end lengthens and the ratio increases synchronously.

[0077] The above method is very convenient and intuitive when inputting different proportions, which is easy for users to use and also makes it easier to obtain data in the background.

[0078] Of course, it is possible that some users are still unwilling to provide the ratio between their concern for logistics speed and quality. In this case, you only need to directly match a default template for such users.

[0079] S102 allocates weights of recommendation indicators according to the user portrait, and the recommendation indicators include logistics speed, quality, after-sales service, and price.

[0080] In addition to logistics speed and quality, instant retail users also pay attention to after-sales service and price. Of course, the attention paid to after-sales service and price is far less than that paid to logistics speed and quality.

[0081] The attention paid to logistics speed and quality is also related to the attention paid to after-sales service and price. Generally speaking, in instant retail, users who pay attention to logistics speed are generally not very sensitive to price. For example, users who buy medicine late at night only care about logistics speed, and they will not care too much about the price as long as it is within a reasonable range. On the contrary, users who pay attention to quality do not pay much attention to after-sales service. The reason is that in instant retail, the products with greater quality differences are food and other products that do not have much after-sales service.

[0082] In this embodiment, the method for allocating weights of recommendation indicators according to user portraits includes:

[0083] S1021 initializes the weights of the recommended indicators and assigns an initial weight to each recommended indicator.

[0084] In order to avoid the case where the user does not provide the ratio between the degree of concern for logistics speed and quality, this embodiment directly matches a default template for such users. This step can also use the template when initializing the user.

[0085] For example, the logistics speed has a weight of 40%, the quality has a weight of 40%, the after-sales service has a weight of 10%, and the price has a weight of 10%.

[0086] S1022 reallocates the logistics speed weight and the quality weight according to the ratio between the user's concerns about logistics speed and quality.

[0087] In step S101, a user portrait is obtained. For example, if the ratio of the user's attention to logistics speed and quality is 60%:40%, then the logistics speed weight and quality weight are reallocated according to the ratio. After the reallocation, the logistics speed weight is 48%, and the quality weight is 32%.

[0088] S103 obtains the product name of the product required by the user according to the search keyword input by the user, and collects the system time and geographical location when the user inputs the search keyword.

[0089] When a user purchases a product, he or she will first enter a keyword about the product in the search box, so the product name of the product that the user needs to purchase can be easily parsed based on the keyword. When entering the search keyword, the system time and geographic location are obtained synchronously.

[0090] S104 calculates the urgency based on the product name, the system time when the user inputs the search keyword, and the geographical location.

[0091] The urgency of the user's purchase of the product can be roughly determined based on the product name, search time, and address of the searched product. For example, buying medicine at home at midnight is likely to be very urgent. Searching for food at work during working hours is likely not too urgent. Specifically, this step package includes:

[0092] S1041 presets a plurality of urgency scoring templates to construct a template library, wherein the urgency scoring template performs urgency scoring according to the input commodity type, purchase time type, and delivery address type.

[0093] In this step, several scoring templates can be preset in advance, for example: product type (drugs) + purchase time type (night) + delivery address (home) = urgency score (100); product type (food) + purchase time type (working hours) + delivery address (workplace) = urgency score (20);

[0094] According to the permutations and combinations, several scoring templates can be constructed for backup.

[0095] In this embodiment, the delivery address defaults to the geographic location when the search keyword is entered.

[0096] S1042 classifies the goods according to the goods names to obtain the goods types; classifies the time of purchasing the goods according to the system time to obtain the purchase time type; and classifies the address of purchasing the goods according to the geographical location to obtain the delivery address type.

[0097] In this embodiment, the preset classification is: the commodity types include medicines, food, and daily necessities; the purchase time types include working hours, meal time, meal preparation time, after-meal time, and night time; the delivery address types include home address, work address, and out-of-town address.

[0098] According to the above classification, multiple scoring templates can be constructed. It is understandable that, based on the above classification, the commodity type, time type and location type can be further refined. The more detailed the classification, the more scoring templates can be constructed, and the more accurate the urgency score will be.

[0099] S1043 matches an urgency score template from a template library based on the commodity type, purchase time type, and delivery address type to obtain an urgency score.

[0100] The product type, time type and address type of the current user are matched with corresponding templates from the template library to obtain the urgency score of the current user to purchase the product.

[0101] S105 adjusts the weight of the recommendation indicator according to the urgency.

