User portrait construction and precise marketing delivery method for virtual shopping scenarios

Abstract

This invention discloses a method for constructing user profiles and precise marketing delivery in virtual shopping scenarios, belonging to the field of virtual shopping and intelligent marketing technology. The method includes the following steps: S1, constructing a virtual shopping scenario and obtaining the user's virtual avatar's interaction records within the virtual shopping scenario; S2, dividing the interaction records into interaction segments to form a scenario feature sequence; S3, generating a scenario-based user profile based on the scenario feature sequence and determining the current shopping stage; S4, calculating the slot attributes for each marketing display slot and filtering candidate marketing display slots; S5, executing marketing delivery based on the current shopping stage, the scenario-based user profile, and the candidate marketing display slots, outputting marketing content, and updating the scenario-based user profile based on the user's virtual avatar's subsequent interactions with the marketing content. Through steps S1-S5, this invention ensures that the output of marketing content is adapted to the current perspective and the current target product display area.

Description

Technical Field

[0001] This invention relates to the fields of virtual shopping and intelligent marketing technology, and in particular to methods for building user profiles and precise marketing delivery for virtual shopping scenarios. Background Technology

[0002] With the increasing application of virtual reality, augmented reality, and 3D interactive technologies in the shopping field, virtual shopping scenarios based on virtual store areas, product display areas, and try-on interactive areas are gradually becoming an important form of shopping. In these virtual shopping scenarios, users typically enter the virtual shopping environment through a virtual avatar and perform operations such as browsing, comparing, trying on, and adding items to their cart around the product display area.

[0003] In existing technologies, user profiling typically focuses on statistically analyzing user actions during the general shopping process and then using the results to deliver marketing content. This approach fails to adequately consider the continuous interactive changes in virtual shopping scenarios, making it difficult to effectively integrate user actions such as staying in the product display area, switching, returning, trying on items, and adding items to the cart. Consequently, it struggles to create contextualized user profiles that are truly suited to the virtual shopping environment.

[0004] Furthermore, existing technologies in marketing campaigns typically focus on selecting marketing content based on user profiles, while neglecting to consider the spatial characteristics of marketing display slots, current perspective relationships, and interaction density. Especially in virtual shopping scenarios, if marketing content is placed in an inappropriate display slot, it can easily obstruct the current target product display area or interfere with users' comparison, try-on, and add-to-cart actions, thus affecting continuous interaction in the virtual shopping environment. Simultaneously, existing technologies lack sufficient granularity in distinguishing between the browsing, comparison, try-on, and near-order stages, and the control of marketing placement at different stages is not clear enough, making it difficult to differentiate and restrict marketing content based on the current shopping stage. After marketing content is displayed, updates to product preference, price sensitivity, and marketing acceptance are usually rather coarse, failing to create a closed-loop update process that matches the virtual shopping scenario.

[0005] Therefore, there is an urgent need to provide methods for building user profiles and precise marketing targeting in virtual shopping scenarios to solve the above problems. Summary of the Invention

[0006] The technical problem to be solved by this invention is to overcome the shortcomings of the above-mentioned existing technology in terms of insufficient precision in marketing and to provide a method for building user profiles and precise marketing for virtual shopping scenarios.

[0007] To address the aforementioned technical problems, one technical solution adopted by this invention is: providing a method for building user profiles and precise marketing targeting in virtual shopping scenarios, comprising the following steps: S1. Construct a virtual shopping scene that includes multiple virtual store areas, product display areas, try-on interaction areas, and marketing display slots, and obtain the scene interaction records of the user's virtual avatar in the virtual shopping scene; S2. Divide the interaction segments according to the scene interaction records, and count the dwell time in the product display area, the number of times the same product category is switched, the number of times the same product display area is returned, the number of times the try-on is triggered and the number of times the product is added to the cart, to form a scene feature sequence; S3. Generate a scenario-based user profile based on the scenario feature sequence, including scenario attention, product preference, price sensitivity, and marketing acceptance, and determine the current shopping stage; S4. Calculate the slot attributes for each of the marketing display slots and filter candidate marketing display slots; S5. Execute marketing campaigns based on the current shopping stage, the contextualized user profile, and the candidate marketing display slots, output marketing content, and update the contextualized user profile based on the user's virtual avatar's subsequent interactions with the marketing content.

[0008] The present invention is further configured such that the scene interaction record in step S1 includes at least: user identifier, recording time, scene coordinates, view direction, identifier of the virtual store area, identifier of the product display area, target product identifier, operation type, and product category corresponding to the operation type; Step S2, which divides the interaction segments according to the scene interaction records, includes: using the time interval between adjacent scene interaction records not greater than a first preset time threshold, and the product display area identifier or the target product identifier corresponding to adjacent scene interaction records being the same, as the conditions for dividing the same interaction segment.

[0009] The present invention is further configured such that: determining the current shopping stage in step S3 includes: When the dwell time in the product display area is greater than or equal to the second preset time threshold, the number of times the same product category is switched is less than the first preset number threshold, the number of times the try-on is triggered is zero, and the number of times the product is added to the cart is zero, it is determined to be in the browsing stage; When the number of times the same product category is switched is greater than or equal to the first preset number threshold, the number of times the same product display area is returned is greater than or equal to the second preset number threshold, and two consecutive interaction segments meet the aforementioned conditions, it is determined to be a comparison stage; When the number of times the try-on is triggered is greater than or equal to the third preset number threshold, and the continuous gaze duration of the target product corresponding to the current scene interaction record after the try-on trigger operation is greater than or equal to the third preset time threshold, it is determined to be the try-on stage; When the number of times the product is added to the cart is greater than or equal to one, and the number of times the product display area corresponding to the same product category is returned within the fourth preset time threshold after the product is added to the cart is less than or equal to the fourth preset number threshold, and two consecutive interaction segments meet the aforementioned conditions, it is determined to be the near order placement stage.

[0010] The present invention is further configured such that the current shopping stage is updated according to the following migration rules: The browsing phase only proceeds to the comparison phase when two consecutive interactive segments both satisfy the determination criteria of the comparison phase. The comparison phase transitions to the fitting phase when the current interactive segment meets the determination criteria of the fitting phase. The comparison stage or the fitting stage will only move to the near order stage if two consecutive interaction segments meet the determination conditions of the near order stage. When the current shopping stage is the comparison stage or the try-on stage, and no try-on trigger operation or add-to-cart operation occurs in two consecutive interaction segments, and the dwell time of the product display area is less than the second preset time threshold, the process will migrate to the browsing stage.

