Automatic interaction method and system based on large retail model

Abstract

The invention relates to the technical field of computers, in particular to an automatic interaction method and system based on a large retail model, and the method comprises the following steps: S1, recording user behaviors in real time; s2, acquiring a matched asset matrix in a preset asset library; s3, sequentially associating a plurality of assets in the pushing asset matrix, judging the assets preferred by the user according to the user behavior record by adopting a retail large model, and sequentially generating a preference asset list according to the preference degree; and S4, dividing and establishing a retrieval matrix according to a preset sorting rule of the selected asset characteristics, and performing retrieval in an asset matrix corresponding to the asset characteristics in the asset library through the retrieval matrix to obtain a matched asset matrix. According to the method, assets with high integrating degree can be automatically and optimally pushed to the retail large model according to the user portraits and the user behaviors, the occupation of computing power resources can be reduced, and the recommendation effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an automatic interaction method and system based on a retail macro model. Background Art

[0002] The retail industry is the sales and service part of commercial activities that directly faces consumers. It covers various sales channels from physical stores to e-commerce. With the advancement of technology and changes in consumer preferences, the retail industry is undergoing a profound transformation. More and more transaction activities are transferred to online platforms. At the same time, with the development of AI technology and big data analysis, in order to optimize product recommendations and improve transaction efficiency, AI technology will be used to provide each customer with unique products and services based on user preferences and historical behaviors, such as the retail recommendation method, system and medium based on data processing of graph database disclosed in a Chinese patent (Announcement No.: CN115797020B). In this patented technology, the method includes: collecting product interaction information and consumption characteristic information and inputting them into the retail event graph database to generate a retail relationship model organization tree, and generating a target consumer group feature portrait, then extracting consumption tendency data and comparing the similarity of the product data information in the graph database to obtain product retail intention data, processing the consumption tendency data and the product retail intention data to obtain the supply and marketing matching correlation index, and then comparing it with the preset retail relationship fitting threshold. The value is compared with the threshold to determine the effectiveness of the retail product recommendation; the consumption tendency data is obtained by using the graph database, and the retail intention data obtained by comparison with the product data information is processed for fit and the retail product recommendation is determined, so as to realize the recommendation and judgment technology of consumer groups and retail products through graph database technology and big data. When it works, it is necessary not only to analyze the user's information and behavior, but also to pre-process the stored products. The general processing method is to set up data mapping relationships, and process the correlation between products through numerous data mapping relationships. Several related products are added to the products through manual, specification library division or AI technology automatic processing, and finally the storage of related products is realized. Although it can reduce the computing power resources occupied by the AI ​​model to a certain extent, the large-scale mapping relationship table will also bring greater data storage pressure, and the subsequent maintenance of adding and reducing data operations is more complicated, and the operating cost is higher. Summary of the invention

[0003] The technical problem to be solved by the present invention is as follows: the existing product recommendation method relies on AI technology. When faced with large-scale product processing, it will occupy more computing resources, resulting in poor recommendation effect, difficult storage of products, high data storage pressure, and complex operations of adding and subtracting data during subsequent maintenance, resulting in high operating costs.

[0004] In order to solve the above technical problems, the first aspect of the present invention adopts the following technical solution: an automatic interaction method based on a retail macro model, comprising the following steps: S1: Access users, obtain historical user portraits, and record user behavior in real time; S2: Call the preset retail model to obtain the matching asset matrix in the preset asset library according to the user portrait and user behavior; S3: According to the compatibility between the user and the asset, several assets in the asset matrix are sequentially associated and pushed, and the retail model is used to determine the assets preferred by the user based on the user behavior records, and a list of preferred assets is generated in order of preference; S4: Obtain several assets in the preferred asset list, divide and establish a search matrix according to the preset sorting rules of the selected asset characteristics, search the asset matrix corresponding to the asset characteristics in the asset library through the search matrix, obtain the matching asset matrix, and go to step S3.

[0005] When the present invention is working, it can automatically optimize and push assets with high compatibility to users according to user portraits and user behaviors through the retail big model, and can cyclically obtain assets preferred by users during the push process. At the same time, a retrieval matrix is ​​established according to the asset characteristics of the preferred assets to search in the asset library, so that the retail big model only needs to analyze the users, reducing the process of analyzing the products. The scale of data that needs to be called when pushing products can be greatly reduced only through feature matching. While greatly reducing the computing power resources occupied, the data security factor of the asset library can also be improved.