[0102] The purpose of this step is to further adjust the weights assigned to the user profile based on the urgency, so as to make adaptive adjustments for this shopping with the user profile as the general tone. The specific steps include:

[0103] The logistics speed weight is K1, the quality weight is K2, the after-sales service weight is K3, and the price weight is K4;

[0104] When the urgency score is greater than the first urgency threshold, K1=(K1+K2)*80%, K2=(K1+K2)*20%, K3=(K3+K4)*80%, K4=(K3+K4)*20%, K1+K2+K3+K4=1;

[0105] When the urgency score is between the second urgency threshold and the first urgency threshold, K1=(K1+K2)*60%, K2=(K1+K2)*40%, K3=(K3+K4)*60%, K4=(K3+K4)*40%, K1+K2+K3+K4=1;

[0106] When the urgency score is between the third urgency threshold and the second urgency threshold, K1=(K1+K2)*40%, K2=(K1+K2)*60%, K3=(K3+K4)*40%, K4=(K3+K4)*60%, K1+K2+K3+K4=1;

[0107] When the urgency score is less than the third urgency threshold, K1=(K1+K2)*20%, K2=(K1+K2)*80%, K3=(K3+K4)*20%, K4=(K3+K4)*80%, K1+K2+K3+K4=1.

[0108] It can be foreseen that the first urgency threshold>the second urgency threshold>the third urgency threshold, thereby forming three continuous intervals.

[0109] For example, if a user's current urgency score is 92, which is greater than the first urgency threshold, the weights assigned to this user based on the user profile are: K1=48%, K2=32%, K3=10%, K4=10%.

[0110] According to the above adjustment rules, K1=64%, K2=16%, K3=16%, K4=4%.

[0111] Of course, including the weight allocation based on the user portrait in step S102 and the weight adjustment in this step, the specific values ​​of the weights can be adaptively modified according to the specific circumstances, and this step does not impose any specific restrictions or solidification.

[0112] S106 scores merchants that sell the products required by the user according to the recommendation index, and recommends merchants to the user based on the merchant scores.

[0113] The methods for obtaining merchant ratings in this step specifically include:

[0114] S1061 obtains user evaluations of the merchant, and the individual scores of each recommendation indicator of the user evaluations.

[0115] Suppose a merchant has 1,000 user reviews, and each user review consists of a single score for each recommended indicator, namely, a separate score for logistics speed, quality, after-sales service, and price.

[0116] S1062 calculates the average of the individual scores of each recommendation indicator of all users.

[0117] The average value of the individual score of each recommendation indicator can be obtained by adding up the individual scores of each recommendation indicator in 1,000 user reviews and dividing by the corresponding number.

[0118] S1063 multiplies the average of the individual scores of each recommendation indicator by the weight of the recommendation indicator to obtain an adjusted score.

[0119] For example, among 1,000 user reviews, the average scores of logistics speed, quality, after-sales service and price are 95, 90, 100 and 98 respectively. The weights adjusted in step S105 are multiplied by the corresponding recommendation indicators, namely 95*64%=60.8, 90*16%=14.4, 100*16%=16, and 98*4%=3.92.

[0120] S1064 sums the adjusted scores of all recommendation indicators to obtain the merchant score.

[0121] The merchant rating can be obtained by summing up the individual weighted scores obtained in the above steps, that is, 60.8+14.4+16+3.92=95.12.

[0122] After obtaining the merchant ratings, this embodiment provides two merchant recommendation methods, including:

[0123] First, the merchants are sorted in descending order according to their ratings, and the goods sold by merchants whose rankings are higher than the ranking threshold are recommended to the user. For example, the top 10 merchants with the highest ratings are recommended to the user.

[0124] The second is to recommend to users the products needed by users that are sold by merchants whose merchant scores are higher than the score threshold. For example, merchants whose merchant scores are higher than 90 are recommended to users.

[0125] It is understandable that the recommendation is made to the user by displaying the goods sold by the merchant, rather than directly recommending the merchant.

[0126] S107 modifies the user profile based on the user's post-purchase evaluation and acceptance of the recommended products.

[0127] In this implementation, after recommending a merchant to a user, the user profile can also be modified based on the user's feedback, thereby achieving closed-loop management, specifically including:

[0128] When the number of times that the user does not accept the recommended merchant exceeds the number threshold, the method for adjusting the user profile includes:

[0129] The average score A1 of the user for a certain recommendation indicator is compared with the average score A2 of all other users for the certain indicator. When A1<A2, the weight of the recommendation indicator is increased by a first percentage; when A1≥A2 and A1 is not a full score, the weight of the recommendation indicator is increased by a second percentage; when A1 is a full score, the weight of the recommendation indicator is reduced by a third percentage; wherein the first percentage>the second percentage.

[0130] For example, when a user does not choose a merchant recommended by the system for purchase three times in a row, it indicates that there is a problem in the weight setting of the recommendation parameters, so adjustments are made.

[0131] In this embodiment, the basis for adjustment is the user's evaluation. This evaluation is not only the evaluation of a single service, but also a comprehensive evaluation of all users.