[0011] The present invention is further configured such that: the calculation of slot attributes and the filtering of candidate marketing display slots in step S4 includes: For each of the marketing display slots, slot attributes are calculated. The ratio of the projected overlap area of ​​the marketing display slot on the product display area corresponding to the current scene interaction record to the projected area of ​​the product display area corresponding to the current scene interaction record is determined as the occlusion coefficient. The number of scene interaction records generated in the area where the marketing display slot is located within the fifth preset time threshold is determined as the interaction density. The duration for which the marketing display slot is continuously located within the current viewpoint and the viewpoint deflection rate is not greater than the first preset deflection rate is determined as the visible duration. Candidate marketing display slots are selected from those marketing display slots whose occlusion coefficient is less than or equal to the first preset occlusion threshold, whose interaction density is less than or equal to the first preset density threshold, whose visible duration is greater than or equal to the sixth preset time threshold, and whose marketing display slot has an angle with the current viewing direction less than or equal to the first preset angle threshold. When the current shopping stage is the browsing stage, the order is sorted in descending order of the visible duration, ascending order of the angle between the marketing display slot and the current viewing direction, and ascending order of the interaction density. When the current shopping stage is the comparison stage or the fitting stage, the items are sorted in ascending order of the occlusion coefficient, ascending order of the interaction density, and descending order of the visible duration. When the current shopping stage is the near-order stage, the products are sorted in ascending order of spatial distance between the marketing display slot and the product display area corresponding to the product added to the cart, descending order of visible duration, and ascending order of the angle between the marketing display slot and the current viewing direction. When there are multiple candidate marketing display slots with the same sorting result, they are sorted in ascending order according to the slot identifier, and the candidate marketing display slot that is ranked first is determined as the target marketing display slot.

[0012] The present invention is further configured such that: the marketing campaign execution in step S5 includes: A pre-generated product category association table is prepared. The table includes the following steps: First, the number of orders in historical orders where any first product category and any second product category appear simultaneously is counted. The ratio of this simultaneous order count to the total number of orders containing the first product category is determined as the co-purchase ratio. When the co-purchase ratio is greater than or equal to a second pre-defined ratio threshold, the second product category is recorded as a candidate associated product category of the first product category. A pre-defined product main accessory mapping table is read, and product categories that have a main accessory mapping relationship with the corresponding product of the first product category are recorded as candidate associated product categories of the first product category. From the candidate associated product categories, product categories identical to the first product category and product categories that have a substitution relationship with the first product category are removed to form an associated product category record corresponding to the first product category. The substitution relationship is determined by a pre-defined product substitution table. The associated product category is the product category that corresponds one-to-one with the product category to which the product added to cart is recorded in the preset product category association table; When the current shopping stage is the browsing stage, from the marketing content whose product category is consistent with the product category ranked first in the product preference in the contextualized user profile, the marketing content in the contextualized user profile whose marketing acceptance is greater than or equal to the first preset acceptance threshold corresponding to the marketing content display format is selected; When the current shopping stage is the comparison stage or the try-on stage, from the marketing content whose product category is consistent with the product category of the target product corresponding to the current scene interaction record and whose price range is the same as or adjacent to the price range of the target product corresponding to the current scene interaction record, the marketing content in the scenario-based user profile whose marketing acceptance is greater than or equal to the second preset acceptance threshold corresponding to the marketing content display format is selected. When the current shopping stage is the near-order stage, from the marketing content whose product category is the same as the product category to which the product added to the cart is to or whose related product category is the same as the product category to which the product added to the cart is to, the marketing content in the contextualized user profile whose marketing acceptance is greater than or equal to the third preset acceptance threshold corresponding to the marketing content display format is selected. When multiple marketing content exists after filtering, they are sorted in ascending order of price difference and descending order of marketing acceptance corresponding to the display format of the marketing content in the contextualized user profile. The marketing content ranked first is taken as the target marketing content. Adjacent price ranges are two price ranges whose price range number difference is equal to one. Output the target marketing content.

[0013] The present invention is further configured such that: the marketing campaign and the updated contextualized user profile in step S5 also include the following rules: During the first preset silent period after the user's virtual avatar enters the product display area corresponding to the current scene interaction record, the marketing content will not be output. When the current shopping stage is the comparison stage or the try-on stage, and the occlusion coefficient is greater than the second preset occlusion threshold or the interaction density is greater than the second preset density threshold, the marketing content is not output. When the current shopping stage is the fitting stage, and the user's virtual image is still in the fitting interaction area within the twelfth preset time threshold after the fitting trigger operation occurs, the marketing content will not be output. Once the marketing content has been output, if the user's virtual avatar leaves the product display area corresponding to the current scene interaction record, performs a return operation, performs a try-on trigger operation, or the marketing display slot is outside the current view range for a consecutive thirteenth preset time threshold, the output of the marketing content will be terminated. The contextualized user profile includes a set of product preferences based on product category, a set of price sensitivity based on price range, and a set of marketing acceptance based on display format. When updating the contextualized user profile, the updates are performed sequentially in the order of product preference, price sensitivity, and marketing acceptance, and boundary constraints are applied according to the following rules: when the updated product preference is greater than a first preset upper limit, the corresponding product preference is set to the first preset upper limit; when the updated price sensitivity is greater than a second preset upper limit, the corresponding price sensitivity is set to the second preset upper limit; when the updated marketing acceptance is greater than a third preset upper limit, the corresponding marketing acceptance is set to the third preset upper limit; when the updated product preference, price sensitivity, or marketing acceptance is less than the corresponding preset lower limit, it is set to the corresponding preset lower limit. The second preset occlusion threshold is less than or equal to the first preset occlusion threshold, and the second preset density threshold is less than or equal to the first preset density threshold.

[0014] The present invention is further configured such that: when updating the contextualized user profile based on the subsequent interactions, the product preference is updated according to the following rules: When a user's virtual avatar performs a viewing pause operation on the product corresponding to the marketing content, and the continuous gaze duration is greater than or equal to the seventh preset time threshold, the product preference degree of the product category corresponding to the product will be increased by the first preset preference increment. When a user's virtual avatar performs a click-to-enter or try-on-to-wear operation on the product corresponding to the marketing content, the product preference of the product category corresponding to the product is increased by a second preset preference increment, and the second preset preference increment is greater than the first preset preference increment. When a user's virtual avatar adds a product to their cart corresponding to the marketing content, the product preference for the product category corresponding to that product is increased by a third preset preference increment, and the third preset preference increment is greater than the second preset preference increment. When a user's virtual avatar performs a close operation on the marketing content, and the close operation occurs within an eighth preset time threshold after the marketing content is output, the product preference degree of the product category corresponding to the product is reduced by the first preset preference reduction.

[0015] The present invention is further configured such that: when updating the contextualized user profile based on the subsequent interactions, the price sensitivity is updated according to the following rules: When a user's virtual avatar performs a click-to-enter, try-on-the-wear, or add-to-cart operation on the product corresponding to the marketing content, the target price range corresponding to the product is obtained; When the target price range is the same as the price range of the target product corresponding to the current scene interaction record, or when the target price range is adjacent to the price range of the product for which the add-to-cart operation is performed, the price sensitivity of the target price range is increased by a first preset price increment. When the target price range is more than two preset price levels higher than the price range of the target product corresponding to the current scene interaction record, and the user's virtual avatar performs a close operation within the ninth preset time threshold, the price sensitivity corresponding to the target price range is reduced by the first preset price reduction. When a user's virtual avatar continuously adds products to their cart for multiple marketing content items, and the price range of the products for the multiple marketing content items is the same, the price sensitivity corresponding to the same price range is increased by a second preset price increment, and the second preset price increment is greater than the first preset price increment.