[0006] Preferably, in step S1, when accessing a user, obtaining a historical user portrait, and recording the user behavior in real time, the following steps are adopted: accessing a user, obtaining a historical user portrait, and recording the user behavior in real time when the user browses, and the recorded user behavior includes at least one of user search information, user browsing information, user page dwell time, and user click records.

[0007] Preferably, in step S2, when calling the preset retail macro model to obtain a matching asset matrix in the preset asset library according to the user portrait and user behavior, the following steps are adopted: A1: Obtain the user's user profile and user behavior, call the preset retail model to split several asset features related to the user according to the user profile and user behavior, dynamically weight several asset features according to the current scenario, use the aggregate statistics method to obtain the statistical results of several asset features and arrange and store them to obtain the asset feature set preferred by the user; A2: The asset feature set is used to locate matching assets in the asset matrix corresponding to the selected asset feature type in the preset asset library, and the asset matrix of suitable data size is extracted from the asset matrix according to the matching assets using preset interception specifications.

[0008] Preferably, the asset library is established by the following steps: B1: Determine the associated hierarchical distribution of several asset features, establish several source nodes, each source node is filled with asset features of the corresponding level as several first child nodes, and the several first child nodes are arranged in sequence into row vectors according to a preset sorting rule; B2: Each first sub-node is filled with a number of second sub-nodes whose asset features are of a corresponding level, and the second sub-nodes are sequentially arranged into a column vector according to a preset sorting rule; B3: Call the retail model to split each asset into several asset features, add the index of the asset to each asset feature, and configure the several asset features as matching second child nodes; B4: Traverse several row vectors of several source nodes to form the elements of the first row of several asset matrices respectively, and fill the corresponding column vectors to form elements of several columns respectively, so as to generate several asset matrices and store them as asset libraries.

[0009] When the present invention is working, it can realize the storage of assets. By splitting asset features and storing assets through category classification, the association operations between several assets are reduced, and the data mapping relationship table can be saved when the asset scale is large. The pressure of data storage can be reduced to a certain extent. At the same time, it is also convenient to update the asset library when adding or reducing assets, and the workflow is optimized. By establishing several asset matrices, the hidden association information between different assets can be excavated, and the conversion rate of asset push can be improved while ensuring the compatibility of asset push, and no additional computing power resources are required, with high economic benefits.

[0010] Preferably, in step S3, according to the compatibility between the user and the asset, several assets in the asset matrix are sequentially associated and pushed, the retail big model is used to determine the user's preferred assets based on the user's behavior records, and a list of preferred assets is generated in sequence according to the degree of preference. The following steps are adopted: the retail big model is used to locate the matching assets in the matching asset matrix according to the user's preference feature set, and at the same time, at least one predicted preferred asset is predicted according to the user's preference feature set according to the retail big model, the associated push direction is determined in the asset matrix through a preset push algorithm, and several assets in the asset matrix are sequentially pushed, the retail big model is used to determine the user's preferred assets based on the user's behavior records, and a list of preferred assets is generated in sequence according to the degree of preference.

[0011] Preferably, in step S3, when determining the associated push direction in the asset matrix by a preset push algorithm and sequentially pushing a number of assets in the asset matrix, the following steps are adopted: C1: Obtain the asset features corresponding to the matched assets and the position of the predicted preferred assets in the current asset matrix, calculate the initial vector from the matched assets to the predicted preferred assets, and calculate the unit vector, and calculate the position of the next preferred asset based on the unit vector and the preset step size; C2: According to the position of the next preferred asset, the matching asset matrix is ​​intercepted by the preset interception specification. When the matching assets in the current asset matrix are pushed, the process goes to step C3; C3: When there is a new preferred asset in the current asset matrix, go to step C4, otherwise, go to step C5; C4: Calculate the unit vector from the current preferred asset to the preset preferred asset, and the unit vector from the current preferred asset to the new preferred asset, obtain the corrected unit vector by vector addition, calculate the position of the next preferred asset according to the unit vector and the preset step length, and judge whether the position of the next preferred asset is the predicted position or whether the number of cycles reaches the preset threshold. If yes, end and record all the preferred assets to generate a preferred asset list, and go to step S4. Otherwise, go to step C2; C5: Dynamically adjust the step size according to the number of loops, calculate the position of the next preferred asset according to the unit vector and the accumulated step size, and determine whether the position of the next preferred asset is the predicted position or whether the number of loops reaches the preset threshold. If yes, end and record all preferred assets to generate a preferred asset list and go to step S4. Otherwise, go to step C2.