[0132] For example, when the average user evaluation of logistics speed is 80, while the average score of other users is 90, it means that the user is very demanding on logistics speed and the weight of logistics speed needs to be significantly increased, for example, by 30% on the original basis.

[0133] For example, another user's average score for logistics speed is 95, which is higher than the average score of 90 for other users. This means that the user is still relatively concerned about logistics speed, but is more easily satisfied. Therefore, the weight of logistics speed is slightly increased, for example, by 5% on the original basis.

[0134] For example, the average of another user's evaluation of logistics speed is full marks, which means that the user does not care much about the logistics speed and just gives good reviews out of habit. Therefore, the weight of the logistics speed can be appropriately reduced, for example, by 10% on the original basis.

[0135] It can be foreseen that after the user portrait is corrected, the next recommendation activity will be based on the corrected user portrait.

[0136] Embodiment 2

[0137] like Figure 2 As shown, the present invention also provides an instant retail commodity recommendation system, comprising:

[0138] User portrait building module, used to obtain the ratio of users' attention to logistics speed and quality, and form a user portrait;

[0139] A weight allocation module is used to allocate weights of recommendation indicators according to user portraits, and the recommendation indicators include logistics speed, quality, after-sales service, and price;

[0140] The urgency calculation module is used to obtain the product name of the product required by the user according to the search keyword input by the user, and collect the system time and geographical location when the user inputs the search keyword; calculate the urgency according to the product name, the system time and geographical location when the user inputs the search keyword;

[0141] A weight adjustment module is used to adjust the weight of the recommendation indicator according to the urgency;

[0142] The recommendation module is used to rate the merchants selling the goods required by the user according to the recommendation index, and recommend the goods sold by the merchants to the user based on the merchant ratings;

[0143] The user portrait correction module is used to correct the user portrait based on the user's post-purchase evaluation and acceptance of the recommended products.

[0144] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0145] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a computer terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0146] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A method for recommending instant retail products, characterized in that include: Obtain the ratio of users’ concerns about logistics speed and quality to form a user profile; Allocate weights of recommendation indicators according to user portraits, where the recommendation indicators include logistics speed, quality, after-sales service, and price; Obtain the product name of the product required by the user according to the search keyword entered by the user, and collect the system time and geographical location when the user enters the search keyword; Calculate the urgency according to the product name, the system time and the geographical location; Adjusting the weight of the recommended indicator according to the urgency; Scoring a merchant that sells the product required by the user according to the recommendation index, and recommending the product sold by the merchant to the user based on the merchant score; Modify the user profile based on the user's post-purchase evaluation and acceptance of the recommended products; The step of calculating the urgency according to the product name, the system time and the geographical location specifically includes: A plurality of urgency scoring templates are preset to construct a template library, wherein the urgency scoring template performs urgency scoring according to the input commodity type, purchase time type, and delivery address type; Classify the goods according to the goods names to obtain the goods types; classify the time of purchase of the goods according to the system time to obtain the purchase time type; classify the address of purchase of the goods according to the geographical location to obtain the delivery address type; Matching an urgency score template from a template library based on the commodity type, the purchase time type, and the delivery address type to obtain an urgency score; The steps of assigning weights to the recommended indicators according to urgency include: The logistics speed weight is K1, the quality weight is K2, the after-sales service weight is K3, and the price weight is K4; When the urgency score is greater than the first urgency threshold, K1=(K1+K2)*80%, K2=(K1+K2)*20%, K3=(K3+K4)*80%, K4=(K3+K4)*20%, K1+K2+K3+K4=1; When the urgency score is between the second urgency threshold and the first urgency threshold, K1=(K1+K2)*60%, K2=(K1+K2)*40%, K3=(K3+K4)*60%, K4=(K3+K4)*40%, K1+K2+K3+K4=1; When the urgency score is between the third urgency threshold and the second urgency threshold, K1=(K1+K2)*40%, K2=(K1+K2)*60%, K3=(K3+K4)*40%, K4=(K3+K4)*60%, K1+K2+K3+K4=1; When the urgency score is less than the third urgency threshold, K1=(K1+K2)*20%, K2=(K1+K2)*80%, K3=(K3+K4)*20%, K4=(K3+K4)*80%, K1+K2+K3+K4=1.

2. A method for recommending instant retail products according to claim 1, characterized in that The step of obtaining the ratio between the user's attention to logistics speed and quality specifically includes: When a user registers, a progress bar is displayed on the user registration page; one end of the progress bar is the logistics speed, and the other end is the quality; a slider that can slide along the progress bar is provided on the progress bar; the user changes the ratio between the logistics speed end and the quality end by dragging the slider, and the sum of the logistics speed end ratio and the quality end ratio is a constant value.