[0016] The present invention is further configured such that: when updating the contextualized user profile based on the subsequent interactions, the marketing acceptance is updated according to the following rules: When a user's virtual avatar performs a view-holding operation on the marketing content, and the continuous gaze duration is greater than or equal to the tenth preset time threshold, the marketing acceptance of the corresponding display format of the marketing content is increased by the first preset acceptance increment. When a user's virtual avatar performs a click-to-enter operation, a try-on-to-trigger operation, or an add-to-cart operation on the marketing content, the marketing acceptance of the corresponding display format of the marketing content is increased by a second preset acceptance increment, and the second preset acceptance increment is greater than the first preset acceptance increment; When a user's virtual avatar closes the marketing content, or when no action occurs within the eleventh preset time threshold, such as lingering view, clicking to enter, trying on clothes, or adding to cart, the marketing acceptance of the corresponding display format of the marketing content is reduced by the first preset acceptance reduction. When marketing content corresponding to the same display format triggers the close operation twice consecutively, the marketing acceptance rate corresponding to the display format is reduced by a second preset acceptance reduction, and the second preset acceptance reduction is greater than the first preset acceptance reduction.

[0017] The beneficial effects of this invention are as follows: 1. This invention generates a scene feature sequence based on scene interaction records, and then generates a scene-based user profile based on this sequence, thereby ensuring that the user profile remains consistent with the continuous interactions in the virtual shopping scene; 2. This invention differentiates and controls marketing campaigns based on the current shopping stage, allowing different marketing content filtering methods to correspond to the browsing stage, comparison stage, try-on stage, and near-order stage. 3. This invention filters candidate marketing display slots based on occlusion coefficient, interaction density, and visible duration, thereby adapting the marketing content output to the current viewpoint and the current target product display area. 4. This invention forms a closed-loop update process for contextualized user profiles by sequentially updating and boundary-constraining product preference, price sensitivity, and marketing acceptance based on subsequent interactions. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.

[0020] Please see Figure 1 The user profile building and targeted marketing methods for virtual shopping scenarios include the following steps: S1. Construct a virtual shopping scene that includes multiple virtual store areas, product display areas, try-on interaction areas, and marketing display slots, and obtain the scene interaction records of the user's virtual avatar in the virtual shopping scene; S2. Divide the interaction segments according to the scene interaction records, and count the dwell time in the product display area, the number of times the same product category is switched, the number of times the same product display area is returned, the number of times the try-on is triggered and the number of times the product is added to the cart, to form a scene feature sequence; S3. Generate a scenario-based user profile based on the scenario feature sequence, including scenario attention, product preference, price sensitivity, and marketing acceptance, and determine the current shopping stage; S4. Calculate the slot attributes for each marketing display slot and filter candidate marketing display slots; S5. Execute marketing campaigns based on the current shopping stage, contextualized user profiles, and candidate marketing display slots, output marketing content, and update the contextualized user profiles based on the user's subsequent interactions with the marketing content using their virtual avatar.

[0021] This method forms a sequence of scene features and a scene-based user profile by recording scene interactions, and combines the current shopping stage, candidate marketing display slots and marketing content to achieve adaptive delivery, thereby improving the accuracy of the profile and the matching of the delivery in the virtual shopping scene.

[0022] A user's virtual avatar includes at least: a user identifier, a virtual avatar location, a virtual avatar orientation, a viewing direction, and an action state corresponding to the virtual avatar location, the virtual avatar orientation, and the viewing direction.

[0023] User ID is used to distinguish different users; Virtual Avatar Position is used to represent the scene coordinates of the user's virtual avatar in the virtual shopping scenario; Virtual Avatar Orientation is used to represent the direction the user's virtual avatar is facing; View Direction is used to represent the current viewing direction of the user's virtual avatar; Action State is used to represent the action changes of the user's virtual avatar when performing actions such as view pausing, clicking to enter, trying on clothes, adding to cart, and returning.

[0024] The steps involved in building a virtual shopping scenario include: S101. Divide the virtual store area into multiple virtual store areas according to the preset scene layout, and arrange multiple product display areas in each virtual store area to form a collection of product display areas; S102. Arrange fitting interaction areas in adjacent positions of at least some of the product display areas, and ensure that the fitting interaction areas are accessible to the corresponding product display areas to form a set of fitting interaction areas; S103. Marketing display slots are arranged at visible path locations, turning locations, or stopping locations between the product display area and the try-on interaction area to form a set of marketing display slots; S104. Assign unified scene coordinates and unified identifier relationships to the virtual store area, product display area, try-on interaction area and marketing display slot, thereby generating a virtual shopping scene that includes the virtual store area, product display area, try-on interaction area and marketing display slot.

[0025] In step S1, the scene interaction record includes at least: user identifier, recording time, scene coordinates, view direction, virtual store area identifier, product display area identifier, target product identifier, operation type, and product category corresponding to the operation type. Methods for obtaining scene interaction records include: S110. After the user's virtual avatar enters the virtual shopping scene, continuously read the scene coordinates and viewing direction of the user's virtual avatar in the virtual shopping scene, and determine the virtual store area identifier and the product display area identifier of the user's virtual avatar based on the scene coordinates. S111. Based on the viewing direction and the product display content corresponding to the product display area, determine the target product identifier corresponding to the current scene interaction record, and determine the product category corresponding to the target product identifier based on the target product identifier; S112. Based on the user's virtual avatar's perspective dwell on the target product icon, click to enter, try on, add to cart, and return behavior, generate an operation type, and combine the user icon, recording time, scene coordinates, perspective direction, virtual store area icon, product display area icon, target product icon, operation type, and the product category corresponding to the operation type to form a scene interaction record.

[0026] Step S2 involves dividing the interaction segments based on the scene interaction records, including: using the time interval between adjacent scene interaction records not exceeding a first preset time threshold, and the product display area identifier or target product identifier corresponding to adjacent scene interaction records being the same, as the condition for dividing the same interaction segment.

[0027] The steps for generating a contextualized user profile are as follows: S301. Classify and summarize the dwell time in the product display area, the number of times the same product category is switched, the number of times the same product display area is returned, the number of times the try-on is triggered, and the number of times the product is added to the cart in the scene feature sequence to form interactive statistical results organized by product category; S302. Generate scene attention based on interaction statistics results, wherein the dwell time in the product display area, the number of times returning to the same product display area, and the duration of continuous gaze are jointly mapped to the degree of continuous attention to the product display area or virtual store area; S303. Generate product preference based on interaction statistics, wherein the interaction intensity corresponding to view dwell, click to enter, try-on trigger, and add to cart is jointly mapped to the degree of preference for product category; S304. Generate price sensitivity based on interaction statistics and target price range, wherein interaction results that are consistent with or adjacent to the price range of the target product corresponding to the current scenario interaction record are mapped to the degree of acceptance of the price range; S305. Generate marketing acceptance based on interaction statistics and marketing content display format, wherein the display format feedback of view stay, click to enter, try-on trigger, add to cart and close is mapped to the degree of acceptance of the display format; S306. Write scenario attention, product preference, price sensitivity, and marketing acceptance into the same profile record according to a unified user identifier to generate scenario-based user profiles.