[0012] When the present invention is working, the push algorithm is used to determine the associated push direction in the asset matrix, and the sequential push of several assets in the asset matrix can be realized. The push path can be corrected according to the real-time changes of user preferences, and the effect of asset push can be further improved, and the conversion rate can be improved. At the same time, the final push target can be corrected, and the user portrait can be further improved. Through multiple cycles of push, the user portrait can be accurately portrayed, thereby assisting the retail big model to push assets with higher fit to users.

[0013] Preferably, step C2 further includes the following steps: when a matching asset matrix is ​​intercepted according to the position of the next preferred asset through a preset interception specification, the interception path and the asset matrix intercepted each time are synchronously recorded to obtain a number of optimized push matrices, and the several optimized push matrices are processed through the retail big model to obtain a number of asset push paths for reducing the consumption of computing resources.

[0014] Preferably, in step S4, a number of assets in the preferred asset list are obtained, and a search matrix is ​​established according to the preset sorting rules of the selected asset features. The asset matrix corresponding to the asset features in the asset library is searched through the search matrix to obtain a matching asset matrix. When entering step S3, the following steps are adopted: D1: Obtain several assets in the preferred asset list, cluster their asset features and determine at least one common asset feature and at least two similar asset features; D2: Determine the matrix position of the selected asset according to the preset sorting rules based on the similar asset characteristics, obtain several groups of matrix positions, calculate the distance of each matrix position in each group of matrix positions and verify them, and eliminate the related asset characteristics whose distance exceeds the threshold; D3: Obtain the preset interception specifications to establish a matrix, and fill in the relevant asset features that have passed the distance verification in several group matrix positions to obtain a search matrix; D4: Search the asset matrix corresponding to the asset features in the asset library by template matching through the search matrix, obtain the matching asset matrix, and go to step S3.

[0015] When the present invention is working, a search matrix obtained by filling in a plurality of groups of matrix positions is used to search in the asset matrix. This can effectively mine the user's hidden preferences while ensuring that the intercepted asset matrix is ​​highly consistent with the current user, without occupying additional computing resources, with high calling efficiency and good push effect.

[0016] Preferably, in step D3, a preset interception specification is obtained to establish a matrix, and relevant asset features that have passed the distance verification in several groups of matrix positions are filled in. When the retrieval matrix is ​​obtained, the following steps are adopted: a preset interception specification is obtained to establish a matrix, all elements are traversed according to several groups of matrix positions, appropriate asset features are filled in in turn, and it is ensured that the corresponding asset features meet the distance constraints of each group of matrix positions, and the retrieval matrix is ​​obtained repeatedly in turn.

[0017] To solve the above technical problems, the second aspect of the present invention adopts the following technical solution: an automatic interaction system based on a retail big model, applying an automatic interaction method based on a retail big model as described above, the automatic interaction system includes a processor and a memory, the memory stores at least one instruction, at least one program, code set or instruction set, the instruction, program, code set or instruction set is loaded by the processor and executes an automatic interaction method based on a retail big model as described above.

[0018] The beneficial technical effects of the present invention include: 1. The present invention can automatically optimize and push highly compatible assets to users according to user portraits and user behaviors through the retail big model, and can cyclically obtain assets preferred by users during the push process. At the same time, a retrieval matrix is ​​established according to the asset characteristics of the preferred assets to search in the asset library, so that the retail big model only needs to analyze the users, reducing the process of analyzing goods. The scale of data that needs to be called when pushing goods can be greatly reduced only through feature matching. While greatly reducing the computing power resources occupied, it can also improve the data security factor of the asset library.

[0019] 2. The present invention can realize the storage of assets. By splitting asset features and storing assets through category classification, the association operations between several assets are reduced, and the data mapping relationship table can be saved when the asset scale is large. The pressure of data storage can be reduced to a certain extent. At the same time, it is also convenient to update the asset library when adding or reducing assets, and the workflow is optimized. By establishing several asset matrices, the hidden association information between different assets can be excavated, which can improve the conversion rate of asset push while ensuring the compatibility of asset push, and there is no need to occupy additional computing power resources, with high economic benefits.

[0020] 3. The present invention determines the associated push directions in the asset matrix through a push algorithm, realizes the sequential push of several assets in the asset matrix, can correct the push path according to real-time changes in user preferences, can further improve the effect of asset push, and improve the conversion rate. At the same time, it can also realize the correction of the final push target, can further improve the user portrait, and complete the accurate portrayal of the user portrait through multiple cycles of push, thereby assisting the retail big model to push assets with higher fit to users.

