Object recommendation method and device, electronic equipment, storage medium and program product

Abstract

The invention provides an object recommendation method. The method comprises the following steps: receiving order query information sent by a user; obtaining candidate orders according to the order query information; determining an order query result folding position based on the order query information and the candidate orders; determining a target order from the candidate orders based on the order query result folding position; determining a to-be-recommended target object; and returning the target order and the to-be-recommended target object. The invention further provides an object recommendation device, electronic equipment, a computer readable storage medium and a computer program product.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to an object recommendation method, device, electronic device, storage medium, and program product. Background Art

[0002] With the development of e-commerce, each user has an increasing number of orders. When a user initiates an order query, the number of orders ultimately displayed also increases. It's understandable that users typically initiate order queries to find a specific order, but there are typically many more orders that meet the relevance of the order query. Orders ranked lower in the order list are less likely to be clicked by users, but this can hinder the display of recommended items. Furthermore, when users want to fulfill repeat purchase demands through order queries, item recommendations are also an important way to meet these demands, especially when previously purchased items have expired. However, orders that partially meet the order query relevance but not the user's needs can hinder the display of recommended items, complicating the user's repurchase path and reducing the user experience. Summary of the Invention

[0003] In view of this, embodiments of the present disclosure provide an object recommendation method, apparatus, electronic device, storage medium, and program product, which can solve or partially solve the above-mentioned problems to a certain extent.

[0004] In some embodiments of the present disclosure, the object recommendation method described in the embodiments of the present disclosure may include: receiving order query information sent by a user; obtaining candidate orders based on the order query information; determining the folding position of the order query results based on the order query information and the candidate orders; determining the target order from the candidate orders based on the folding position of the order query results; determining the target object to be recommended; and returning the target order and the target object to be recommended.

[0005] In some embodiments of the present disclosure, obtaining candidate orders based on the order query information includes: recalling historical orders that match the order query information from all historical orders of the user; sorting the historical orders based on at least one feature associated with the historical orders; and determining the candidate order based on the sorted historical orders.

[0006] In some embodiments of the present disclosure, sorting the historical orders based on at least one feature associated with the historical orders includes: determining, for each feature of the at least one feature, a score of the historical order corresponding to the feature; fusing the scores of the historical orders corresponding to the at least one feature to obtain an overall score of the historical orders; and sorting the historical orders based on the overall score of the historical orders.

[0007] In some embodiments of the present disclosure, determining the folding position of the order query result based on the order query information and the candidate order includes: obtaining the sorting position information of the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order; and determining the folding position of the order query result using a trained folding position determination model based on the order query information, the sorting position information of the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order.

[0008] In some embodiments of the present disclosure, determining the folding position of the order query result based on the order query information and the candidate order includes: obtaining the ranking position information of the candidate order, the score of each target feature of the candidate order corresponding to the at least one feature, the order status information of the candidate order, and the status information of the goods related to the candidate order; and determining the folding position of the order query result using a trained folding position determination model based on the order query information, the ranking position information of the candidate order, the score of each target feature of the candidate order corresponding to the at least one feature, the order status information of the candidate order, and the status information of the goods related to the candidate order.

[0009] In some embodiments of the present disclosure, a model training method for a trained folding position determination model includes: obtaining a sample data of a training data set; wherein the sample data includes: sample order query information, a predetermined number of sample order information corresponding to the sample order query information, and a user's first click position; inputting the sample order query information and the sample order information into a neural network model, and receiving an estimated click position output by the neural network model; determining a training loss based on the user's first click position and the estimated click position; training the neural network model based on the training loss; and returning to the step of obtaining a sample data of the training data set until a preset training end condition is met.

[0010] In some embodiments of the present disclosure, the object recommendation method may further include: before determining the target order from the candidate orders based on the folding position of the order query result, correcting the folding position of the order query result based on a preset correction specification.

[0011] In some embodiments of the present disclosure, the folding position of the order query result corresponds to N orders, and determining the target order from the candidate orders based on the folding position of the order query result includes: based on the folding position of the order query result, selecting the top N candidate orders from the candidate orders as the target order; wherein N is a positive integer.

[0012] In some embodiments of the present disclosure, determining the target object to be recommended includes: recalling an object matching the order query information from an object database as the target object to be recommended.

[0013] In some embodiments of the present disclosure, determining the target object to be recommended includes: extracting object information from the target order; and recalling an object matching the object information from an object database as the target object to be recommended.

[0014] In some embodiments of the present disclosure, the object recommendation method may further include: in response to determining that the number of candidate orders is zero, determining a target object to be recommended, and returning prompt information that no relevant order has been retrieved and the target object to be recommended.

[0015] Corresponding to the above-mentioned object recommendation method, an embodiment of the present disclosure further provides an object recommendation device, comprising:

[0016] Information receiving module, used to receive order query information sent by users;

[0017] An order query module, configured to obtain candidate orders based on the order query information;

[0018] a folding position determining module, configured to determine a folding position of an order query result based on the order query information and the candidate orders;

[0019] An order determination module, configured to determine a target order from the candidate orders based on a folding position of the order query result;

[0020] An object recommendation module, used to determine a target object to be recommended; and

[0021] The information output module is used to return the target order and the target object to be recommended.

[0022] In addition, an embodiment of the present disclosure further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the object recommendation method when executing the computer program.

[0023] An embodiment of the present disclosure further provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the object recommendation method.

[0024] An embodiment of the present disclosure further provides a computer program product, comprising computer program instructions, which, when executed on a computer, enable the computer to execute the object recommendation method.

[0025] It can be seen that in the above-mentioned object recommendation method, device, electronic device, storage medium and program product, order query and object recommendation can be comprehensively considered. When displaying the order query results, the order search result folding function is introduced to hide part of the order information and display the target object to be recommended, thereby breaking the logic that the display priority of the order query results is higher than the object recommendation. Under the premise of not affecting the user's normal order query experience, the order query results are intelligently folded according to user needs. While meeting the user's order query needs, objects can also be recommended to the user, thereby increasing the display probability of the recommended object and improving the user's usage experience.