[0028] Step S3, which determines the current shopping stage, includes: When the duration of the product display area is greater than or equal to the second preset time threshold, the number of product switching times in the same product category is less than the first preset number of times threshold, the number of times the try-on is triggered is zero, and the number of times the product is added to the cart is zero, the product is determined to be in the browsing stage. When the number of times the same product category is switched is greater than or equal to the first preset number threshold, the number of times the same product display area is returned is greater than or equal to the second preset number threshold, and two consecutive interaction segments meet the aforementioned conditions, it is determined to be in the comparison stage; When the number of times the try-on is triggered is greater than or equal to the third preset number threshold, and the continuous gaze duration of the target product corresponding to the current scene interaction record after the try-on trigger operation is greater than or equal to the third preset time threshold, it is determined to be the try-on stage; When the number of times a product is added to the cart is greater than or equal to one, and the number of times the product display area corresponding to the same product category is returned within the fourth preset time threshold after the addition to the cart is less than or equal to the fourth preset number threshold, and two consecutive interaction segments meet the aforementioned conditions, it is determined to be the near order placement stage.

[0029] The current shopping phase is updated according to the following migration rules: The browsing phase only moves to the comparison phase if two consecutive interactive segments meet the criteria for the comparison phase. The comparison phase transitions to the fitting phase when the current interaction segment meets the criteria for the fitting phase. The comparison or try-on stage will only move to the near-order stage if two consecutive interaction segments meet the criteria for the near-order stage. When the current shopping stage is the comparison stage or the try-on stage, and no try-on or add-to-cart actions occur in two consecutive interaction segments, and the dwell time in the product display area is less than the second preset time threshold, the process will move to the browsing stage.

[0030] Step S4, which involves calculating slot attributes and filtering candidate marketing display slots, includes: For each marketing display slot, the slot attributes are calculated separately. Among them, the ratio of the overlapping area of ​​the marketing display slot with the product display area corresponding to the current scene interaction record to the projected area of ​​the product display area corresponding to the current scene interaction record is determined as the occlusion coefficient. The number of scene interaction records generated in the area where the marketing display slot is located within the fifth preset time threshold is determined as the interaction density. The duration of the marketing display slot being continuously within the current view range and the view deflection rate not exceeding the first preset deflection rate is determined as the visible duration. Steps for calculating the overlapping area of ​​the projected product display area: S401. Determine the projection surface under the current view based on the scene coordinates and view direction in the current scene interaction record, and map the product display area corresponding to the current scene interaction record onto the projection surface to form the product display area projection area. S402. Map the marketing display slots to be calculated onto the projection surface to form the marketing display slot projection area; S403. Extract the overlapping part between the product display area projection area and the marketing display slot projection area, and determine the area of ​​the overlapping part as the overlapping area of ​​the marketing display slot projection of the product display area corresponding to the current scene interaction record.

[0031] The method for calculating the viewpoint deflection rate is as follows: extract the previous viewpoint direction and the next viewpoint direction from the interaction records of adjacent scenes, determine the change in direction between the previous viewpoint direction and the next viewpoint direction, and then divide the change in direction by the time interval between the interaction records of adjacent scenes to generate the viewpoint deflection rate.

[0032] From marketing display slots whose occlusion coefficient is less than or equal to the first preset occlusion threshold, whose interaction density is less than or equal to the first preset density threshold, whose visible duration is greater than or equal to the sixth preset time threshold, and whose angle between the marketing display slot and the current viewing direction is less than or equal to the first preset angle threshold, candidate marketing display slots are selected. Steps to obtain the angle between the marketing display slot and the current viewing direction: S404. Determine the current viewpoint position based on the scene coordinates in the current scene interaction record, and determine the center position of the marketing display slot based on the position of the marketing display slot in the virtual shopping scene. S405. Generate the direction of the marketing display slot from the current viewpoint position to the center position of the slot, and extract the viewpoint direction from the current scene interaction record as the current viewpoint direction. S406. Compare the angle between the direction the slot points and the current viewing direction, and determine the angle value corresponding to the angle between the marketing display slot and the current viewing direction.

[0033] When the current shopping stage is the browsing stage, the products are sorted in descending order of visible duration, ascending order of the angle between the marketing display slot and the current viewing direction, and ascending order of interaction density. When the current shopping stage is the comparison stage or the fitting stage, sort them in ascending order of occlusion coefficient, ascending order of interaction density, and descending order of visible duration. When the current shopping stage is the near-order stage, the products are sorted in ascending order by the spatial distance between the marketing display slot and the corresponding product display area of ​​the product added to the cart, descending order by the visible duration, and ascending order by the angle between the marketing display slot and the current viewing direction. The method for generating the spatial distance between the marketing display slot and the corresponding product display area for the product that has been added to the cart is as follows: S407. Extract the product identifier of the product that performed the add-to-cart operation from the scene interaction record corresponding to the add-to-cart operation, and determine the product display area corresponding to the product that performed the add-to-cart operation based on the product identifier; S408. Extract the center position of the marketing display slot and the center position of the product display area corresponding to the product that performs the add-to-cart operation, and unify the center position of the marketing display slot and the center position of the product display area corresponding to the product that performs the add-to-cart operation under the scene coordinates. S409. Generate a position interval based on the position difference between the center position of the marketing display slot and the center position of the product display area corresponding to the product that has been added to the cart in the scene coordinates, and determine the straight-line distance corresponding to the position interval as the spatial distance between the marketing display slot and the product display area corresponding to the product that has been added to the cart.

[0034] When there are multiple candidate marketing display slots with the same ranking result, they are sorted in ascending order according to the slot identifier, and the candidate marketing display slot ranked first is determined as the target marketing display slot.

[0035] Step S5, which involves executing marketing campaigns, includes: A pre-generated product category association table is prepared. This table includes: counting the number of orders in historical orders where any first product category and any second product category appear simultaneously; determining the co-purchase ratio as the ratio of these simultaneous orders to the total number of orders containing the first product category; recording the second product category as a candidate associated product category when the co-purchase ratio is greater than or equal to a second pre-defined threshold; reading a pre-defined product main component mapping table; and recording product categories that have a main component mapping relationship with the corresponding products of the first product category as candidate associated product categories; removing product categories identical to the first product category and those with a substitution relationship with the first product category from the candidate associated product categories to form associated product category records for the first product category, where the substitution relationship is determined by a pre-defined product substitution table. The default product category association table is a category correspondence table built based on historical orders. Each corresponding record in the category correspondence table includes a first product category, a second product category, and the co-purchase ratio of the first product category and the second product category. The first product category is the product category used as the association basis, and the second product category is the product category that appears in historical orders at the same time as the first product category.