[0021] 4. The present invention uses a search matrix obtained by filling in several groups of matrix positions to search in the asset matrix. It can effectively explore the user's hidden preferences while ensuring that the intercepted asset matrix is ​​highly consistent with the current user, without occupying additional computing power resources, with high calling efficiency and good push effect.

[0022] Other features and advantages of the present invention will be disclosed in detail in the following specific embodiments and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described below in conjunction with the accompanying drawings: Figure 1 It is a workflow diagram of an automatic interaction method based on a retail big model; Figure 2 Workflow diagram for establishing asset library; Figure 3 is a workflow diagram of step S3 in an automatic interaction method based on a retail big model; Figure 4This is a workflow diagram of step S4 in an automatic interaction method based on a retail macro model. DETAILED DESCRIPTION

[0024] The technical solutions of the embodiments of the present invention are explained and described below in conjunction with the drawings of the embodiments of the present invention, but the following embodiments are only preferred embodiments of the present invention, not all. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without creative work are all within the protection scope of the present invention.

[0025] In the following description, terms such as "inside", "outside", "up", "down", "left", "right", etc. that indicate directions or positional relationships are only used to facilitate the description of the embodiments and simplify the description, and do not indicate or imply that the referred device or element must have a specific direction, be constructed and operate in a specific direction. Therefore, it should not be understood as a limitation of the present invention. Embodiment 1:

[0026] See also Figure 1 This embodiment discloses an automatic interaction method based on a retail big model, comprising the following steps: S1: Access users, obtain historical user portraits, and record user behavior in real time; S2: Call the preset retail model to obtain the matching asset matrix in the preset asset library according to the user portrait and user behavior; S3: According to the compatibility between the user and the asset, several assets in the asset matrix are sequentially associated and pushed, and the retail model is used to determine the assets preferred by the user based on the user behavior records, and a list of preferred assets is generated in order of preference; S4: Obtain several assets in the preferred asset list, divide and establish a search matrix according to the preset sorting rules of the selected asset characteristics, search the asset matrix corresponding to the asset characteristics in the asset library through the search matrix, obtain the matching asset matrix, and go to step S3.

[0027] When this instance is working, the retail big model can automatically optimize and push highly suitable assets to users based on their portraits and behaviors, and can cyclically obtain assets preferred by users during the push process. At the same time, a retrieval matrix is ​​established based on the asset features of the preferred assets to search in the asset library. This allows the retail big model to only analyze users, reducing the process of analyzing products. The scale of data that needs to be called when pushing products can be greatly reduced through feature matching alone. While greatly reducing the computing power resources occupied, it can also improve the data security factor of the asset library.

[0028] In this embodiment, in step S1, when accessing a user, obtaining a historical user portrait, and recording the user behavior in real time, the following steps are adopted: accessing a user, obtaining a historical user portrait, and recording the user behavior in real time when the user browses. The recorded user behavior includes at least one of user search information, user browsing information, user page dwell time, and user click records. For example, the user browses the material interface, process interface, effect interface, etc. of an asset. By recording this part of the user behavior, the asset features preferred by the user and their corresponding types can be captured.

[0029] In the specific implementation, in step S2, when calling the preset retail model to obtain the matching asset matrix in the preset asset library according to the user portrait and user behavior, the following steps are adopted: A1: Obtain the user's user profile and user behavior, call the preset retail model to split several asset features related to the user according to the user profile and user behavior, dynamically weight several asset features according to the current scenario, use the aggregate statistics method to obtain the statistical results of several asset features and arrange and store them to obtain the asset feature set preferred by the user; A2: The asset feature set is used to locate matching assets in the asset matrix corresponding to the selected asset feature type in the preset asset library, and the asset matrix of suitable data size is extracted from the asset matrix according to the matching assets using preset interception specifications.

[0030] Preferably, when dynamically assigning weights, the weights of each asset feature can be dynamically planned based on user behavior. At this time, the hidden preferences of each user can be mined through the retail big model based on the common preferences of similar customers, which can further improve the asset push effect.

