Search response method and device

Abstract

The invention discloses a search response method and device. The method comprises the following steps: determining a target commodity and a recommended commodity according to a search term input by a user; the target commodity is located in a first area of the search result interface, and the recommended commodity is located in a second area of the search result interface; according to the relationship between the first path of recommended commodities in the recommended commodities and the search terms, determining the correlation of the first path of recommended commodities; according to the relationship between the commodity category of the second path of recommended commodities in the recommended commodities and the target category, determining the correlation of the second path of recommended commodities; for any recommended commodity, determining the display position of the recommended commodity in the second area according to the correlation of the recommended commodity and the tradeability of the recommended commodity; the transactionality represents the transaction conversion effect of the recommended commodity. By adopting the method, when the commodity recommendation sequence is determined, the correlation and trade performance of the commodities are fully considered, the diversity of browsing the commodities by the user is expanded, and the experience feeling of the user is improved.

Description

Technical Field

[0001] The present application relates to the field of e-commerce, and in particular to a search response method and device. Background Art

[0002] With the development of Internet technology, the search and recommendation functions of e-commerce platforms have gradually become an important means to improve user experience and promote business growth. Among them, the "Guess What You Like" module, as an innovative function that combines search and recommendation features, has received widespread attention.

[0003] The "Guess You Like" module is located on the search results interface and is an effective supplement to the search module. Its core is to recommend a variety of products to users through a balance of relevance, user experience and business value.

[0004] The "Guess You Like" module also recommends products to users through a process of recommendation and re-ranking. Current recommendation modules include vector-based recommendations and low-end recommendations from the search module. The re-ranking module further cross-merges and outputs recommended products after screening. Therefore, current solutions have shortcomings in terms of relevance, category breadth, and diversity. For example, vector-based recommendations fail to adequately consider product relevance, while low-end recommendations from the search module include fewer products and lack consideration for category breadth and product diversity. Summary of the Invention

[0005] The present application provides a search response method and device for fully considering the relevance and transaction nature of products when determining the order of product recommendations, broadening the diversity of products browsed by users and improving the user experience.

[0006] In a first aspect, an embodiment of the present application provides a search response method, which can be executed by a search response device, the method comprising: determining a target product pointed to by a search term and a recommended product corresponding to the search term based on a search term input by a user; the target product is located in a first area of ​​the search result interface, and the recommended product is located in a second area of ​​the search result interface; determining the relevance of a first recommended product among the recommended products based on a relationship between the first recommended product and the search term; determining the relevance of a second recommended product among the recommended products based on a relationship between the product category and the target category of the second recommended product; wherein the relationship between the first recommended product and the search term is stronger than the relationship between the second recommended product and the search term; the target category at least includes the product category of the target product; for any recommended product, determining the display position of the recommended product in the second area based on the relevance of the recommended product and the transactionability of the recommended product; the transactionability characterizes the transaction conversion effect of the recommended product.

[0007] Using this method, the relevance of recommended products is determined based on the relationship between the recommended product and the search term, or the relationship between the recommended product's category and the target category. The placement of the recommended product in the second area is then determined based on the transactional value of the recommended product and its relevance. This approach fully considers both the relevance and transactional value of the product when determining the placement of the recommended product, improving the user's shopping experience and increasing the conversion rate of the product.

[0008] In one possible implementation, the display position of the recommended product in the second area is determined based on the relevance of the recommended product and the tradability of the recommended product, including: calculating the ranking position of the recommended product based on the tradability of the recommended product and the relevance of the recommended product; wherein, if the number of target products is less than a first threshold, the importance of the relevance of the recommended product is higher than the importance of the tradability of the recommended product; if the number of target products is not less than the first threshold, the importance of the relevance of the recommended product is lower than the importance of the tradability of the recommended product; and determining the display position of the recommended product in the second area according to the ranking position of the recommended product.

[0009] Using the above method, if the number of target products is less than the first threshold, it means that the number of target products determined by the search terms is small, and the range of products that users can choose in the first area is small. In this case, the recommended products displayed in the second area are more likely to recommend products with higher relevance to the search terms. Therefore, when calculating the ranking position of the recommended products, the relevance of the recommended products is taken into consideration. On the contrary, if the number of target products is not less than the first threshold, it means that the number of target products determined by the search terms is large, and the range of products that users can choose in the first area is large enough. In this case, the recommended products displayed in the second area are more likely to recommend products with higher transaction value. In this way, the relevance and transaction value of products can be fully considered when the number of target products is different, thereby improving the category width and diversity of products browsed by users.

[0010] In one possible implementation, the display position of the recommended products in the second area is determined according to the sorting position of the recommended products, including: dividing each recommended product into multiple recommendation groups according to the sorting position of each recommended product; each recommendation group includes multiple recommended products; for each recommendation group, adjusting the sorting positions of the multiple recommended products in the recommendation group according to the product categories of the multiple recommended products in the recommendation group, wherein the product categories of adjacent recommended products in the adjusted recommendation group are different.