[0026] In some embodiments, the present disclosure introduces an intelligent folding position determination function that can dynamically determine the folding position of order query results based on order query information and related information of candidate orders, rather than folding orders according to a fixed number of displayed orders. This allows for a flexible and effective balance between the user's order query needs and the goal of recommending objects to the user based on the order query information and order query results. In other words, the intelligent folding position determination function described above can achieve the goal of folding other orders that are not the target of the user's search, while still satisfying the user's order query experience. This increases the probability of displaying recommended objects and provides users with a better user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0028] Figure 1 A schematic diagram showing an exemplary system provided by an embodiment of the present disclosure is shown.

[0029] Figure 2 The embodiment shows the implementation process of the object recommendation method described in some embodiments of the present disclosure.

[0030] Figure 3 The embodiment shows the implementation process of the model training method of the folding position determination model described in some embodiments of the present disclosure.

[0031] Figure 4 The internal structure of the object recommendation device described in some embodiments of the present disclosure is shown.

[0032] Figure 5 A more specific schematic diagram of the hardware structure of an electronic device described in some embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0033] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0034] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the usual meanings understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the described object changes, the relative position relationship may also change accordingly.

[0035] It is understandable that before using the technical solutions of each embodiment of the present disclosure, the type, scope of use, usage scenarios, etc. of the personal information involved will be informed to the user in an appropriate manner, and the user's authorization will be obtained.

[0036] For example, in response to a user's active request, a prompt message is sent to the user to clearly inform the user that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the electronic device, application, server, storage medium, or other software or hardware that performs the operation of the disclosed technical solution based on the prompt message.

[0037] As an optional but non-limiting implementation, in response to receiving a user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. Furthermore, the pop-up window may also contain a selection control for the user to select "agree" or "disagree" to provide personal information to the electronic device.

[0038] It is understandable that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0039] As used herein, the term "in response to" refers to a state in which a corresponding event occurs or a condition is satisfied. It will be understood that the timing of executing a subsequent action executed in response to the event or condition is not necessarily strongly correlated with the time when the event occurs or the condition is satisfied. For example, in some cases, a subsequent action may be executed immediately upon the occurrence of the event or the satisfaction of the condition; in other cases, the subsequent action may be executed some time after the occurrence of the event or the satisfaction of the condition.

[0040] As previously mentioned, since an increasing number of orders are typically displayed when users perform order inquiries, some of the displayed order information not only fails to meet user needs but also hinders the display of recommended objects, thereby reducing the user experience. In view of this, embodiments of the present disclosure provide an object recommendation method, apparatus, electronic device, storage medium, and program product that can partially or partially address the aforementioned issues.

[0041] For the sake of clarity in description, before describing the specific technical solutions of the embodiments of the present disclosure, the technical terms involved in the embodiments of the present disclosure are first explained.

[0042] An order can refer to transaction information recorded on an e-commerce platform. Each transaction generated by a user on an e-commerce platform is recorded in the backend system. This single transaction is referred to as an order, and each order is assigned a unique number, also known as an order ID. Information associated with an order typically includes: order ID, order status, order time information, store ID, logistics information, product ID associated with the order, product status, product images or videos, purchase quantity, and purchase price.

[0043] Order query, also known as order search, can be a type of search system. Users can search for orders by entering their order query information in the search box on the order query page. Order query information is typically textual and can be expressed as a query statement. The candidate search content can include all of the user's historical orders.

[0044] The order database may be a database that stores various orders, and each order stored in the order database corresponds to order identification information, so that the order can be uniquely identified by the order identification information. Typically, each order can also be associated with a user ID to identify the entity that placed the order.

[0045] The order click-through rate can refer to the quotient of the number of times a user clicks on an order and the number of times a user sees the order.

[0046] Object recommendations can refer to content recommendations for users on e-commerce platforms, including but not limited to product recommendations, video recommendations, live broadcast recommendations, etc. It is understood that the objects in the above object recommendations can refer to various genres of content displayed on the e-commerce platform, including but not limited to products, videos, live broadcasts, etc.

[0047] The object database may be a database that stores data related to various genres of content provided by the e-commerce platform. The object database is used to store data related to objects.

[0048] A multi-layer perceptron (MLP) is a feedforward neural network. It consists of an input layer, an output layer, and at least one hidden layer (the layer between the input and output layers). Each neuron (or node) is connected to all neurons in the next layer, and these connections have weights.

[0049] Transformer, also known as a converter, can be a deep learning architecture. Its core innovation is to replace the traditional recurrent neural network with the self-attention mechanism, which can significantly improve the performance of natural language processing tasks.

[0050] Immediate interest refers to topics or areas of interest to a user at the current moment. It is a manifestation of a user's short-term interests and is usually triggered by the user's current situation, needs, or behavior. For example, when a user enters a keyword into a search engine, the user's immediate interest is content related to that keyword; when a user browses a specific topic on social media, the user's immediate interest is content related to that specific topic.

[0051] Long-term interests refer to topics or areas that users have been paying attention to and interested in for a long time. These interests may be related to the user's career, hobbies, lifestyle, etc., and usually do not change significantly in a short period of time.

[0052] Short-term interests can refer to topics or areas of interest for a user over a short period of time. Unlike long-term interests, short-term interests are often triggered by specific events, situations, or needs and can change over a short period of time. For example, in the travel sector, a user's short-term interest might be a specific destination or activity; in the news sector, a user's short-term interest might be a breaking event or hot topic.

[0053] Figure 1 FIG. 1 is a schematic diagram of an exemplary system 100 provided by an embodiment of the present disclosure.