[0036] Preset product main and accessory mapping table: A category mapping table established based on the compatibility between products. Each mapping record in the category mapping table includes the main product category and the accessory product category.

[0037] Preset Product Substitution Table: This is a category exclusion table built based on the substitution relationship between products. Each exclusion record in the category exclusion table includes the product category being substituted and the product category being substituted.

[0038] The method for generating the first and second product categories is as follows: extract the combinations of product categories that appear simultaneously from historical orders, determine the base category in the product category combination as the first product category, determine the associated categories that appear together with the base category as the second product category, and then comprehensively write the co-purchase ratio, main accessory mapping relationship and substitution relationship between the second product category and the first product category into the preset product category association table, the preset product main accessory mapping table and the preset product substitution table.

[0039] The associated product category is the product category that corresponds one-to-one with the product category of the product added to the cart in the preset product category association table; When the current shopping stage is the browsing stage, from the marketing content that matches the product category and the product preference ranking in the contextual user profile, filter the marketing content whose marketing acceptance corresponding to the marketing content display format is greater than or equal to the first preset acceptance threshold in the contextual user profile. The "product preference ranking first" method specifically refers to first sorting the product preferences of each product category in the contextualized user profile from high to low to form a product preference ranking sequence. Then, the product categories that are within a preset top range are extracted from the product preference ranking sequence and used as the product categories with the highest product preference ranking. When multiple product categories have the same product preference, the contextual attention of the multiple product categories is compared first, and the product category with higher contextual attention is ranked first. Then, when the contextual attention is the same, the number of times the multiple product categories are added to the cart is compared, and the product category with more times added to the cart is ranked first. This ensures that the product categories with the highest product preference ranking simultaneously reflect product preference, contextual attention, and number of times added to the cart.

[0040] When the current shopping stage is the comparison stage or the try-on stage, from the marketing content whose product category is the same as the target product corresponding to the current scene interaction record and whose price range is the same as or adjacent to the target product corresponding to the current scene interaction record, the marketing content in the contextualized user profile whose marketing acceptance is greater than or equal to the second preset acceptance threshold is selected. When the current shopping stage is the near-order stage, from the marketing content whose product category is the same as the product category of the product added to the cart or the related product category of the product category of the product added to the cart, the marketing content whose marketing acceptance is greater than or equal to the third preset acceptance threshold in the contextual user profile is selected. When multiple marketing content exists after filtering, they are sorted in ascending order of price difference and descending order of marketing acceptance corresponding to the display format of the marketing content in the contextualized user profile. The marketing content ranked first is taken as the target marketing content. Adjacent price ranges are two price ranges whose price range number difference is equal to one. Output targeted marketing content.

[0041] Step S5, which involves executing marketing campaigns and updating the contextualized user profile, also includes the following rules: During the first preset silent period after a user's virtual avatar enters the product display area corresponding to the current scene interaction record, no marketing content will be output. When the current shopping stage is the comparison stage or the try-on stage, and the occlusion coefficient is greater than the second preset occlusion threshold or the interaction density is greater than the second preset density threshold, no marketing content will be output. When the current shopping stage is the try-on stage, and the user's virtual avatar is still in the try-on interaction area within the twelfth preset time threshold after the try-on trigger operation occurs, no marketing content will be output. Once marketing content has been output, if the user's virtual avatar leaves the product display area corresponding to the current scene interaction record, performs a return operation, performs a try-on trigger operation, or the marketing display slot is outside the current view range for the thirteenth consecutive preset time threshold, the output of marketing content will be terminated. Contextualized user profiles include a set of product preferences based on product category, a set of price sensitivity based on price range, and a set of marketing acceptance based on display format; When updating the contextualized user profile, updates are performed sequentially in the order of product preference, price sensitivity, and marketing acceptance, with boundary constraints executed according to the following rules: when the updated product preference is greater than the first preset upper limit, the corresponding product preference is set to the first preset upper limit; when the updated price sensitivity is greater than the second preset upper limit, the corresponding price sensitivity is set to the second preset upper limit; when the updated marketing acceptance is greater than the third preset upper limit, the corresponding marketing acceptance is set to the third preset upper limit; when the updated product preference, price sensitivity, or marketing acceptance is less than the corresponding preset lower limit, it is set to the corresponding preset lower limit. The second preset occlusion threshold is less than or equal to the first preset occlusion threshold, and the second preset density threshold is less than or equal to the first preset density threshold.

[0042] When updating the contextualized user profile based on subsequent interactions, the product preference is updated according to the following rules: When a user's virtual avatar performs a viewing lingering operation on the product corresponding to the marketing content, and the continuous viewing time is greater than or equal to the seventh preset time threshold, the product preference of the product category corresponding to the product will be increased by the first preset preference increment. When a user's virtual avatar clicks on or tries on a product corresponding to the marketing content, the product preference for the corresponding product category is increased by a second preset preference increment, and the second preset preference increment is greater than the first preset preference increment. When a user's virtual avatar adds a product to their cart corresponding to the marketing content, the product preference for the corresponding product category is increased by a third preset preference increment, and the third preset preference increment is greater than the second preset preference increment. When a user's virtual avatar closes a marketing message, and the closing action occurs within the eighth preset time threshold after the marketing message is output, the product preference score of the corresponding product category is reduced by the first preset preference reduction.

[0043] When updating the contextualized user profile based on subsequent interactions, price sensitivity is updated according to the following rules: When a user's virtual avatar performs actions such as clicking to enter, trying on, or adding to cart on products corresponding to marketing content, obtain the target price range for the products; When the target price range is the same as the price range of the target product corresponding to the current scene interaction record, or when the target price range is adjacent to the price range of the product for which the add-to-cart operation is performed, the price sensitivity of the target price range will be increased by the first preset price increment. When the target price range is more than two preset price levels higher than the price range of the target product corresponding to the current scene interaction record, and the user's virtual avatar performs a close operation within the ninth preset time threshold, the price sensitivity corresponding to the target price range is reduced by the first preset price reduction. When a user's virtual avatar continuously adds products to their cart for multiple marketing content items, and the price range of the products for multiple marketing content items is the same, the price sensitivity corresponding to the same price range will be increased by a second preset price increment, and the second preset price increment will be greater than the first preset price increment.