[0011] By adopting the above method, adjacent recommended products in the recommendation group are broken up into different product categories, avoiding the sequential display of products of the same product category, and improving the diversity of products browsed by users.

[0012] In one possible implementation, the number of recommendation groups is negatively correlated with the number of target products. The method further includes: if the number of recommended products is less than a second threshold, obtaining P popular products, where the popular products are determined based on historical orders of multiple users; and using the p popular products as the last p products in the second area.

[0013] By adopting the above method, even when the number of target products is small, it is still possible to combine popular products and recommend a sufficient number of products to users, thereby improving the user's shopping experience and increasing the conversion rate of products.

[0014] In one possible implementation, the target product and the first recommended product are determined by a search module, and the correlation between the target product and the search term is greater than the correlation between the first recommended product and the search term; the second recommended product is determined based on the recommendation module.

[0015] By adopting the above method, the recommended products are divided into the first recommended products and the second recommended products, so that when calculating the relevance of the products, different calculation methods are used, which can more accurately reflect the relevance of each product.

[0016] In one possible implementation, the relevance of the first group of recommended products in the recommended products is determined based on the relationship between the first group of recommended products and the search term, including: determining a first score of the first group of recommended products based on whether the first group of recommended products in the recommended products contains the search term; and / or determining a second score of the first group of recommended products based on whether the text matching value between the first group of recommended products in the recommended products and the search term exceeds a third threshold; and / or determining a third score of the first group of recommended products based on whether the relevance score output by the search module of the first group of recommended products is greater than a fourth threshold; and determining the relevance of the first group of recommended products based on at least one score of the first group of recommended products.

[0017] In one possible implementation, the target category includes a first target category set and a second target category set, the first target category set is a multi-level product category corresponding to at least one target product, and the second target category set is a multi-level product category of the product indicated by the search term; there is an affiliation between the high category level and the low category level; based on the relationship between the product category of the second-link recommended product and the target category in the recommended products, the relevance of the second-link recommended product is determined, including: for any second-link recommended product, determining the category level that matches the product category of the second-link recommended product from the target category, and determining the relevance of the second-link recommended product based on the matched category level; wherein, the category level is negatively correlated with the relevance; when the matched category level belongs to the first target category set, the represented correlation is higher than that of the second target category set.

[0018] In a second aspect, an embodiment of the present application provides a search response device, which includes a determination module, wherein the determination module is used to determine, based on a search term input by a user, a target product pointed to by the search term and a recommended product corresponding to the search term; the target product is located in a first area of ​​the search result interface, and the recommended product is located in a second area of ​​the search result interface; based on the relationship between a first recommended product in the recommended products and the search term, the relevance of the first recommended product is determined; based on the relationship between the product category of a second recommended product in the recommended products and the target category, the relevance of the second recommended product is determined; wherein the relationship between the first recommended product and the search term is stronger than the relationship between the second recommended product and the search term; the target category at least includes the product category of the target product; for any recommended product, the display position of the recommended product in the second area is determined based on the relevance of the recommended product and the transaction nature of the recommended product; the transaction nature represents the transaction conversion effect of the recommended product.

[0019] In one possible implementation, the device also includes a calculation module, which is used to calculate the ranking position of the recommended product based on the transactionability of the recommended product and the relevance of the recommended product; wherein, if the number of the target products is less than a first threshold, the importance of the relevance of the recommended product is higher than the importance of the transactionability of the recommended product; if the number of the target products is not less than the first threshold, the importance of the relevance of the recommended product is lower than the importance of the transactionability of the recommended product; the determination module is also used to determine the display position of the recommended product in the second area according to the ranking position of the recommended product.

[0020] In one possible implementation, the device further includes a grouping module, which is used to divide each recommended product into multiple recommendation groups according to the ranking position of each recommended product; each recommendation group includes multiple recommended products; the device further includes an adjustment module, which is used to adjust the ranking positions of the multiple recommended products in the recommendation group according to the product categories of the multiple recommended products in the recommendation group, wherein the product categories of adjacent recommended products in the adjusted recommendation group are different.

[0021] In one possible implementation, the number of recommendation groups is negatively correlated with the number of target products; the device also includes an acquisition module, which is used to obtain P popular products if the number of recommended products is less than a second threshold, and the popular products are determined based on historical orders of multiple users; the determination module is also used to use the p popular products as the last p products in the second area.

[0022] In one possible implementation, the target product and the first recommended product are determined by a search module, and the correlation between the target product and the search term is greater than the correlation between the first recommended product and the search term; the second recommended product is determined based on the recommendation module.