[0054] like Figure 1As shown, system 100 may include terminal device 102, terminal device 104, and server 106. A medium (e.g., a network) providing a communication link may be included between terminal device 102, terminal device 104, and server 106. The network may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0055] Exemplarily, the terminal devices 102 and 104 may be installed with an application (APP) or software that can implement a product ordering service. These applications or software usually provide an order query function to facilitate users to query their own historical orders. The terminal devices 102 and 104 here can be hardware or software. When the terminal devices 102 and 104 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, laptop computers (Laptops) and desktop computers (PCs), etc. When the terminal devices 102 and 104 are software, they can be installed in the electronic devices listed above. It can be implemented as multiple software or software modules (for example, to provide distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0056] The server 106 may be a server that provides a product ordering service, such as a backend server that provides support for applications or software on the terminal device 102 or the terminal device 104. The server 106 here may also be hardware or software. When the server 106 is hardware, it may be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server 106 is software, it may be implemented as multiple software or software modules (for example, for providing distributed services), or as a single software or software module. No specific limitation is given here.

[0057] It should be understood that Figure 1 The numbers of terminal devices, users and servers in the embodiment are merely illustrative. Any number of terminal devices, users and servers may be used as required.

[0058] As an example scenario, server 106 may provide a product ordering service, which typically includes an order query function. Users 112 and 114 may use applications on their terminal devices 102 and 104 to submit order query requests containing order query information to server 106. Server 106 searches for relevant orders in the user's entire order history based on the order query information submitted by user 112 and 114 via terminal devices 102 and 104. Server 106 may then provide the resulting order list to users 112 and 114 via terminal devices 102 and 104.

[0059] Based on the above system 100, in order to solve the problem of ineffective object recommendation in the order query process in the related art, the present embodiment provides an object recommendation method. The object recommendation method provided by the present embodiment is described below with reference to specific embodiments and drawings.

[0060] Figure 2 The implementation process of the object recommendation method described in the embodiment of the present disclosure is shown. Figure 2 As shown, the object recommendation method described in the embodiment of the present disclosure can generally be executed by the above-mentioned server 106, and can specifically include the following steps.

[0061] In step 210, order query information sent by the user is received.

[0062] In step 220, candidate orders are obtained according to the order query information.

[0063] In step 230 , the folding position of the order query result is determined based on the order query information and the candidate orders.

[0064] In step 240 , a target order is determined from the candidate orders based on the folding position of the order query result.

[0065] In step 250 , the target object to be recommended is determined.

[0066] In step 260 , the target order and the target object to be recommended are returned.

[0067] It can be seen that in the above-mentioned object recommendation method, order query and object recommendation can be comprehensively considered. When displaying the order query results, the order search result folding function is introduced to hide part of the order information and display the target object to be recommended, thereby breaking the logic that the display priority of the order query results is higher than the object recommendation. Under the premise of not affecting the user's normal order query experience, the order query results are intelligently folded according to user needs. While meeting the user's order query needs, objects can also be recommended to the user, which increases the display probability of the recommended object, shortens the user's order path, and greatly improves the user experience.

[0068] In addition, it is generally believed that if there are too many folded orders in the order query result display page, users may not be able to find their target order, which may lead to a poor user experience. To this end, some embodiments of the present disclosure may introduce an intelligent folding position judgment function, which can dynamically determine the folding position of the order query result based on the order query information and candidate orders, rather than folding orders according to a fixed number. Therefore, it is possible to flexibly and effectively balance the needs of user order queries and the goal of recommending objects to users based on the order query information and order query results. In other words, through the above-mentioned intelligent folding position judgment function, it is possible to fold other orders that are not the user's target search while satisfying the user's experience of order query, thereby increasing the display probability of the recommended objects and providing users with a better user experience.

[0069] The following describes in detail the various steps of the object recommendation method according to the embodiment of the present disclosure with reference to specific examples and accompanying drawings.

[0070] For the above step 210, illustratively, Figure 1 Users 112 and 114 enter the order query information through a search box on an order query function page provided in an application on terminal devices 102 and 104 that enables product ordering services. The application on terminal devices 102 and 104 may generate an order query request based on the order query information entered by the user and send the generated order query request to server 106. Server 106 then retrieves the order query results and feeds them back to terminal devices 102 and 104 for display. In some embodiments, the order query information may be text information and may typically include at least one query keyword.

[0071] Regarding step 220 above, after receiving the order query information input by the user via terminal devices 102 or 104, server 106 may first retrieve historical orders from the user's entire order history that match the received order query information. Typically, there are multiple historical orders that match the order query information. It is understood that the user's entire order history may be stored in an order database. An order database can be understood as a database that stores various orders, and each order stored in the order database is associated with order identification information, allowing the order to be uniquely identified. Furthermore, each order stored in the order database is also associated with a user ID. Thus, in the above step, all historical orders of the user can be retrieved from the order database based on the user ID of the user who sent the order query information. It should be noted that the historical orders may include ongoing orders, completed orders, and closed orders. Active orders may include orders awaiting shipment, orders awaiting payment, shipped orders, and refunded / after-sales orders. Closed orders may include unpaid orders and orders with refund costs. After recalling historical orders matching the received order query information from the user's entire historical order history, server 106 may further sort the recalled historical orders based on at least one characteristic associated with the historical orders and determine the candidate orders based on the sorted historical orders. As can be seen, step 220 may primarily include the process of recalling historical orders and the process of sorting historical orders. In particular, in some embodiments, step 220 may further include a process of selecting historical orders to control the number of candidate orders determined, thereby adapting to the input requirements of the folding position determination model used in subsequent steps.

[0072] Regarding the recall process of the above-mentioned historical orders, as mentioned above, the above-mentioned order query information can be text information, usually including at least one query keyword. In this case, the server 106 can determine the user's needs by analyzing the order query information. For example, the above-mentioned analysis may include operations such as word segmentation and rewriting. Specifically, the search terms can be generated based on the order query information through the above-mentioned analysis, and then a text search can be performed based on the generated search terms to recall historical orders that match the search terms from all historical orders of the user. In some embodiments of the present disclosure, the relevant information of a historical order may include: order identification information, order time information, order status information, order product information (for example, it may include product identification, product images and / or videos, purchase quantity or price information, etc.), order store information, and order logistics information, etc.

[0073] Specifically, in the process of recalling historical orders, in some embodiments, the order query information can be directly used as the above-mentioned search term, and then the historical orders of the above-mentioned user containing the above-mentioned order query information can be searched from the order database.