[0031] See also Figure 2 As a further improvement of this embodiment, the asset library is established by the following steps: B1: Determine the associated hierarchical distribution of several asset features, establish several source nodes, each source node is filled with asset features of the corresponding level as several first child nodes, and the several first child nodes are arranged in sequence into row vectors according to a preset sorting rule; B2: Each first sub-node is filled with a number of second sub-nodes whose asset features are of a corresponding level, and the second sub-nodes are sequentially arranged into a column vector according to a preset sorting rule; B3: Call the retail model to split each asset into several asset features, add the index of the asset to each asset feature, and configure the several asset features as matching second child nodes; B4: Traverse several row vectors of several source nodes to form the elements of the first row of several asset matrices respectively, and fill the corresponding column vectors to form elements of several columns respectively, so as to generate several asset matrices and store them as asset libraries.

[0032] When this instance is working, it can realize the storage of assets. By splitting asset features and storing assets through category classification, it reduces the association operations between several assets, can save data mapping relationship tables when the asset scale is large, can reduce the pressure of data storage to a certain extent, and also facilitates updating the asset library when adding or reducing assets, optimizes the workflow, and can also mine hidden association information between different assets by establishing several asset matrices, which can improve the conversion rate of asset push while ensuring the fit of asset push, and does not require additional computing power resources, with high economic benefits.

[0033] In specific implementation, the sorting rules can be appropriately divided according to the type of each asset feature. For non-numerical features, manual or big data analysis is required to determine the correlation for sequential division. For numerical features, they can be sequentially divided by numerical sorting. When dividing, they need to be sorted at intervals according to the preset distribution gradient, and the positions with no data are set to zero. This can ensure the correlation between similar asset features in the asset matrix. Compared with manual or AI model automatic addition of mapping relationships, it not only improves the correlation efficiency, but also facilitates the excavation of hidden correlation information between assets, which can further improve the fit of asset push.

[0034] See also Figure 4 Preferably, in step S4, a plurality of assets in the preferred asset list are obtained, and a search matrix is ​​established according to the preset sorting rules of the selected asset features. The asset matrix corresponding to the asset features in the asset library is searched through the search matrix to obtain a matching asset matrix. When entering step S3, the following steps are adopted: D1: Obtain several assets in the preferred asset list, cluster their asset features and determine at least one common asset feature and at least two similar asset features; D2: Determine the matrix position of the selected asset according to the preset sorting rules based on the similar asset characteristics, obtain several groups of matrix positions, calculate the distance of each matrix position in each group of matrix positions and verify them, and eliminate the related asset characteristics whose distance exceeds the threshold; D3: Obtain the preset interception specifications to establish a matrix, and fill in the relevant asset features that have passed the distance verification in several group matrix positions to obtain a search matrix; D4: Search the asset matrix corresponding to the asset features in the asset library by template matching through the search matrix, obtain the matching asset matrix, and go to step S3.

[0035] When this example is working, a search matrix obtained by filling in several groups of matrix positions is used to search in the asset matrix. This can effectively mine the user's hidden preferences while ensuring that the intercepted asset matrix is ​​highly consistent with the current user, without occupying additional computing power resources, with high calling efficiency and good push effect.

[0036] As a further improvement of the present embodiment, in the step D2, when calculating the distance between each matrix position in each group of matrix positions, a Chebyshev distance calculation operation may be used to calculate the maximum distance between the corresponding matrices in the asset matrix under different sorting rules. In actual work, the row coordinates of two matrix positions generally have a small deviation, and their vertical coordinates have a large deviation. The use of the Chebyshev distance calculation operation can reduce the calculation pressure. Of course, any other suitable calculation method may also be used to achieve the distance calculation requirements of the present embodiment.

[0037] Preferably, in step D3, a preset interception specification is obtained to establish a matrix, and relevant asset features that have passed the distance verification in several groups of matrix positions are filled. When the retrieval matrix is ​​obtained, the following steps are adopted: the preset interception specification is obtained to establish a matrix, all elements are traversed according to several groups of matrix positions, appropriate asset features are filled in turn and it is ensured that the corresponding asset features meet the distance constraints of each group of matrix positions, and the retrieval matrix is ​​obtained repeatedly in turn. Preferably, several groups of matrix positions can also be screened to control the number of fillings, which can improve the efficiency of template matching and reduce the occupation of computing resources while ensuring the matching efficiency. Embodiment 2:

[0038] See also Figure 3 This embodiment provides an automatic interaction method based on a retail big model. The similarities with other embodiments are not repeated here, and the differences are described in detail below.