[0023] In one possible implementation, the determination module is also used to determine the first score of the first recommended product in the recommended products based on whether the first recommended product in the recommended products contains the search term; and / or determine the second score of the first recommended product in the recommended products based on whether the text matching value between the first recommended product in the recommended products and the search term exceeds a third threshold; and / or determine the third score of the first recommended product in the recommended products based on whether the relevance score output by the search module of the first recommended product is greater than a fourth threshold; determine the relevance of the first recommended product based on at least one score of the first recommended product.

[0024] In one possible implementation, the target category includes a first target category set and a second target category set, the first target category set is a multi-level product category corresponding to at least one target product, and the second target category set is a multi-level product category of the product indicated by the search term; there is an affiliation between the high category level and the low category level; the determination module is also specifically used to, for any second-recommended product, determine the category level that matches the product category of the second-recommended product from the target category, and determine the relevance of the second-recommended product based on the matched category level; wherein, the category level is negatively correlated with the relevance; when the matched category level belongs to the first target category set, the represented correlation is higher than that of the second target category set.

[0025] In a third aspect, an embodiment of the present application further provides a search response device, which includes a memory and a processor, wherein the memory is used to store computer programs or instructions; the processor is used to call the computer programs or instructions stored in the memory to execute a method as in any possible implementation of the first aspect.

[0026] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which instructions are stored. When a computer reads and executes the instructions, the computer executes the method in any possible implementation of the first aspect.

[0027] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein instructions are stored in the computer program product. When a computer reads and executes the instructions, the computer executes the method in any possible implementation of the first aspect above. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0029] Figure 1 A schematic diagram of a search result interface provided in an embodiment of the present application;

[0030] Figure 2 A flowchart corresponding to a search response method provided in an embodiment of the present application;

[0031] Figure 3 A schematic diagram of a process for determining the relevance of a first path of recommended products provided in an embodiment of the present application;

[0032] Figure 4 A schematic diagram of a process for determining the relevance of second-path recommended products provided in an embodiment of the present application;

[0033] Figure 5 Another flowchart corresponding to the search response method provided in this application;

[0034] Figure 6 A schematic diagram of an internal module of a search response device 6000 provided in an embodiment of the present application;

[0035] Figure 7 A structural diagram of a search response device 7000 provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] To make the objectives, technical solutions, and advantages of this application more clear, this application will be further described in detail below with reference to the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0037] Currently, in some shopping apps, after a user enters a search term in the search box, multiple target products related to the search term will be displayed in the search results interface. After the target product display interface, there are also some other recommended products. Figure 1 This is a schematic diagram of a search result interface provided by an embodiment of the present application. Figure 1 As shown, if a user enters "Hanging Water Fish" in the search box, multiple target products will appear on the search results interface, such as "Hanging Water Fish 1" and "Hanging Water Fish 2." After the last target product is output, some recommended products will appear on the search results interface, such as eggs and tomatoes. The recommended products after the target product can be called products output by the Guess You Like module. The Guess You Like module is just a name provided in the embodiment of this application.

[0038] The products corresponding to the traditional Guess You Like module are also determined through steps such as recall, rough sorting, fine sorting and re-ranking. The number of products determined through the recall, rough sorting, fine sorting and re-ranking steps is getting smaller and smaller. Traditional recall schemes may include vector-based recall, such as using the item to item (I2I) linkage model for recall. For example, if the product is beer, the result output by the I2I linkage model based on the co-occurrence probability of the user purchasing the product may be diapers. It can also include low-end recall of the search module. For example, the relevance model of the search module will score the products determined by the search module according to the search terms. It can be understood that the products are divided into products that are strongly related to the search terms, medium-related products and weakly related products according to the scores. Among them, products with lower scores, that is, weakly related products, can be used as products in the Guess You Like module. The recall step can also include other types of recalls, which will not be elaborated here.

[0039] Traditional re-ranking schemes output recalled, rough-ranked, and refined products according to specific rules. For example, products identified by the search module as low-end recalled are ranked ahead of other recalled products. However, while the low-end recalled products identified by the search module take into account the relevance between the product and the search term, they exclude a large number of products. While other recalled products have a certain degree of diversity, their relevance to the search term is insufficiently considered. Furthermore, the impact of product price is not considered for recalled products.

[0040] Based on this, the present application provides a search response method for fully considering the relevance and transaction nature of products when determining the order of product recommendations, broadening the diversity of products browsed by users and improving the user experience.

[0041] Figure 2 This is a flowchart of a search response method provided in an embodiment of the present application. The process can be executed by a search response device, such as Figure 2 As shown, the process includes the following steps:

[0042] Step 201 : The search response device determines the target product pointed to by the search term and the recommended product corresponding to the search term according to the search term input by the user.

[0043] Exemplarily, the search response device includes a search module, and the search module includes a correlation model. The search response device receives a search term input by a user, and the correlation model in the search module scores the products according to the search term, and divides the products into products that are strongly correlated with the search term, medium correlated products, and weakly correlated products. Optionally, strongly correlated products and medium correlated products can be determined as target products pointed to by the search term, and weakly correlated products can be determined as the first recommended products among the recommended products corresponding to the search term. In other words, the target product and the first recommended product are determined by the search module, and the correlation between the target product and the search term is greater than the correlation between the first recommended product and the search term.