[0074] In other embodiments, the order query information may be first segmented (which may further include removing stop words), and the resulting query keywords may be used as the search terms. The query keywords may be understood as key information in the order query information. The order database is then searched for the user's historical orders that match each query keyword.

[0075] Taking into account that at least one query keyword extracted from the order query information may have different expressions from the information stored in the order database, the historical orders of the user queried from the order database based on each query keyword may be missed. Therefore, in some other embodiments, the query keywords after word segmentation can be further rewritten to obtain synonyms, antonyms and / or intention words corresponding to each query keyword as rewritten words, and the rewritten words obtained after rewriting are respectively used as the above-mentioned search terms. Among them, rewriting each query keyword can also be called expansion processing of each query keyword. Further, the determined search term can be matched with each order-related information in all the historical orders of the user to determine the historical orders of the user that match the search term.

[0076] As can be seen, the number of search terms determined by one or more of the above methods may be multiple. In this case, the recall process can be performed separately based on multiple recall links currently known and / or to be developed in the future, and the recall results can be merged to obtain the historical orders matching the search terms. For example, in some embodiments, one recall link can directly recall based on order query information; another or more recall links can recall based on coarse-grained word segmentation; yet another or more recall links can recall based on rewritten words, and so on. It can be understood that by setting up multiple recall links to perform the recall process, the problem of missed recalls can be effectively avoided. Furthermore, the recall process can have two possible outcomes: a recall result and no recall result. If no recall result is obtained, a prompt message indicating that no matching orders were found will be returned to the terminal device. If a recall result is obtained, a sorting process can be further performed to sort the recall results, ultimately obtaining the candidate order information.

[0077] Regarding the above-mentioned sorting process of historical orders, sorting historical orders based on at least one feature associated with the historical orders may specifically include: for each historical order and each feature in the at least one feature, determining the score of the historical order corresponding to the above feature; for each historical order, fusing the scores of each feature in the at least one feature corresponding to the historical order to obtain the total score of each historical order; and sorting the historical orders based on the total score of each historical order. In particular, if only one feature is involved in the above-mentioned sorting, the step of fusing the scores of each feature in the at least one feature corresponding to the historical order may be omitted, and the score of the historical order corresponding to the above feature may be directly used as the total score of the historical order. It should be noted that the features used in the above-mentioned sorting process may be any pre-set features associated with the historical orders.

[0078] Specifically, in some embodiments, historical orders can be sorted based on a single feature. For example, the at least one feature associated with the historical orders can be the time information of the historical orders, and the score for the time information can be directly used as the total score of the historical orders. In some examples, the time information itself can represent the score for the time information, and the historical orders can be sorted in the forward or reverse order of time according to the time information in the historical orders. For another example, the at least one feature associated with the historical orders can be the matching degree between the order query information and the historical orders, and the score for the matching degree can be directly used as the total score of the historical orders. In some examples, the matching degree itself can represent the score for the matching degree, and the historical orders can be sorted in descending order according to the matching degree between the order query information and the historical orders. It can be understood that the method of sorting historical orders based on a single feature has the characteristics of simple implementation and low computational complexity.

[0079] In other embodiments, historical orders can be ranked by comprehensively considering multiple features. For example, the at least one feature associated with historical orders may include: the time information of the historical order, the degree of match between the order query information and the historical order, and the click-through rate of the historical order. In this case, multiple scores can be determined for each historical order, including a time score (determined based on the time information of the historical order), a relevance score (determined based on the degree of match between the order query information and the historical order), and a click-through rate score (determined based on the click-through rate of the historical order). The time score is used to assess the time proximity of an order. Generally, orders with closer time (generally, higher time scores) have a higher probability of meeting user needs. The relevance score is used to assess the relevance between an order and the order query information entered by the user. Generally, orders with higher relevance (generally, higher relevance scores) have a higher probability of meeting user needs. The click-through rate score is used to assess the degree to which an order meets user needs. Generally, orders with higher click-through rate scores have a higher probability of being selected by the user. It will be understood that in order to achieve score fusion, the scores for each feature should be normalized. Then, score fusion is performed for each historical order to obtain the total score of each historical order. In the embodiment of the present disclosure, a specific fusion method can be used, for example, a weighted summation method, to fuse the scores of various features of the historical orders to obtain the total score of the historical orders. Among them, the weight coefficient corresponding to each feature can be flexibly determined according to the actual situation, and the embodiment of the present disclosure is not limited to this. It can be understood that the method of sorting historical orders based on multiple features can comprehensively evaluate the possibility of an order meeting the user's needs from multiple dimensions, thereby effectively achieving the goal of sorting orders that meet user needs first.

[0080] As previously mentioned, in some embodiments, to accommodate the input requirements of the folding position determination model used in subsequent steps, step 220 may further include a historical order selection process to control the number of candidate orders determined. In this case, based on the maximum number of candidate orders input as defined by the folding position determination model, a corresponding number of historical orders may be selected from the sorted historical orders in sorted order as candidate orders. It will be appreciated that in some embodiments, the maximum number of candidate orders input as defined by the folding position determination model may be the maximum number of orders that can be displayed on the order query result display page.

[0081] It can be seen from this that after the above-mentioned historical order recall process and historical order sorting process, or further after the above-mentioned historical order selection process, the above-mentioned candidate orders can be obtained, and the above-mentioned candidate orders are sorted historical orders.

[0082] With respect to the above step 230 , the folding position of the above order query result may be determined based on a trained folding position determination model.

[0083] It should be noted that the order query result folding position can represent the number of orders displayed on the order query result display page. For example, if the order query result folding position is 3, it means that the first three candidate orders found on the order query result display page will be displayed, and the fourth candidate order will be hidden.

[0084] Specifically, in some embodiments of the present disclosure, the above step 230 may include: first, obtaining the sorting position information of the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order; then, based on the order query information, the sorting position information of the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order, using the trained folding position determination model to determine the folding position of the order query result. The sorting position information of the candidate order represents the number of the candidate order after the sorting of the candidate orders. The order status information may refer to the information of the current status of the order. The current status may be, for example, in progress, completed, closed, etc. Among them, the ongoing status can be further subdivided into pending shipment, pending payment, shipped, refunded / after-sales, etc. The status information of the goods includes valid (indicating that it can be ordered again) or expired (indicating that it cannot be ordered again), etc. In this way, the folding position of the order query results is determined based on the sorting position information, order status information and status information of the products related to the candidate orders of each candidate order, which can more intelligently determine where to fold. Therefore, while satisfying the user's experience of order query, other orders that are not the user's target search can be folded, providing users with a better user experience.