3. The instant retail commodity recommendation method according to claim 1, characterized in that The steps of allocating weights of recommendation indicators according to user portraits include: Initialize the weights of the recommended indicators and assign an initial weight to each recommended indicator; The logistics speed weight and quality weight are reallocated according to the ratio between users' concerns about logistics speed and quality.

4. The instant retail commodity recommendation method according to claim 3, characterized in that: The commodity types include medicines, food, and daily necessities; the purchase time types include working hours, meal time, meal preparation time, after-meal time, and night time; the delivery address types include home address, work address, and out-of-town address.

5. The instant retail commodity recommendation method according to claim 1, characterized in that The steps of scoring merchants selling products required by users according to the recommendation indicators specifically include: Obtain user reviews of the merchant, including individual scores of each recommendation indicator in the user reviews; Calculate the average score of each recommendation indicator for all users; Multiply the average of the individual scores of each recommended indicator by the weight of the recommended indicator to obtain the adjusted score; The adjusted scores of all recommendation indicators are summed up to obtain the merchant score.

6. The instant retail commodity recommendation method according to claim 1, characterized in that The steps of recommending products to users based on merchant ratings include: Sort the merchants in descending order according to their ratings, and recommend to the user the products needed by the user sold by merchants whose rankings are higher than the ranking threshold; or The products required by the users sold by merchants whose merchant scores are higher than the score threshold are recommended to the users.

7. The instant retail commodity recommendation method according to claim 1, characterized in that The steps to modify the user portrait based on the user's post-purchase evaluation and acceptance of the recommended products are as follows: When the number of times that the user does not accept the recommended merchants exceeds the number threshold, the user profile is adjusted; the adjustment steps specifically include: The average score A1 of the user for a certain recommendation indicator is compared with the average score A2 of all other users for the certain indicator. When A1<A2, the weight of the recommendation indicator is increased by a first percentage; when A1≥A2 and A1 is not a full score, the weight of the recommendation indicator is increased by a second percentage; when A1 is a full score, the weight of the recommendation indicator is reduced by a third percentage; wherein the first percentage>the second percentage.

8. A computer-readable storage medium having computer program instructions stored therein, characterized in that: When the computer program instructions are executed by at least one processor, they are used to implement the instant retail product recommendation method as described in any one of claims 1 to 7.

9. An instant retail product recommendation system, characterized in that include: User portrait building module, used to obtain the ratio of users' attention to logistics speed and quality, and form a user portrait; A weight allocation module is used to allocate weights of recommendation indicators according to user portraits, and the recommendation indicators include logistics speed, quality, after-sales service, and price; The urgency calculation module is used to obtain the product name of the product required by the user according to the search keyword input by the user, and collect the system time and geographical location when the user inputs the search keyword; calculate the urgency according to the product name, the system time and geographical location when the user inputs the search keyword; A weight adjustment module is used to adjust the weight of the recommendation indicator according to the urgency; The recommendation module is used to rate the merchants selling the goods required by the user according to the recommendation index, and recommend the goods sold by the merchants to the user based on the merchant ratings; User portrait correction module, used to correct user portraits based on the user's post-purchase evaluation and acceptance of recommended products; The urgency calculation module is specifically used to preset a number of urgency scoring templates to build a template library, wherein the urgency scoring template performs urgency scoring according to the input commodity type, purchase time type, and delivery address type; and classifies commodities according to commodity names to obtain commodity types; classifies the time of commodity purchase according to system time to obtain purchase time types; classifies the address of commodity purchase according to geographical location to obtain delivery address types; and matches the urgency scoring template from the template library based on the commodity type, purchase time type, and delivery address type to obtain the urgency score; The weight adjustment module allocates weights to the recommendation indicators according to the urgency, specifically including: The logistics speed weight is K1, the quality weight is K2, the after-sales service weight is K3, and the price weight is K4; When the urgency score is greater than the first urgency threshold, K1=(K1+K2)*80%, K2=(K1+K2)*20%, K3=(K3+K4)*80%, K4=(K3+K4)*20%, K1+K2+K3+K4=1; When the urgency score is between the second urgency threshold and the first urgency threshold, K1=(K1+K2)*60%, K2=(K1+K2)*40%, K3=(K3+K4)*60%, K4=(K3+K4)*40%, K1+K2+K3+K4=1; When the urgency score is between the third urgency threshold and the second urgency threshold, K1=(K1+K2)*40%, K2=(K1+K2)*60%, K3=(K3+K4)*40%, K4=(K3+K4)*60%, K1+K2+K3+K4=1; When the urgency score is less than the third urgency threshold, K1=(K1+K2)*20%, K2=(K1+K2)*80%, K3=(K3+K4)*20%, K4=(K3+K4)*80%, K1+K2+K3+K4=1.

Images