[0044] Among them, the target product is located in the first area of ​​the search result interface, and the recommended product is located in the second area of ​​the search result interface. The first area and the second area can be arranged up and down in the search result interface, or in other arrangements. This application does not impose any specific restrictions on this.

[0045] The recommended products corresponding to the search term also include second-level recommended products. The second-level recommended products are determined based on the recommendation module. The recommendation module includes query to category (Q2C) recall, that is, the recalled products are determined based on the relationship between the search term and the product category. For example, the search term is Gala fruit. According to the search term, the second-level product category corresponding to Gala fruit is determined to be real fruit, the third-level product category is apple, and the leaf-level product category is specialty apple. Based on the second-level product category, the third-level product category and the leaf-level product category, multiple products under each product category are determined. For example, real fruit includes blueberries and cantaloupes, apples include rock sugar heart apples, and specialty apples include Yantai apples.

[0046] The second-path product recommendation also includes selecting N products based on products that are strongly, moderately, and weakly correlated with the search term, and determining similar products to each of the N products using the I2I substitution model. For example, N / 3 products are selected from each of the strongly, moderately, and weakly correlated categories. The I2I substitution model identifies similar products to a product. For example, a similar product to a Shandong Red Fuji apple is a Shaanxi Red Fuji apple. Based on the N selected products, multiple similar products are determined for each product as products in the second-path product recommendation.

[0047] Second-path product recommendations also include other models, such as the Query to Item (Q2I) model, which we won't elaborate on here. It's important to note that if the second-path product recommendations include items that are identical to both the target item and the first-path product recommendations, the identical items are removed from the second-path recommendations to ensure that the same item appears only once in the search results interface.

[0048] In step 202, the search response device determines the relevance of the first recommended products based on the relationship between the first recommended products and the search term, and determines the relevance of the second recommended products based on the relationship between the product category of the second recommended products and the target category.

[0049] Since the second-channel recommended products are all determined based on the I2I substitution model, Q2I or Q2C, which are based on vector recall, products based on vector recall cannot guarantee the correlation between products and search terms. In addition, the second-channel recommended products only retain products that are different from the target product and the first-channel recommended products. Therefore, the relationship between the first-channel recommended products and the search terms is stronger than the relationship between the second-channel recommended products and the search terms.

[0050] Determine the relevance of the first group of recommended products based on the relationship between the first group of recommended products and the search term, including: determining a first score of the first group of recommended products based on whether the first group of recommended products in the recommended products contains the search term; and / or, determining a second score of the first group of recommended products based on whether the text matching value between the first group of recommended products in the recommended products and the search term exceeds a third threshold; and / or, determining a third score of the first group of recommended products based on whether the relevance score output by the search module of the first group of recommended products in the recommended products is greater than a fourth threshold; determine the relevance of the first group of recommended products based on at least one score of the first group of recommended products.

[0051] For example, Figure 3A flow chart of determining the relevance of the first recommended product provided by the embodiment of the present application. Specifically, the relevance can be reflected according to the relevance score. For each recommended product in the first recommended product, the initial relevance score is set to 0, and it is determined whether the product name of the recommended product contains the search term. If it does, the relevance score of the product is increased by the first score; if not, the relevance score of the product is not increased; continue to determine whether the text match value between the name of the product and the search term exceeds the third threshold. If it exceeds, the relevance score of the product is increased by the second score. If it does not exceed, the second score of the product is not increased; the text match value can be calculated based on the bm25 best match. bm25 is used to measure the relevance between the document and the query. The third threshold is flexibly set according to the actual scenario; continue to determine whether the relevance score of the product is greater than the fourth threshold. If so, the relevance score of the product is increased by the third score. If not, the third score of the product is not increased. The relevance score of the product is determined by the relevance model in the search module. Among them, the third threshold and the fourth threshold are not equal. The first score can be less than the second score, and the second score can be less than the third score. In this way, the relevance score of each first recommended product can be obtained.

[0052] Optionally, for products in the first recommendation path whose relevance scores are below a preset threshold, they can be added to the second recommendation path. In this way, the recommended products can be divided into first-path recommended products with better relevance and second-path recommended products with worse relevance. It will be appreciated that after adding first-path recommended products with relevance scores below the preset threshold to the second recommendation path, the relevance scores of these products can be recalculated.

[0053] Furthermore, the relevance of the second-tier recommended products is determined based on the relationship between the product categories of the second-tier recommended products and the target categories in the recommended products. Specifically, the target categories at least include the product categories of the target products, the target categories include a first target category set and a second target category set, the first target category set includes at least one multi-level product category corresponding to the target product, the target product is a product that is strongly correlated and moderately correlated with the search term as determined by the correlation model, and the multi-level product category corresponding to the target product is determined based on at least one target product. In a possible embodiment, the product levels are divided into primary product categories, secondary product categories, tertiary product categories, and leaf-level product categories, and based on at least one target product, the primary product categories, secondary product categories, tertiary product categories, and leaf-level product categories corresponding to the target product are determined as the first target category set.