[0044] When updating the contextualized user profile based on subsequent interactions, marketing acceptance is updated according to the following rules: When a user's virtual avatar performs a viewing action on marketing content, and the continuous viewing time is greater than or equal to the tenth preset time threshold, the marketing acceptance of the corresponding display format of the marketing content will be increased by the first preset acceptance increment. When a user's virtual avatar clicks on, tries on, or adds items to the shopping cart in response to marketing content, the marketing acceptance rate of the corresponding display format of the marketing content is increased by a second preset acceptance increment, and the second preset acceptance increment is greater than the first preset acceptance increment. When a user's virtual avatar closes the marketing content, or if no action occurs within the eleventh preset time threshold, such as viewing pause, clicking to enter, trying on clothes, or adding to cart, the marketing acceptance of the corresponding display format of the marketing content will be reduced by the first preset acceptance reduction. When marketing content corresponding to the same display format triggers the close operation twice consecutively, the marketing acceptance rate corresponding to the display format will be reduced by the second preset acceptance reduction, and the second preset acceptance reduction is greater than the first preset acceptance reduction.

[0045] It should be noted that in this invention, subsequent interactions include at least the user's virtual avatar's viewpoint lingering on the marketing content, clicking to enter, trying on / off triggering, adding to cart, and closing. "Same product category" refers to products of the same product category. Continuous gaze duration refers to the cumulative value of the corresponding recording duration when the user's virtual avatar's viewpoint direction continuously points to the same target product in continuous scene interaction records, and no viewpoint deviation exceeding the judgment criteria is formed between corresponding continuous scene interaction records. Price difference refers to the difference between the price range of the product corresponding to the marketing content to be filtered and the price range of the target product corresponding to the current scene interaction record, or the price range of the product corresponding to the added-to-cart operation. Multiple price ranges are numbered sequentially from low to high to form price range serial numbers, and adjacent price ranges are two price ranges whose price range serial number difference is equal to one. Each marketing display slot is pre-assigned a unique slot identifier, and the area where each marketing display slot is located is pre-determined when constructing the virtual shopping scene.

[0046] It should also be noted that the terms "first preset time threshold," "second preset time threshold," "third preset time threshold," "fourth preset time threshold," "fifth preset time threshold," "sixth preset time threshold," "seventh preset time threshold," "eighth preset time threshold," "ninth preset time threshold," "tenth preset time threshold," "eleventh preset time threshold," "twelfth preset time threshold," "thirteenth preset time threshold," "first preset number of times threshold," "second preset number of times threshold," "third preset number of times threshold," "fourth preset number of times threshold," "first preset preference increment," "second preset preference increment," "third preset preference increment," "first preset preference decrease," "first preset price increment," "second preset price increment," and "first preset..." appearing in the claims and this embodiment... The following parameters are set: price reduction, first preset acceptance increment, second preset acceptance increment, first preset acceptance reduction, second preset acceptance reduction, first preset deflection rate, first preset occlusion threshold, second preset occlusion threshold, first preset density threshold, second preset density threshold, first preset included angle threshold, second preset ratio threshold, first preset upper limit value, second preset upper limit value, third preset upper limit value, and corresponding preset lower limit value. These parameters can be determined based on the statistical results of historical scene interaction records in the virtual shopping scenario, the distribution of product display areas, the distribution of marketing display slots, and the calibration of marketing content delivery effects. The values ​​are based on the ability to distinguish between different current shopping stages, different marketing acceptance states, and different marketing display slot adaptation states. Specific Implementation In a virtual shopping scenario that includes 3 virtual store areas, 12 product display areas, 4 try-on interaction areas and 18 marketing display slots, first complete S101 to S104 according to the preset scenario layout to obtain unified scenario coordinates and unified identification relationships. After the user's virtual avatar enters the virtual shopping scene, the scene interaction records are continuously acquired according to S110 to S112. The user identifier is U01, the recording time interval is 2 seconds, the scene coordinates are located in the same product display area, the viewing direction is continuously pointing to the same target product identifier, and the operation types are view stay, click to enter, try on trigger, and add to cart. The product category corresponding to the operation type is the first product category. If the first preset time threshold is set to 5 seconds, then the time interval between adjacent scene interaction records is not greater than the first preset time threshold, and the product display area identifiers corresponding to adjacent scene interaction records are the same, so they are divided into the same interaction segment; within this interaction segment, the product display area stays for 18 seconds, the number of times the same product category is switched is 3, the number of times the same product display area is returned is 2, the number of times the try-on is triggered is 1, and the number of times the item is added to the cart is 1, thus forming a scene feature sequence; Then execute S301 to S306 to obtain a scenario-based user profile with high scenario attention, 0.78 product preference, 0.64 price sensitivity, and 0.71 marketing acceptance. Take the second preset time threshold as 8 seconds, the first preset number threshold as 2, the second preset number threshold as 1, the third preset number threshold as 1, the third preset time threshold as 6 seconds, the fourth preset time threshold as 60 seconds, and the fourth preset number threshold as 1. Then, the user's virtual image first meets the judgment conditions of the comparison stage. Then, after the trial operation, the continuous gaze duration on the target product corresponding to the current scenario interaction record reaches 7 seconds, so it enters the trial stage. After the add-to-cart operation, the number of times it returns to the product display area of ​​the same product category within the fourth preset time threshold is 0, so it enters the near-order stage. Then, S401 to S409 are executed for each marketing display slot. The first preset deflection speed is 30 degrees per second. The overlapping area of ​​the projection of a certain marketing display slot with the product display area corresponding to the current scene interaction record is 12 from the current viewpoint. The projection area of ​​the product display area corresponding to the current scene interaction record is 80. Therefore, the occlusion coefficient is 12 divided by 80, which is 0.15. The number of scene interaction records generated in the area where the marketing display slot is located within the fifth preset time threshold of 20 seconds is 4, so the interaction density is 4, the visible duration is 5 seconds, and the angle between the marketing display slot and the current viewpoint direction is 18 degrees. The first preset occlusion threshold is 0.20, the first preset density threshold is 6, the sixth preset time threshold is 3 seconds, and the first preset angle threshold is 25 degrees. Then, the marketing display slot enters the candidate marketing display slot. In the near-order stage, the spatial distance between the marketing display slot and the product display area corresponding to the product added to cart is 6, while the spatial distance between the other marketing display slot is 9. Therefore, the former is ranked first. Then, the marketing campaign is executed. First, a preset product category association table is generated from historical orders. The co-purchase ratio of the first product category and the second product category is 0.36, the second preset ratio threshold is 0.20, and the first product category and the second product category have a main accessory mapping relationship in the preset product main accessory mapping table. The first product category and the second product category do not have a substitution relationship in the preset product substitution table. Therefore, the second product category is written into the associated product category that corresponds one-to-one with the product category of the product added to cart. Then, from the marketing content whose product category is the same as the product category of the product added to cart or the associated product category, the marketing content whose marketing acceptance corresponding to the marketing content display format is greater than or equal to the third preset acceptance threshold of 0.60 is selected. If there are two marketing content, they are sorted in ascending order by price difference and descending order by marketing acceptance corresponding to the marketing content display format in the contextual user profile. The marketing content ranked first is taken as the target marketing content and output. After output, if the user's virtual avatar performs a view stay operation, click to enter, or add to cart operation on the product corresponding to the marketing content, the product preference will be updated sequentially according to the first preset preference increment of 0.05, the second preset preference increment of 0.10, and the third preset preference increment of 0.15; the price sensitivity will be updated according to the first preset price increment of 0.06; and the marketing acceptance will be updated according to the second preset acceptance increment of 0.12. If the user's virtual avatar closes the app within 4 seconds (eighth preset time threshold), the product preference score is reduced by 0.04 (first preset preference reduction). If no action occurs within 6 seconds (eleventh preset time threshold), including view lingering, clicking to enter, trying on clothes, or adding to cart), the marketing acceptance score is reduced by 0.05 (first preset acceptance reduction). Boundary constraints are applied according to the first preset upper limit of 1.00, the second preset upper limit of 1.00, the third preset upper limit of 1.00, and the corresponding preset lower limit of 0.