[0085] Specifically, in some other embodiments of the present disclosure, the above-mentioned step 230 may include: first, obtaining the sorting position information of the candidate order, the score of each target feature in the at least one feature corresponding to the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order; then, based on the order query information, the sorting position information of the candidate order, the score of each target feature in the at least one feature corresponding to the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order, the trained folding position determination model is used to determine the folding position of the order query result. In this way, the folding position of the order query result is determined according to the sorting position information of each candidate order, the score of the feature, the order status information, and the status information of the goods related to the candidate order, which can more intelligently determine where to fold, thereby folding other orders that are not the user's target search while satisfying the user's experience of order query, thereby providing the user with a better user experience.

[0086] In some other embodiments of the present disclosure, the user's characteristics can be used as one of the input features of the folding position determination model on the basis of the above-mentioned input information. In some specific examples, the above-mentioned user characteristics can be obtained based on the information associated with the above-mentioned candidate orders or based on the user's behavior data. The above-mentioned user's behavior data may refer to all the user's interactive behavior data on the e-commerce platform obtained with the user's authorization, such as click behavior data, browsing behavior data, etc. By injecting the user's characteristics into the folding position determination model, the folding position determination model can be further assisted in determining the folding position of the order query result, so that the output order query result folding position is more in line with the user's own characteristics, that is, more accurate.

[0087] It should be understood that, as mentioned above, before using the technical solutions disclosed in the embodiments of this application, the type, scope of use, and usage scenarios of the personal information involved in this application should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations. For example, in response to receiving an active request from the user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require the acquisition and use of the user's personal information. Thus, the user can independently choose whether to provide personal information to the electronic device, application, server, storage medium, or other software or hardware that performs the operation of the technical solution of this application based on the prompt message.

[0088] As an optional but non-limiting implementation method, in response to receiving an active request from the user, the method of sending a prompt message to the user can be, for example, a pop-up window, and the prompt message can be presented in the form of text in the pop-up window. In addition, the pop-up window can also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device. It is understandable that the above notification and user authorization process are only illustrative and do not constitute a limitation on the implementation method of this application. Other methods that meet the relevant laws and regulations can also be applied to the implementation method of this application. It is understandable that the user personal information data involved in this technical solution (including but not limited to the data itself, the acquisition, storage and use of the data) should comply with the requirements of the relevant laws, regulations and relevant provisions, and shall not violate public order and good morals.

[0089] In addition, it should be noted that in order to ensure the load of the folding position determination model, the maximum number of candidate orders input into the folding position determination model can be set. The above maximum number can be designed based on the performance load of the neural network and the posterior statistics of the data. In some embodiments, the above maximum number can also be set as the maximum number of candidate orders displayed on the order query result display page. When the maximum number of candidate orders input into the folding position determination model is set, and the total number of historical orders recalled in the above step 220 is greater than the above maximum number, in the above step 220, after sorting the historical orders, the above maximum number of historical orders ranked first can be selected from the sorted historical orders based on the above maximum number as the candidate orders.

[0090] In the embodiment of the present disclosure, the model training method of the trained folding position determination model can be Figure 3 The process shown is implemented. Figure 3 As shown, the above model training method may include the following steps.

[0091] In step 310, a sample data of the training data set is obtained.

[0092] In an embodiment of the present disclosure, the above-mentioned sample data may include: sample order query information, a predetermined number of sample order information corresponding to the sample order query information, and the user's first click position.

[0093] It is understood that the sample data in the training dataset can be obtained by processing historical order query data stored on the product ordering platform. The sample order query information can be the order query information actually entered by the user during the order query process. The predetermined number of sample order information corresponding to the sample order query information can be the predetermined number of order information included in the order query results actually returned to the user. For example, the sample order information can include: order ranking position information, order status information, and status information of the products related to the order. In addition, in some embodiments, the sample order information can also include: scores corresponding to various features of the order, such as the order time score, order relevance score, and order click-through rate score. Furthermore, the predetermined number can be the maximum number of candidate orders for an input fold position determination model designed based on the performance load of the neural network and data posterior statistics. Typically, the predetermined number can also refer to the maximum number of candidate orders displayed on the order query result display page. The user's first click position can refer to the position of the order that the user first clicked on the order query result display page, that is, the order on which the user first clicked on the order query result display page. In particular, if the user does not actually click on any order displayed on the order query result display page and does not click on the option to view more orders, the user's first click position can be set to 0. If the user does not actually click on any order displayed on the order query result display page but clicks on the option to view more orders, the user's first click position can be set to the maximum number of candidate orders input into the folding position determination model plus 1. That is, assuming that the maximum number of candidate orders that can be input into the folding position determination model is M, if the user does not actually click on any order displayed on the order query result display page but clicks on the option to view more orders, the user's first click position can be set to M+1, where M is a positive integer.

[0094] In other embodiments of the present disclosure, the sample data may further include user characteristics. The user characteristics may be obtained based on information associated with the candidate order or based on the user's behavioral data. Furthermore, as previously mentioned, the user's behavioral data is obtained with the user's authorization.

[0095] In step 320, the sample order query information and the sample order information are input into the neural network model, and the estimated click position output by the neural network model is received.

[0096] In the embodiments of the present disclosure, the above-mentioned neural network model can adopt various neural network structures such as MLP or Transformer.

[0097] At step 330 , a training loss is determined based on the user's first click location and the estimated click location.

[0098] In some embodiments of the present disclosure, cross entropy loss or ordinal regression can be used as the loss function for determining the training loss. It should be noted that the above loss function is only for illustration, and the embodiments of the present disclosure do not limit the specific method for determining the training loss.

[0099] At step 340 , the neural network model is trained based on the training loss.