[0054] The second target category set is a multi-level product category for the product indicated by the search term. In one possible embodiment, the product levels are divided into primary, secondary, tertiary, and leaf-level product categories. Based on the search term, for example, "Gala Apple," the primary product category corresponding to Gala Apple is determined to be "Fruit," the secondary product category is "Solid Fruit," the tertiary product category is "Apple," and the leaf-level product category is "Specialty Apple." It is understood that there is an affiliation between the higher and lower category levels in the first and second target category sets. That is, the secondary product category belongs to the primary product category, and the tertiary product category belongs to the secondary product category.

[0055] Optionally, the second target category set may not include the primary product category, because the primary product category includes too broad a range of products, and the primary product category in the second target category set has no matching significance in the matching process.

[0056] Based on the relationship between the product category of the second-recommended product and the target category in the recommended products, the relevance of the second-recommended products is determined, including: for any second-recommended product, determining the category level that matches the product category of the second-recommended product from the target category, and determining the relevance of the second-recommended product based on the matching category level.

[0057] For example, Figure 4 A flow chart of determining the relevance of the second-path recommended products provided in an embodiment of the present application. Specifically, for each recommended product in the second-path recommended products, set the initial relevance score to 0, determine whether the target category contains the first-level product category of the recommended product, if so, determine whether the first-level product category of the recommended product hits the first target category set, if so, increase the relevance score of the recommended product by a fourth score, if the target category does not contain the first-level product category of the recommended product, the relevance score of the recommended product remains 0. Continue to determine whether the target category contains the second-level product category of the recommended product, if so, determine whether the second-level product category of the recommended product hits the first target category set, if so, increase the relevance score of the recommended product by a fifth score, if not, determine whether the second-level product category of the recommended product hits the second target category set, if so, increase the relevance score of the recommended product by a seventh score. Continue to determine whether the target category contains the leaf-level product category of the recommended product. If so, determine whether the leaf-level product category of the recommended product hits the first target category set. If so, increase the relevance score of the recommended product by the sixth score. If not, determine whether the leaf-level product category of the recommended product hits the second target category set. If so, increase the relevance score of the recommended product by the eighth score. In this way, the relevance score of each second-path recommended product can be obtained. It should be noted that Figure 4The corresponding target categories are divided into first-level product categories, second-level product categories and leaf-level product categories.

[0058] The category level is negatively correlated with the correlation, meaning that the higher the category level of the item being recommended, the lower the correlation. The first-level category is higher than the second-level category, which in turn is higher than the third-level category. In other words, the score for a first-level category is lower than the score for a second-level category; the score for a second-level category is lower than the score for a third-level category. Therefore, the fourth score is lower than the fifth score, the fifth score is lower than the sixth score, and the seventh score is lower than the eighth score. It should be noted that both the sixth and eighth scores are lower than the third score because the correlation between the second-level recommended item and the search term is weaker than the correlation between the first-level recommended item and the search term. Therefore, the sixth and eighth scores for the second-level recommended item cannot be higher than the third score.

[0059] Furthermore, the seventh score is calculated from the ninth score and the scores corresponding to the secondary product categories in the second target category set. The seventh score is equal to the ninth score multiplied by the scores corresponding to the secondary product categories in the second target category set. The eighth score is calculated from the tenth score and the scores corresponding to the leaf-level product categories in the second target category set. The eighth score is equal to the tenth score multiplied by the scores corresponding to the leaf-level product categories in the second target category set. Because the scores corresponding to each product category in the second target category set can be different, after hitting the second target category set, the score can be calculated based on the scores corresponding to each product category hit.

[0060] Optionally, when the matched product category level belongs to the first target category set, the correlation of the representation is higher than that of the second target category set.

[0061] Step 203 : The search response device determines, for any recommended product, a display position of the recommended product in the second area according to the relevance of the recommended product and the transaction value of the recommended product.

[0062] Specifically, transactionality represents the transaction conversion effect of the recommended product. Transactionality can be reflected by the transaction score. The calculation formula of the transaction score is as follows:

[0063] gmvScore=ctcvrScore*(price*weight1) weight2

[0064] Among them, gmvScore represents the transaction score, ctcvrScore represents the click-through conversion rate of the product, and the click-through conversion rate is the probability that the product is clicked and purchased. Price represents the price of the product. Weight1 and weight2 represent the first weight and the second weight. The values ​​of the first weight and the second weight are greater than 0 and less than 1. The values ​​of the first weight and the second weight are set according to actual needs.