[0048] This embodiment can form a scenario-based user profile based on the scenario interaction record, and output marketing content in combination with the current shopping stage and candidate marketing display slots, thereby improving the accuracy of the profile, the matching of the delivery, and the stability of the update.

[0049] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for building user profiles and implementing precise marketing in virtual shopping scenarios, characterized by: Includes the following steps: S1. Construct a virtual shopping scene that includes multiple virtual store areas, product display areas, try-on interaction areas, and marketing display slots, and obtain the scene interaction records of the user's virtual avatar in the virtual shopping scene; S2. Divide the interaction segments according to the scene interaction records, and count the dwell time in the product display area, the number of times the same product category is switched, the number of times the same product display area is returned, the number of times the try-on is triggered and the number of times the product is added to the cart, to form a scene feature sequence; S3. Generate a scenario-based user profile based on the scenario feature sequence, including scenario attention, product preference, price sensitivity, and marketing acceptance, and determine the current shopping stage; S4. Calculate the slot attributes for each of the marketing display slots and filter candidate marketing display slots; S5. Execute marketing campaigns based on the current shopping stage, the contextualized user profile, and the candidate marketing display slots, output marketing content, and update the contextualized user profile based on the user's virtual avatar's subsequent interactions with the marketing content.

2. The method for constructing user profiles and conducting precise marketing for virtual shopping scenarios according to claim 1, characterized in that: The scene interaction record in step S1 includes at least: user identifier, recording time, scene coordinates, view direction, identifier of the virtual store area, identifier of the product display area, target product identifier, operation type, and product category corresponding to the operation type; Step S2, which divides the interaction segments according to the scene interaction records, includes: using the time interval between adjacent scene interaction records not greater than a first preset time threshold, and the product display area identifier or the target product identifier corresponding to adjacent scene interaction records being the same, as the conditions for dividing the same interaction segment.

3. The method for constructing user profiles and conducting precise marketing in virtual shopping scenarios according to claim 2, characterized in that: Determining the current shopping stage in step S3 includes: When the dwell time in the product display area is greater than or equal to the second preset time threshold, the number of times the same product category is switched is less than the first preset number threshold, the number of times the try-on is triggered is zero, and the number of times the product is added to the cart is zero, it is determined to be in the browsing stage; When the number of times the same product category is switched is greater than or equal to the first preset number threshold, the number of times the same product display area is returned is greater than or equal to the second preset number threshold, and two consecutive interaction segments meet the aforementioned conditions, it is determined to be a comparison stage; When the number of times the try-on is triggered is greater than or equal to the third preset number threshold, and the continuous gaze duration of the target product corresponding to the current scene interaction record after the try-on trigger operation is greater than or equal to the third preset time threshold, it is determined to be the try-on stage; When the number of times the product is added to the cart is greater than or equal to one, and the number of times the product display area corresponding to the same product category is returned within the fourth preset time threshold after the product is added to the cart is less than or equal to the fourth preset number threshold, and two consecutive interaction segments meet the aforementioned conditions, it is determined to be the near order placement stage.

4. The method for constructing user profiles and conducting precise marketing for virtual shopping scenarios according to claim 3, characterized in that: The current shopping phase is updated according to the following migration rules: The browsing phase only proceeds to the comparison phase when two consecutive interactive segments both satisfy the determination criteria of the comparison phase. The comparison phase transitions to the fitting phase when the current interactive segment meets the determination criteria of the fitting phase. The comparison stage or the fitting stage will only move to the near order stage if two consecutive interaction segments meet the determination conditions of the near order stage. When the current shopping stage is the comparison stage or the try-on stage, and no try-on trigger operation or add-to-cart operation occurs in two consecutive interaction segments, and the dwell time of the product display area is less than the second preset time threshold, the process will migrate to the browsing stage.

5. The method for constructing user profiles and conducting precise marketing for virtual shopping scenarios according to claim 4, characterized in that: Step S4, which involves calculating slot attributes and filtering candidate marketing display slots, includes: For each of the marketing display slots, slot attributes are calculated. The ratio of the projected overlap area of ​​the marketing display slot on the product display area corresponding to the current scene interaction record to the projected area of ​​the product display area corresponding to the current scene interaction record is determined as the occlusion coefficient. The number of scene interaction records generated in the area where the marketing display slot is located within the fifth preset time threshold is determined as the interaction density. The duration for which the marketing display slot is continuously located within the current viewpoint and the viewpoint deflection rate is not greater than the first preset deflection rate is determined as the visible duration. Candidate marketing display slots are selected from those marketing display slots whose occlusion coefficient is less than or equal to the first preset occlusion threshold, whose interaction density is less than or equal to the first preset density threshold, whose visible duration is greater than or equal to the sixth preset time threshold, and whose marketing display slot has an angle with the current viewing direction less than or equal to the first preset angle threshold. When the current shopping stage is the browsing stage, the order is sorted in descending order of the visible duration, ascending order of the angle between the marketing display slot and the current viewing direction, and ascending order of the interaction density. When the current shopping stage is the comparison stage or the fitting stage, the items are sorted in ascending order of the occlusion coefficient, ascending order of the interaction density, and descending order of the visible duration. When the current shopping stage is the near-order stage, the products are sorted in ascending order of spatial distance between the marketing display slot and the product display area corresponding to the product added to the cart, descending order of visible duration, and ascending order of the angle between the marketing display slot and the current viewing direction. When there are multiple candidate marketing display slots with the same sorting result, they are sorted in ascending order according to the slot identifier, and the candidate marketing display slot that is ranked first is determined as the target marketing display slot.