[0100] In step 350 , the process returns to step 310 until a preset training end condition is met.

[0101] It can be understood that in the embodiments of the present disclosure, the above-mentioned training end conditions may include: reaching a preset training round or the neural network model has converged, etc.

[0102] It can be seen from this that after the above training process, the folding position determination model obtained can determine the folding position of the order query result based on the input order query information, the sorting position information of the order, the order status information of the order, and the status information of the order-related goods, and can even include the score of each target feature in at least one feature corresponding to the order and / or the characteristics of the user. By using the folding position determination model to determine the folding position of the order query result, for each order query request, the order query information and the relevant information of the candidate order can be considered, and even the user's own characteristics can be considered. The user's needs can be intelligently judged, and the folding position of the order query result can be flexibly determined while ensuring the user's experience in order querying, rather than selecting a fixed folding position to fold the order. Therefore, it is possible to flexibly and effectively balance the needs of the user's order query and the goal of recommending objects to the user, effectively increase the display probability of the recommended object, shorten the user's order path, and optimize the user's usage experience.

[0103] After determining the folding position of the order query result, and before step 240, the object recommendation method may further include: correcting the folding position of the order query result based on pre-set correction specifications. Specifically, the correction rules may include: an order query result folding position exceeding limit correction rule, an output abnormality correction rule, and an operation control correction rule, etc. Among them, the order query result folding position exceeding limit correction rule may include: pre-setting a maximum threshold and a minimum threshold for the order query result folding position, and in response to determining that the folding position of the order query result output by the folding position determination model is greater than the above maximum threshold, correcting the folding position of the order query result to the above maximum threshold or a pre-set default position value; and in response to determining that the folding position of the order query result output by the folding position determination model is less than the above minimum threshold, correcting the folding position of the order query result to the above minimum threshold or a pre-set default position value. The output abnormality correction rule may include: in response to determining that the folding position determination model does not output the folding position of the order query result or outputs an abnormal value, correcting the folding position of the order query result to a pre-set default position value. Among them, the above-mentioned default position value can be any value that is less than or equal to the above-mentioned maximum value threshold and greater than or equal to the above-mentioned minimum value threshold, and can be flexibly set according to actual conditions. The operation control correction rule may include: for order query information or order query results that meet specific conditions, the folding position of the above-mentioned order query result can be increased or decreased based on a pre-set adjustment method. For example, assuming that it can be determined through the order query information that the queried order belongs to certain specific categories that need to be recommended, the folding position of the above-mentioned order query result can be reduced in a pre-set adjustment method, for example, the folding position of the order query result is reduced by a fixed step value, so that the query result returned to the terminal has less order information and more recommended objects. And assuming that it can be determined through the order query information that the queried order belongs to certain specific categories that need to be reduced in recommendation, the folding position of the order query result can be increased in a pre-set adjustment method, for example, the folding position of the order query result is increased by a fixed step value, so that the query result returned to the terminal has more order information and fewer recommended objects.

[0104] With respect to step 240 above, a candidate order ranked higher in the order query result can be selected as the target order based on the collapsed position of the order query result. Specifically, assuming that the collapsed position of the order query result corresponds to N orders, the N candidate orders ranked higher in the order query result can be selected as the target order; where N is a positive integer. As can be seen from the preceding steps, the candidate orders are historical orders that have already been sorted. Therefore, in step 240 above, the N candidate orders ranked higher in the order query result can be directly selected as the target order, which can basically ensure that the selected target order has a high probability of being the historical order that the user is actually looking for.

[0105] Regarding the above step 250, the process of determining the target object to be recommended can generally be performed in a variety of ways.

[0106] In some embodiments, target objects to be recommended can be determined based on order query information. Specifically, objects matching the order query information can be retrieved from an object database as the target objects to be recommended. It is understood that the above process can generally include an object recall process, an object ranking process, and an object selection process. Specifically, the object recall process can refer to the recall process of historical orders described above. For example, search terms are first generated based on the order query (which may include one or more steps such as word segmentation, stop word removal, and rewriting). Based on the obtained search terms, objects matching the order query information are retrieved from the object database through one or more recall links. These objects can be referred to as candidate recommendation objects. The object database can be understood as a database that stores e-commerce search results for various genres, and the object database is used to provide object information data. Furthermore, the object ranking process can also refer to the ranking process of historical orders described above. For example, the scores of candidate recommendation objects for at least one feature are first determined. The scores of the candidate recommendation objects for at least one feature are then merged to obtain the total score of the candidate recommendation objects. Finally, the candidate recommendation objects are ranked according to their total score. Finally, a predetermined number of candidate recommendation objects ranked first are selected from the candidate recommendation objects after the above sorting as the target objects to be recommended. It should be noted that the features involved in scoring the candidate recommendation objects may be different from the features involved in sorting the historical orders. For example, the features involved in scoring the candidate recommendation objects may generally include relevance, click-through rate, transaction volume, etc. It can be seen that by providing users with target objects to be recommended that match the order query information, the goal of using the order query information input by the user as the user's immediate interest to obtain recommendation objects related to the user's immediate interest is achieved, so that recommended objects that meet the user's immediate needs can be recommended to the user, which improves the accuracy of object recommendation. Therefore, when the user wants to make a repeat purchase through order query, the user's repeat purchase needs can be more effectively met, especially when the goods in the order queried by the user have expired, which greatly improves the user's usage experience.

[0107] In other embodiments, the target objects to be recommended can be determined based on the target order. Specifically, object information can first be extracted from the determined target order; then, based on the extracted object information, objects matching the object information are retrieved from the object database as the target objects to be recommended. The object information can be at least one of object identification information and an image / video of the object. The above process can also primarily include an object recall process, an object sorting process, and an object selection process. Specifically, the object recall process can refer to the recall process described in the above embodiment, the object sorting process can refer to the sorting process described in the above embodiment, and the object selection process can refer to the selection process described in the above embodiment, and will not be repeated here. As can be seen, by providing the user with target objects to be recommended that match the target order information (i.e., the order query result), the goal of using the target order information as the user's immediate interest to obtain recommended objects related to the user's immediate interest is achieved. This allows the user to be recommended objects that meet their immediate needs, improving the accuracy of object recommendations. This can more effectively meet the user's repurchase needs when the user wishes to make a repeat purchase through order query, especially when the product in the order being queried has expired, greatly improving the user's user experience.