[0039] In this embodiment, in the step S3, according to the compatibility between the user and the asset, several assets in the asset matrix are sequentially associated and pushed, the retail big model is used to determine the user's preferred assets according to the user behavior record, and the preferred asset list is generated in sequence according to the degree of preference. The following steps are adopted: the retail big model is used to locate the matching assets in the matching asset matrix according to the user's preference feature set, and at the same time, at least one predicted preferred asset is predicted according to the user's preference feature set according to the retail big model, the associated push direction is determined in the asset matrix through a preset push algorithm, and several assets in the asset matrix are sequentially pushed, the retail big model is used to determine the user's preferred assets according to the user behavior record, and the preferred asset list is generated in sequence according to the degree of preference.

[0040] In the specific implementation, in step S3, the associated push direction is determined in the asset matrix by a preset push algorithm, and several assets in the asset matrix are pushed in sequence, and the following steps are adopted: C1: Obtain the asset features corresponding to the matched assets and the position of the predicted preferred assets in the current asset matrix, calculate the initial vector from the matched assets to the predicted preferred assets, and calculate the unit vector, and calculate the position of the next preferred asset based on the unit vector and the preset step size; C2: According to the position of the next preferred asset, the matching asset matrix is ​​intercepted by the preset interception specification. When the matching assets in the current asset matrix are pushed, the process goes to step C3; C3: When there is a new preferred asset in the current asset matrix, go to step C4, otherwise, go to step C5; C4: Calculate the unit vector from the current preferred asset to the preset preferred asset, and the unit vector from the current preferred asset to the new preferred asset, obtain the corrected unit vector by vector addition, calculate the position of the next preferred asset according to the unit vector and the preset step length, and judge whether the position of the next preferred asset is the predicted position or whether the number of cycles reaches the preset threshold. If yes, end and record all the preferred assets to generate a preferred asset list, and go to step S4. Otherwise, go to step C2; C5: Dynamically adjust the step size according to the number of loops, calculate the position of the next preferred asset according to the unit vector and the accumulated step size, and determine whether the position of the next preferred asset is the predicted position or whether the number of loops reaches the preset threshold. If yes, end and record all preferred assets to generate a preferred asset list and go to step S4. Otherwise, go to step C2.

[0041] When this example is working, the push algorithm is used to determine the associated push direction in the asset matrix, and to push several assets in the asset matrix in sequence. The push path can be modified according to real-time changes in user preferences, which can further improve the effect of asset push and increase conversion rate. At the same time, the final push target can be corrected, and the user portrait can be further improved. Through multiple cycles of push, the user portrait can be accurately portrayed, thereby assisting the retail model in pushing more suitable assets to users.

[0042] As a further improvement of this embodiment, in step C1, the asset features corresponding to the matched assets and the position of the predicted preferred assets in the current asset matrix are obtained, the initial vector from the matched assets to the predicted preferred assets is calculated, and the unit vector is calculated. When the position of the next preferred asset is calculated according to the unit vector and the preset step length, the following steps are adopted. The asset feature position corresponding to the matched assets is: , predicting the position of the preferred asset is , when calculating the initial vector from the matched assets to the predicted preferred assets, the following formula is used: ; in: is the initial vector of matching assets to predicted preferred assets; When calculating the unit vector from the matched asset to the predicted preferred asset, the following formula is used: ; in: is the unit vector from the matched asset to the predicted preferred asset; When calculating the next preferred asset position, the following formula is used: ; ; in: is the actual step from the current asset to the next preferred asset; It is the rounding operation after step length calculation; s is the preset step size, and the step size setting is adjustable; for the location of the next preferred asset; In step C4, the following formula is used to calculate the unit vector from the current preferred asset to the preset preferred asset and the unit vector from the current preferred asset to the new preferred asset: ; ; in: is the unit vector from the current preferred asset to the preset preferred asset; is the unit vector from the current preferred asset to the new preferred asset; When the corrected unit vector is obtained by vector addition, the following formula is used: ; ; in: The correction vector is obtained by adding the unit vector from the current preferred asset to the preset preferred asset and the unit vector from the current preferred asset to the new preferred asset; is the unit vector of the correction vector; Calculate the vector length for pair by taking the Euclidean norm; When calculating the next preferred asset position, the following formula is used: ; ; In step C5, the step size is dynamically adjusted according to the number of cycles, and the following formula is used to calculate the position of the next preferred asset according to the unit vector and the accumulated step size: ; ; in: is the step length after accumulation.