[0065] Determining a display position of the recommended item in the second area based on the relevance of the recommended item and the transactionability of the recommended item includes: calculating a ranking position of the recommended item based on the transactionability of the recommended item and the relevance of the recommended item. If the number of target items is less than a first threshold, the importance of the relevance of the recommended item is higher than the importance of the transactionability of the recommended item; if the number of target items is not less than the first threshold, the importance of the relevance of the recommended item is lower than the importance of the transactionability of the recommended item.

[0066] That is, the number of target products in the first area of ​​the search results interface is determined. If the number of target products is less than the first threshold, it means that the number of target products determined based on the search terms is small, and the user has a narrow range of products to choose from in the first area. In this case, the recommended products displayed in the second area are more likely to recommend products with a higher relevance to the search terms. Therefore, when calculating the ranking position of the recommended products, the relevance of the recommended products is more considered. Conversely, if the number of target products is not less than the first threshold, it means that the number of target products determined based on the search terms is large, and the user has a wide range of products to choose from in the first area. In this case, the recommended products displayed in the second area are more likely to recommend products with higher transaction value rather than products with a higher relevance.

[0067] Specifically, the ranking score is calculated based on the relevance score and transaction score of the recommended products, as shown in the following formula:

[0068] finalScore=gmvWeight*gmvScore+relateWeight*relateScore

[0069] Where gmvScore represents the transaction score of the recommended product, gmvWeight represents the weight of the transaction score, relateScore represents the relevance score of the recommended product, and relateWeight represents the weight of the relevance score. If the number of target products is less than a first threshold, the weight of the relevance score is greater than the weight of the transaction score; if the number of target products is not less than the first threshold, the weight of the relevance score is less than the weight of the transaction score. Alternatively, if the number of target products is less than the first threshold, the weight of the relevance score is greater than 1, and the weight of the transaction score is less than 1; if the number of target products is not less than the first threshold, the weight of the relevance score is less than 1, and the weight of the transaction score is greater than 1.

[0070] The ranking position of the recommended product is determined according to the ranking score of the recommended product, and the display position of the recommended product in the second area is determined according to the ranking position of the recommended product.

[0071] Determine the display position of the recommended products in the second area according to the ranking position of the recommended products, including: dividing each recommended product into multiple recommendation groups according to the ranking position of each recommended product, each recommendation group includes multiple recommended products, and for each recommendation group, adjusting the ranking positions of the multiple recommended products in the recommendation group according to the product categories of the multiple recommended products in the recommendation group, wherein the product categories of adjacent recommended products in the adjusted recommendation group are different.

[0072] Specifically, the recommended products are initially ranked based on their ranking scores from high to low, and the recommended products are divided into multiple recommendation groups. For example, if there are 1,000 recommended products, these 1,000 recommended products can be divided into 5 recommendation groups, each containing 200 recommended products. For each recommendation group, if the 200 recommended products have 50 leaf-level product categories, namely Product Category 1, Product Category 2, Product Category 3, and Product Category 50, then the recommended products in the recommendation group are re-ranked based on the ranking scores from high to low, from Product Category 1 to Product Category 50. Specifically, among the 200 recommended products, the first recommended product belongs to product category 1, and the second recommended product also belongs to product category 1. Then, the product category of the third recommended product will be determined. If the product category of the third recommended product does not belong to product category 1, the third recommended product will be determined as the second ranking position; if the product category of the third recommended product still belongs to product category 1, the product category of the next recommended product will be determined, until a recommended product that does not belong to product category 1 is found and determined as the second ranking position. As for the recommended product with the second highest ranking score, since it is in the same product category as the recommended product with the highest ranking score, it can be ranked in the 51st position or in the third position to ensure that the product categories of adjacent recommended products are different. In this way, the user experience can be improved, because if the product categories of adjacent recommended products are the same, it will affect the user's shopping experience.

[0073] Optionally, the number of recommendation groups is negatively correlated with the number of target products, that is, the number of recommendation groups tends to decrease as the number of target products increases. Continuing with the above example, there are 1,000 recommended products divided into 5 recommendation groups. If the number of target products is less than the fifth threshold, the two recommendation groups with higher ranking scores will be merged into 1 group, and the merged group will be 4 groups; if the number of target products is not less than the fifth threshold, the three recommendation groups with higher ranking scores will be merged into 1 group, and the merged group will be 3 groups. This application does not limit the number of recommendation groups.

[0074] Optionally, if the number of target products is not less than the fifth threshold and less than the sixth threshold, the groups are merged into 3 groups; if the number of target products is not less than the sixth threshold, the groups are merged into 2 groups.

[0075] The number of recommendation groups is determined based on the number of target products. If the number of target products is large, fewer recommendation groups can be created. This allows for a greater number of recommended products in each group, with a greater variety of recommended products. This increases the diversity of recommended products within a given number of recommended products. For example, if there are 40 recommended products, if a recommendation group contains more than 40 product categories, the 40 recommended products will be of different categories. If a recommendation group contains fewer than 40 product categories, the 40 recommended products will likely be of the same category.