[0108] In some other embodiments, the target objects to be recommended can also be determined based on the user's characteristics. As mentioned above, in some specific examples, the above-mentioned user characteristics can be obtained based on the information associated with the above-mentioned candidate orders or based on the user's behavioral data. Moreover, the above-mentioned user's behavioral data can refer to the user's behavioral data obtained with the user's authorization. Considering that the user's preferences or preferences can be understood based on the user's authorized user behavioral data, such as the user's long-term interests and short-term interests. Therefore, the embodiments of the present disclosure can determine the user's long-term interests and / or short-term interests based on the user's authorized user behavioral data. Then, based on the user's long-term interests and / or short-term interests, candidate recommendation objects are obtained from the object database. In some optional embodiments, when determining the user's long-term interests and / or short-term interests, the user's behavioral data can be analyzed to determine the user's long-term interests, such as objects that the user has been paying attention to and interested in for a long period of time, and the user's short-term interests, such as objects that the user has been interested in for a short period of time. Then, all objects related to the user's long-term interests and / or short-term interests are obtained from the object database as the above-mentioned target objects to be recommended. It can be seen that the above embodiment can determine the user's long-term interests and / or short-term interests based on the user's behavioral data authorized by the user, and then recommend objects to the user based on the user's long-term interests and / or short-term interests, so as to provide the user with recommended objects that meet the user's demands based on the user's preferences, thereby improving the user's shopping experience and providing favorable conditions for improving the object conversion rate.

[0109] It should be noted that in the embodiments of the present disclosure, the above-mentioned multiple methods of determining the target objects to be recommended can be combined. For example, the first part of candidate recommendation objects can be determined based on order query information; the second part of candidate recommendation objects can be determined based on the target order; the third part of candidate recommendation objects can be determined based on the user's characteristics; then, all or part of the above-mentioned first part of candidate recommendation objects and / or second part of candidate recommendation objects and / or third part of candidate recommendation objects can be used as the above-mentioned target objects to be recommended. The above-mentioned method can comprehensively consider the user's immediate interests and / or short-term interests and / or long-term interests when making object recommendations, thereby making it easier to meet the user's needs. In the above examples, the recall and sorting methods for candidate recommendation objects can refer to the previous embodiments and will not be repeated here.

[0110] Furthermore, it should be noted that, in addition to the above-described method, embodiments of the present disclosure may also employ any other object recommendation method to determine the target object to be recommended. For example, recommendations may be made based on the popularity of the object, or recommendations may be made randomly, etc. The embodiments of the present disclosure do not limit the method for determining the target object to be recommended.

[0111] With respect to step 260, after determining the target order and the target object to be recommended, the server 106 may return the target order and the target object to be recommended to the terminal devices 102 and 104. Next, the terminal devices 102 and 104 may display the target order and the target object to be recommended to the users 112 and 114 via the order query result display page, so as to provide the users with order query result feedback and object recommendations.

[0112] As mentioned above, in the process of recalling historical orders, two results may occur, obtaining a recall result and not obtaining a recall result. The processing scheme for obtaining a recall result has been described in detail in the previous embodiment. If no recall result is obtained, the above method may further include: in response to determining that the number of the above candidate orders is zero, the target object to be recommended may be determined first; then, a prompt message that the relevant order has not been retrieved and the determined target object to be recommended are returned. In this case, after receiving the prompt message that the relevant order has not been retrieved and the determined target object to be recommended, the terminal device will display the prompt message that the relevant order has not been retrieved and the target object to be recommended to the user through the order query result display page to recommend the object to the user. The method for determining the target object to be recommended described in the above embodiment can also refer to the previous embodiment and will not be repeated here.

[0113] Corresponding to the above-mentioned object recommendation method, some embodiments of the present disclosure further disclose an object recommendation device. Figure 4The internal structure of the object recommendation device according to the embodiment of the present disclosure is shown. Figure 4 As shown, the above-mentioned object recommendation device may include the following multiple modules:

[0114] The information receiving module 410 is used to receive order query information sent by the user;

[0115] The order query module 420 is used to obtain candidate orders based on the order query information;

[0116] A folding position determining module 430 is configured to determine a folding position of an order query result based on the order query information and the candidate orders;

[0117] An order determination module 440 is configured to determine a target order from candidate orders based on a folding position of the order query result;

[0118] The object recommendation module 450 is used to determine the target object to be recommended; and

[0119] The information output module 460 is used to return the target order and the target object to be recommended.

[0120] It should be noted that the implementation method of each module in the above-mentioned object recommendation device and the specific technical effects that can be achieved can refer to the implementation method of each step in the above-mentioned embodiment, and will not be repeated here.

[0121] From this, it can be seen that the above-mentioned object recommendation device can comprehensively consider order query and object recommendation, and introduce the order search result folding function when displaying the order query results, which can hide part of the order information and display the target object to be recommended, thereby breaking the logic that the display priority of the order query results is higher than the object recommendation. Under the premise of not affecting the user's normal order query experience, the order query results are intelligently folded according to user needs. While meeting the user's order query needs, objects can also be recommended to the user, which increases the display probability of the recommended object, shortens the user's order path, and greatly improves the user's usage experience.

[0122] In addition, it is generally believed that if there are too many folded orders in the order query result display page, users may not be able to find their target orders, which may lead to a poor user experience. To this end, some embodiments of the present disclosure introduce an intelligent folding position judgment function, which can dynamically determine the folding position of the order query results based on the order query information and candidate orders, rather than folding orders according to a fixed number. Therefore, it is possible to flexibly and effectively balance the needs of user order queries and the goal of recommending objects to users based on the order query information and order query results. In other words, through the above-mentioned intelligent folding position judgment function, it is possible to fold other orders that are not the user's target search while satisfying the user's experience of order query, thereby increasing the display probability of the recommended objects and providing users with a better user experience.