[0043] Preferably, in step C2, the following steps are also included: when a matching asset matrix is ​​intercepted according to the position of the next preferred asset through a preset interception specification, the interception path and the asset matrix intercepted each time are synchronously recorded to obtain a number of optimized push matrices; the several optimized push matrices are processed through the retail big model to obtain a number of asset push paths for reducing the consumption of computing resources; in specific implementation, the predicted preferred assets can be set as assets with high economic benefits, at which time the user's preferences can be non-invasively influenced, so that in multiple asset browsings, the user believes that the various asset characteristics of the asset with high economic benefits are in the best combination state, thereby increasing the preference for assets with high economic benefits, and then improving their conversion rate and economic benefits. Embodiment three:

[0044] The present embodiment provides an automatic interaction system based on a retail big model, applying an automatic interaction method based on a retail big model as in the above embodiment, the automatic interaction system includes a processor and a memory, the memory stores at least one instruction, at least one program, code set or instruction set, the instruction, program, code set or instruction set is loaded by the processor and executed as in the above embodiment, an automatic interaction method based on a retail big model.

[0045] The beneficial technical effects of this embodiment include: the present invention can automatically optimize and push highly suitable assets to users according to user portraits and user behaviors through the retail big model, and can cyclically obtain assets preferred by users during the push process, and at the same time establish a retrieval matrix based on the asset characteristics of the preferred assets to search in the asset library, so that the retail big model only needs to analyze the users, reducing the process of analyzing goods, and can greatly reduce the scale of data that needs to be called when pushing goods only through feature matching, while greatly reducing the computing power resources occupied, and can also improve the data security factor of the asset library.

[0046] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Any modification that does not deviate from the functional and structural principles of the present invention will be included in the scope of the claims.

Claims

1. An automatic interaction method based on a retail big model, characterized in that: The following steps are involved: S1: Access users, obtain historical user portraits, and record user behavior in real time; S2: Call the preset retail model to obtain the matching asset matrix in the preset asset library according to the user portrait and user behavior; S3: According to the compatibility between the user and the asset, several assets in the asset matrix are sequentially associated and pushed, and the retail model is used to determine the assets preferred by the user based on the user behavior records, and a list of preferred assets is generated in order of preference; S4: Obtain several assets in the preferred asset list, divide and establish a search matrix according to the preset sorting rules of the selected asset characteristics, search the asset matrix corresponding to the asset characteristics in the asset library through the search matrix, obtain the matching asset matrix, and go to step S3.

2. The automatic interaction method based on the retail big model according to claim 1, characterized in that: In step S1, when accessing a user, obtaining a historical user portrait, and recording the user behavior in real time, the following steps are adopted: accessing a user, obtaining a historical user portrait, and recording the user behavior in real time when the user browses. The recorded user behavior includes at least one of user search information, user browsing information, user page dwell time, and user click records.

3. The automatic interaction method based on the retail big model according to claim 1 is characterized by: In step S2, when calling the preset retail model to obtain a matching asset matrix in the preset asset library according to the user portrait and user behavior, the following steps are adopted: A1: Obtain the user's user profile and user behavior, call the preset retail model to split several asset features related to the user according to the user profile and user behavior, dynamically weight several asset features according to the current scenario, use the aggregate statistics method to obtain the statistical results of several asset features and arrange and store them to obtain the asset feature set preferred by the user; A2: The asset feature set is used to locate matching assets in the asset matrix corresponding to the selected asset feature type in the preset asset library, and the asset matrix of suitable data size is extracted from the asset matrix according to the matching assets using preset interception specifications.

4. The automatic interaction method based on the retail big model according to claim 1 is characterized by: The establishment of the asset library adopts the following steps: B1: Determine the associated hierarchical distribution of several asset features, establish several source nodes, each source node is filled with asset features of the corresponding level as several first child nodes, and the several first child nodes are arranged in sequence into row vectors according to a preset sorting rule; B2: Each first sub-node is filled with a number of second sub-nodes whose asset features are of a corresponding level, and the second sub-nodes are sequentially arranged into a column vector according to a preset sorting rule; B3: Call the retail model to split each asset into several asset features, add the index of the asset to each asset feature, and configure the several asset features as matching second child nodes; B4: Traverse several row vectors of several source nodes to form the elements of the first row of several asset matrices respectively, and fill the corresponding column vectors to form elements of several columns respectively, so as to generate several asset matrices and store them as asset libraries.