6. The method for constructing user profiles and conducting precise marketing in virtual shopping scenarios according to claim 5, characterized in that: Step S5, which involves executing marketing campaigns, includes: A pre-generated product category association table is prepared. The table includes the following steps: First, the number of orders in historical orders where any first product category and any second product category appear simultaneously is counted. The ratio of this simultaneous order count to the total number of orders containing the first product category is determined as the co-purchase ratio. When the co-purchase ratio is greater than or equal to a second pre-defined ratio threshold, the second product category is recorded as a candidate associated product category of the first product category. A pre-defined product main accessory mapping table is read, and product categories that have a main accessory mapping relationship with the corresponding product of the first product category are recorded as candidate associated product categories of the first product category. From the candidate associated product categories, product categories identical to the first product category and product categories that have a substitution relationship with the first product category are removed to form an associated product category record corresponding to the first product category. The substitution relationship is determined by a pre-defined product substitution table. The associated product category is the product category that corresponds one-to-one with the product category to which the product added to cart is recorded in the preset product category association table; When the current shopping stage is the browsing stage, from the marketing content whose product category is consistent with the product category ranked first in the product preference in the contextualized user profile, the marketing content in the contextualized user profile whose marketing acceptance is greater than or equal to the first preset acceptance threshold corresponding to the marketing content display format is selected; When the current shopping stage is the comparison stage or the try-on stage, from the marketing content whose product category is consistent with the product category of the target product corresponding to the current scene interaction record and whose price range is the same as or adjacent to the price range of the target product corresponding to the current scene interaction record, the marketing content in the scenario-based user profile whose marketing acceptance is greater than or equal to the second preset acceptance threshold corresponding to the marketing content display format is selected. When the current shopping stage is the near-order stage, from the marketing content whose product category is the same as the product category to which the product added to the cart is to or whose related product category is the same as the product category to which the product added to the cart is to, the marketing content in the contextualized user profile whose marketing acceptance is greater than or equal to the third preset acceptance threshold corresponding to the marketing content display format is selected. When multiple marketing content exists after filtering, they are sorted in ascending order of price difference and descending order of marketing acceptance corresponding to the display format of the marketing content in the contextualized user profile. The marketing content ranked first is taken as the target marketing content. Adjacent price ranges are two price ranges whose price range number difference is equal to one. Output the target marketing content.

7. The method for constructing user profiles and conducting precise marketing in virtual shopping scenarios according to claim 6, characterized in that: Step S5, which involves executing marketing campaigns and updating the contextualized user profile, also includes the following rules: During the first preset silent period after the user's virtual avatar enters the product display area corresponding to the current scene interaction record, the marketing content will not be output. When the current shopping stage is the comparison stage or the try-on stage, and the occlusion coefficient is greater than the second preset occlusion threshold or the interaction density is greater than the second preset density threshold, the marketing content is not output. When the current shopping stage is the fitting stage, and the user's virtual image is still in the fitting interaction area within the twelfth preset time threshold after the fitting trigger operation occurs, the marketing content will not be output. Once the marketing content has been output, if the user's virtual avatar leaves the product display area corresponding to the current scene interaction record, performs a return operation, performs a try-on trigger operation, or the marketing display slot is outside the current view range for a consecutive thirteenth preset time threshold, the output of the marketing content will be terminated. The contextualized user profile includes a set of product preferences based on product category, a set of price sensitivity based on price range, and a set of marketing acceptance based on display format. When updating the contextualized user profile, the updates are performed sequentially in the order of product preference, price sensitivity, and marketing acceptance, and boundary constraints are applied according to the following rules: when the updated product preference is greater than a first preset upper limit, the corresponding product preference is set to the first preset upper limit; when the updated price sensitivity is greater than a second preset upper limit, the corresponding price sensitivity is set to the second preset upper limit; when the updated marketing acceptance is greater than a third preset upper limit, the corresponding marketing acceptance is set to the third preset upper limit; when the updated product preference, price sensitivity, or marketing acceptance is less than the corresponding preset lower limit, it is set to the corresponding preset lower limit. The second preset occlusion threshold is less than or equal to the first preset occlusion threshold, and the second preset density threshold is less than or equal to the first preset density threshold.

8. The method for constructing user profiles and conducting precise marketing in virtual shopping scenarios according to claim 7, characterized in that: When updating the contextualized user profile based on subsequent interactions, the product preference is updated according to the following rules: When a user's virtual avatar performs a viewing pause operation on the product corresponding to the marketing content, and the continuous viewing time is greater than or equal to the seventh preset time threshold, the product preference of the product category corresponding to the product will be increased by the first preset preference increment. When a user's virtual avatar performs a click-to-enter or try-on-to-wear operation on the product corresponding to the marketing content, the product preference of the product category corresponding to the product is increased by a second preset preference increment, and the second preset preference increment is greater than the first preset preference increment. When a user's virtual avatar adds a product to their cart corresponding to the marketing content, the product preference for the product category corresponding to that product is increased by a third preset preference increment, and the third preset preference increment is greater than the second preset preference increment. When a user's virtual avatar performs a close operation on the marketing content, and the close operation occurs within an eighth preset time threshold after the marketing content is output, the product preference degree of the product category corresponding to the product is reduced by a first preset preference reduction.

9. The method for constructing user profiles and conducting precise marketing for virtual shopping scenarios according to claim 8, characterized in that: When updating the contextualized user profile based on subsequent interactions, price sensitivity is updated according to the following rules: When a user's virtual avatar performs a click-to-enter, try-on-the-wear, or add-to-cart operation on the product corresponding to the marketing content, the target price range corresponding to the product is obtained; When the target price range is the same as the price range of the target product corresponding to the current scene interaction record, or when the target price range is adjacent to the price range of the product for which the add-to-cart operation is performed, the price sensitivity of the target price range is increased by a first preset price increment. When the target price range is more than two preset price levels higher than the price range of the target product corresponding to the current scene interaction record, and the user's virtual avatar performs a close operation within the ninth preset time threshold, the price sensitivity corresponding to the target price range is reduced by the first preset price reduction. When a user's virtual avatar continuously adds items to their cart for multiple marketing content items, and the price range of the multiple marketing content items is the same, the price sensitivity corresponding to the same price range is increased by a second preset price increment, and the second preset price increment is greater than the first preset price increment.

10. The method for constructing user profiles and conducting precise marketing for virtual shopping scenarios according to claim 9, characterized in that: When updating the contextualized user profile based on subsequent interactions, the marketing acceptance rate is updated according to the following rules: When a user's virtual avatar performs a view-holding operation on the marketing content, and the continuous gaze duration is greater than or equal to the tenth preset time threshold, the marketing acceptance of the corresponding display format of the marketing content will be increased by the first preset acceptance increment. When a user's virtual avatar performs a click-to-enter operation, a try-on-to-trigger operation, or an add-to-cart operation on the marketing content, the marketing acceptance of the corresponding display format of the marketing content is increased by a second preset acceptance increment, and the second preset acceptance increment is greater than the first preset acceptance increment; When a user's virtual avatar closes the marketing content, or when no action occurs within the eleventh preset time threshold, such as lingering view, clicking to enter, trying on clothes, or adding to cart, the marketing acceptance of the corresponding display format of the marketing content is reduced by the first preset acceptance reduction. When marketing content corresponding to the same display format triggers the close operation twice consecutively, the marketing acceptance rate corresponding to the display format is reduced by a second preset acceptance reduction, and the second preset acceptance reduction is greater than the first preset acceptance reduction.

Images