[0076] Optionally, if the number of recommended products is less than a second threshold, P popular products are obtained. Popular products are determined based on the historical orders of multiple users, and these p popular products are placed as the last p products in the second region. In other words, if the total number of recommended products is less than the second threshold, popular products from that period are determined based on the user's order history over the past period and added to the end of the second region. This allows users to recommend a sufficient number of products based on popular products, even if the number of recommended products is small, thereby improving their shopping experience and increasing product conversion rates.

[0077] Figure 5 Another flowchart corresponding to the search response method provided in this application does not limit the specific implementation scheme of this application.

[0078] Figure 6 A schematic diagram of an internal module of the search response device 6000 provided in an embodiment of the present application is shown as follows: Figure 6 As shown, the apparatus may include: a determination module 601 , a calculation module 602 , a grouping module 603 , an adjustment module 604 , and an acquisition module 605 .

[0079] Among them, the determination module 601 is used to determine the target product pointed to by the search term and the recommended product corresponding to the search term based on the search term input by the user; the target product is located in the first area of ​​the search result interface, and the recommended product is located in the second area of ​​the search result interface; based on the relationship between the first recommended product in the recommended products and the search term, the relevance of the first recommended product is determined; based on the relationship between the product category of the second recommended product in the recommended products and the target category, the relevance of the second recommended product is determined; wherein, the relationship between the first recommended product and the search term is stronger than the relationship between the second recommended product and the search term; the target category at least includes the product category of the target product; for any recommended product, the display position of the recommended product in the second area is determined based on the relevance of the recommended product and the transaction nature of the recommended product; the transaction nature represents the transaction conversion effect of the recommended product.

[0080] In one possible implementation, the device also includes a calculation module 602, which is used to calculate the ranking position of the recommended product based on the transactionability of the recommended product and the relevance of the recommended product; wherein, if the number of target products is less than a first threshold, the importance of the relevance of the recommended product is higher than the importance of the transactionability of the recommended product; if the number of target products is not less than the first threshold, the importance of the relevance of the recommended product is lower than the importance of the transactionability of the recommended product; the determination module 601 is also used to determine the display position of the recommended product in the second area according to the ranking position of the recommended product.

[0081] In one possible implementation, the device further includes a grouping module 603, which is used to divide each recommended product into multiple recommendation groups according to the ranking position of each recommended product; each recommendation group includes multiple recommended products; the device further includes an adjustment module 604, which is used to, for each recommendation group, adjust the ranking positions of the multiple recommended products in the recommendation group according to the product categories of the multiple recommended products in the recommendation group, wherein the product categories of adjacent recommended products in the adjusted recommendation group are different.

[0082] In one possible implementation, the number of recommendation groups is negatively correlated with the number of target products; the device also includes an acquisition module 605, which is used to obtain P popular products if the number of recommended products is less than a second threshold, and the popular products are determined based on historical orders of multiple users; the determination module 601 is also used to use the p popular products as the last p products in the second area.

[0083] In one possible implementation, the target product and the first recommended product are determined by a search module, and the correlation between the target product and the search term is greater than the correlation between the first recommended product and the search term; the second recommended product is determined based on the recommendation module.

[0084] In one possible implementation, the determination module 601 is also used to determine the first score of the first recommended product in the recommended products based on whether the first recommended product in the recommended products contains the search term; and / or, determine the second score of the first recommended product in the recommended products based on whether the text matching value between the first recommended product in the recommended products and the search term exceeds a third threshold; and / or, determine the third score of the first recommended product in the recommended products based on whether the relevance score output by the search module of the first recommended product is greater than a fourth threshold; determine the relevance of the first recommended product based on at least one score of the first recommended product.

[0085] In one possible implementation, the target category includes a first target category set and a second target category set, the first target category set is a multi-level product category corresponding to at least one target product, and the second target category set is a multi-level product category of the product indicated by the search term; there is an affiliation between the high category level and the low category level; the determination module 601 is also specifically used to, for any second-recommended product, determine the category level that matches the product category of the second-recommended product from the target category, and determine the relevance of the second-recommended product based on the matched category level; wherein the category level is negatively correlated with the relevance; when the matched category level belongs to the first target category set, the represented relevance is higher than that of the second target category set.

[0086] Figure 7 This is a schematic diagram of the structure of a search response device 7000 provided in an embodiment of the present application. Figure 7 As shown, it includes at least one processor 701 and a memory 702 connected to the at least one processor 701. The specific connection medium between the processor 701 and the memory 702 is not limited in the embodiment of the present application. Figure 7 For example, the processor 701 and the memory 702 are connected via a bus. The bus can be divided into an address bus, a data bus, a control bus, and the like.

[0087] In the embodiment of the present application, the memory 702 stores instructions that can be executed by at least one processor 701. The at least one processor 701 can implement the steps of the above-mentioned search response method by executing the instructions stored in the memory 702.