5. The automatic interaction method based on the retail big model according to claim 4 is characterized by: In the step S3, according to the compatibility between the user and the asset, several assets in the asset matrix are sequentially associated and pushed, the retail big model is used to determine the user's preferred assets based on the user's behavior records, and the preferred asset list is generated in sequence according to the degree of preference. The following steps are adopted: the retail big model is used to locate the matching assets in the matching asset matrix according to the user's preference feature set, and at the same time, at least one predicted preferred asset is predicted according to the user's preference feature set according to the retail big model, the associated push direction is determined in the asset matrix through a preset push algorithm, and several assets in the asset matrix are sequentially pushed, the retail big model is used to determine the user's preferred assets based on the user's behavior records, and the preferred asset list is generated in sequence according to the degree of preference.

6. The automatic interaction method based on the retail big model according to claim 5 is characterized by: In step S3, when determining the associated push direction in the asset matrix through a preset push algorithm and sequentially pushing a number of assets in the asset matrix, the following steps are adopted: C1: Obtain the asset features corresponding to the matched assets and the position of the predicted preferred assets in the current asset matrix, calculate the initial vector from the matched assets to the predicted preferred assets, and calculate the unit vector, and calculate the position of the next preferred asset based on the unit vector and the preset step size; C2: According to the position of the next preferred asset, the matching asset matrix is ​​intercepted by the preset interception specification. When the matching assets in the current asset matrix are pushed, the process goes to step C3; C3: When there is a new preferred asset in the current asset matrix, go to step C4, otherwise, go to step C5; C4: Calculate the unit vector from the current preferred asset to the preset preferred asset, and the unit vector from the current preferred asset to the new preferred asset, obtain the corrected unit vector by vector addition, calculate the position of the next preferred asset according to the unit vector and the preset step length, and judge whether the position of the next preferred asset is the predicted position or whether the number of cycles reaches the preset threshold. If yes, end and record all the preferred assets to generate a preferred asset list, and go to step S4. Otherwise, go to step C2; C5: Dynamically adjust the step size according to the number of loops, calculate the position of the next preferred asset according to the unit vector and the accumulated step size, and determine whether the position of the next preferred asset is the predicted position or whether the number of loops reaches the preset threshold. If yes, end and record all preferred assets to generate a preferred asset list and go to step S4. Otherwise, go to step C2.

7. The automatic interaction method based on the retail big model according to claim 6 is characterized by: In the step C2, the following steps are also included: when a matching asset matrix is ​​intercepted according to the position of the next preferred asset through a preset interception specification, the interception path and the asset matrix intercepted each time are synchronously recorded to obtain a number of optimized push matrices, and a number of optimized push matrices are processed through the retail big model to obtain a number of asset push paths for reducing the consumption of computing resources.

8. The automatic interaction method based on the retail big model according to claim 1 is characterized by: In step S4, several assets in the preferred asset list are obtained, and a search matrix is ​​established according to the preset sorting rules of the selected asset features. The asset matrix corresponding to the asset features in the asset library is searched through the search matrix to obtain a matching asset matrix. When entering step S3, the following steps are adopted: D1: Obtain several assets in the preferred asset list, cluster their asset features and determine at least one common asset feature and at least two similar asset features; D2: Determine the matrix position of the selected asset according to the preset sorting rules based on the similar asset characteristics, obtain several groups of matrix positions, calculate the distance of each matrix position in each group of matrix positions and verify them, and eliminate the related asset characteristics whose distance exceeds the threshold; D3: Obtain the preset interception specifications to establish a matrix, and fill in the relevant asset features that have passed the distance verification in several group matrix positions to obtain a search matrix; D4: Search the asset matrix corresponding to the asset features in the asset library by template matching through the search matrix, obtain the matching asset matrix, and go to step S3.

9. The automatic interaction method based on the retail big model according to claim 8, characterized in that: In step D3, a preset interception specification is obtained to establish a matrix, and relevant asset features that have passed the distance verification in several groups of matrix positions are filled in. When the retrieval matrix is ​​obtained, the following steps are adopted to obtain the preset interception specification to establish a matrix, traverse all elements according to several groups of matrix positions, fill in appropriate asset features in turn and ensure that the corresponding asset features meet the distance constraints of each group of matrix positions, and repeatedly obtain the retrieval matrix in turn.

10. An automatic interaction system based on a retail model, using an automatic interaction method based on a retail model as claimed in any one of claims 1 to 9, characterized in that: The automatic interaction system includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the instruction, program, code set or instruction set is loaded and executed by the processor as an automatic interaction method based on a retail large model as described in any one of claims 1 to claim 9.

Images