[0088] The processor 701 is the control center of the computer device. It can connect various parts of the computer device using various interfaces and lines, and perform resource settings by running or executing instructions stored in the memory 702 and calling data stored in the memory 702. Optionally, the processor 701 may include one or more processing units. The processor 701 may integrate an application processor and a modem processor. The application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understood that the modem processor may not be integrated into the processor 701. In some embodiments, the processor 701 and the memory 702 may be implemented on the same chip. In some embodiments, they may also be implemented on separate chips.

[0089] The processor 701 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed by a hardware processor, or can be executed by a combination of hardware and software modules in the processor.

[0090] The memory 702 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 702 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (Random Access Memory, RAM), a static random access memory (Static Random Access Memory, SRAM), a programmable read-only memory (Programmable Read Only Memory, PROM), a read-only memory (Read Only Memory, ROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 702 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 702 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0091] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0092] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0093] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0095] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A search response method, characterized in that: The method comprises: Determining, based on a search term input by a user, a target product pointed to by the search term and a recommended product corresponding to the search term; the target product is located in a first area of ​​the search result interface, and the recommended product is located in a second area of ​​the search result interface; Determining the relevance of a first path of recommended products based on a relationship between the first path of recommended products and the search term; determining the relevance of a second path of recommended products based on a relationship between a product category of the second path of recommended products and a target category; wherein the relationship between the first path of recommended products and the search term is stronger than the relationship between the second path of recommended products and the search term; and the target category includes at least the product category of the target product; For any recommended product, the display position of the recommended product in the second area is determined based on the relevance of the recommended product and the transaction value of the recommended product; the transaction value represents the transaction conversion effect of the recommended product.

2. The method according to claim 1, characterized in that Determining a display position of the recommended product in the second area according to the relevance of the recommended product and the transaction value of the recommended product includes: The ranking position of the recommended product is calculated based on the transactionability of the recommended product and the relevance of the recommended product; wherein, if the number of the target products is less than a first threshold, the importance of the relevance of the recommended product is higher than the importance of the transactionability of the recommended product; if the number of the target products is not less than the first threshold, the importance of the relevance of the recommended product is lower than the importance of the transactionability of the recommended product; Determine the display position of the recommended product in the second area according to the ranking position of the recommended product.

3. The method according to claim 2, characterized in that Determining a display position of the recommended product in the second area according to the ranking position of the recommended product includes: According to the ranking position of each recommended product, each recommended product is divided into multiple recommendation groups; each recommendation group includes multiple recommended products; For each recommendation group, the sorting positions of the multiple recommended products in the recommendation group are adjusted according to the product categories of the multiple recommended products in the recommendation group, wherein the product categories of adjacent recommended products in the adjusted recommendation group are different.

4. The method according to claim 3, characterized in that The number of recommendation groups is negatively correlated with the number of target products; The method further comprises: If the number of recommended products is less than the second threshold, obtain P popular products, where the popular products are determined based on historical orders of multiple users; The p popular products are taken as the last p products in the second area.

5. The method according to any one of claims 1 to 4, characterized in that The target product and the first recommended product are determined by a search module, and the correlation between the target product and the search term is greater than the correlation between the first recommended product and the search term; The second recommended products are determined according to the recommendation module.

6. The method according to claim 5, characterized in that Determining the relevance of a first group of recommended products in the recommended products based on a relationship between the first group of recommended products and the search term includes: determining a first score for a first group of recommended products among the recommended products based on whether the first group of recommended products contains the search term; and / or, determining a second score of a first group of recommended products in the recommended products based on whether a text match value between the first group of recommended products and the search term exceeds a third threshold; and / or, determining a third score of a first group of recommended products among the recommended products based on whether the relevance score of the first group of recommended products output by the search module is greater than a fourth threshold; Determine the relevance of the first path of recommended products based on at least one score of the first path of recommended products.

7. The method according to any one of claims 1 to 4, characterized in that The target categories include a first target category set and a second target category set, wherein the first target category set is a multi-level product category corresponding to at least one target product, and the second target category set is a multi-level product category of the product indicated by the search term; a high-level category and a low-level category have an affiliation relationship; Determining the relevance of the second-path recommended products according to the relationship between the product categories of the second-path recommended products and the target category includes: For any second-link recommended product, determine the category level that matches the product category of the second-link recommended product from the target category, and determine the relevance of the second-link recommended product based on the matched category level; wherein, the category level is negatively correlated with the relevance; when the matched category level belongs to the first target category set, the represented relevance is higher than that belonging to the second target category set.

8. A search response device, characterized in that: include: Memory, used to store computer programs or instructions; A processor, configured to call the computer program or instruction stored in the memory to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when a computer reads and executes the instructions, the computer is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that The computer program product stores instructions, and when a computer reads and executes the instructions, the computer is caused to perform the method according to any one of claims 1 to 7.

Images