[0123] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the object recommendation method described in any of the above embodiments when executing the computer program.

[0124] Figure 5 1 is a schematic diagram showing the hardware structure of a more specific electronic device provided in this embodiment. The device may include: a processor 2010, a memory 2020, an input / output interface 2030, a communication interface 2040, and a bus 2050. The processor 2010, the memory 2020, the input / output interface 2030, and the communication interface 2040 are communicatively connected to each other within the device via the bus 2050.

[0125] The processor 2010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0126] The memory 2020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 2020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 220 and is called and executed by the processor 2010.

[0127] The input / output interface 2030 is used to connect input / output devices to enable information input and output. Input / output devices can be configured as components within the device or externally connected to the device to provide corresponding functions. Input devices may include microphones and various sensors, while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0128] The communication interface 2040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WiFi, Bluetooth, etc.).

[0129] The bus 2050 comprises a path for transmitting information between the various components of the device (eg, the processor 2010 , the memory 2020 , the input / output interface 2030 , and the communication interface 2040 ).

[0130] It should be noted that although the above device only shows the processor 2010, the memory 2020, the input / output interface 2030, the communication interface 2040, and the bus 2050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, those skilled in the art will understand that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0131] The electronic device of the above embodiment is used to implement the corresponding object recommendation method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0132] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the object recommendation method described in any of the above embodiments.

[0133] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0134] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the object recommendation method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0135] Based on the same inventive concept, corresponding to the object recommendation method of any of the above embodiments, the present disclosure further provides a computer program product comprising computer program instructions. In some embodiments, when the computer program instructions are executed on a computer, the computer executes each step of each embodiment of the object recommendation method. Corresponding to the execution subject corresponding to each step in each embodiment of the object recommendation method, the processor executing the corresponding step may belong to the corresponding execution subject.

[0136] The computer program product of the above embodiment is used to enable a processor to execute the object recommendation method described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0137] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Within the scope of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0138] In addition, to simplify the description and discussion, and so as not to obscure the embodiments of the present disclosure, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present disclosure, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure are to be implemented (i.e., these details should be fully within the purview of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it will be apparent to those skilled in the art that the embodiments of the present disclosure may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0139] Although the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0140] The embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. An object recommendation method, comprising: Receive order inquiry information sent by users; Obtain candidate orders according to the order query information; Determining a folding position of an order query result based on the order query information and the candidate orders; Determining a target order from the candidate orders based on a folding position of the order query result; Determine the target audience for recommendation; as well as Return the target order and the target object to be recommended.

2. The method according to claim 1, wherein Acquiring candidate orders according to the order query information includes: Recalling historical orders that match the order query information from all historical orders of the user; sorting the historical orders based on at least one characteristic associated with the historical orders; and The candidate orders are determined based on the sorted historical orders.

3. The method according to claim 2, wherein: The sorting of the historical orders based on at least one feature associated with the historical orders includes: For each of the at least one feature, determining a score of the historical order corresponding to the feature; fusing the scores of the historical orders corresponding to the respective features of the at least one feature to obtain an overall score of the historical orders; and The historical orders are sorted based on their total scores.

4. The method according to claim 1, wherein Determining the folding position of the order query result based on the order query information and the candidate orders includes: Obtaining the sorting position information of the candidate order, the order status information of the candidate order, and the status information of the products related to the candidate order; and The folding position of the order query result is determined using a trained folding position determination model based on the order query information, the sorting position information of the candidate order, the order status information of the candidate order, and the status information of the goods related to the candidate order.

5. The method according to claim 3, wherein Determining the folding position of the order query result based on the order query information and the candidate orders includes: Obtaining ranking position information of the candidate order, a score of the candidate order corresponding to each target feature of the at least one feature, order status information of the candidate order, and status information of products related to the candidate order; and Based on the order query information, the ranking position information of the candidate order, the score of each target feature of the at least one feature corresponding to the candidate order, the order status information of the candidate order, and the status information of the products related to the candidate order, the trained folding position determination model is used to determine the folding position of the order query result.

6. The method according to claim 4 or 5, wherein: The model training method of the trained folding position determination model includes: Obtain a sample data of a training data set; wherein the sample data includes: sample order query information, a predetermined number of sample order information corresponding to the sample order query information, and a user's first click position; Inputting the sample order query information and the sample order information into a neural network model, and receiving an estimated click position output by the neural network model; Determining a training loss based on the user's first click location and the estimated click location; Training the neural network model based on the training loss; and Return to the step of obtaining a sample data of the training data set until a preset training end condition is met.

7. The method according to claim 1, further comprising: Before determining at least one target order from the candidate orders based on the folding position of the order query result, the folding position of the order query result is corrected based on a preset correction specification.

8. The method according to claim 1, wherein The folding position of the order query result corresponds to N orders, and determining the target order from the candidate orders based on the folding position of the order query result includes: based on the folding position of the order query result, selecting the top N candidate orders from the candidate orders as the target orders; wherein N is a positive integer.

9. The method according to claim 1, wherein: The determining of the target object to be recommended includes: recalling an object matching the order query information from an object database as the target object to be recommended.

10. The method according to claim 1, wherein Determining the target object to be recommended includes: Extracting object information from the target order; and An object matching the object information is retrieved from an object database as the target object to be recommended.

11. The method according to claim 1 , further comprising: In response to determining that the number of the candidate orders is zero, a target object to be recommended is determined, and prompt information indicating that no relevant order has been retrieved and the target object to be recommended are returned.

12. An object recommendation device, comprising: Information receiving module, used to receive order query information sent by users; An order query module, configured to obtain candidate orders based on the order query information; a folding position determining module, configured to determine a folding position of an order query result based on the order query information and the candidate orders; An order determination module, configured to determine a target order from the candidate orders based on a folding position of the order query result; The object recommendation module is used to determine the target object to be recommended; as well as The information output module is used to return the target order and the target object to be recommended.

13. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the object recommendation method according to any one of claims 1 to 11 when executing the computer program.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the object recommendation method according to any one of claims 1 to 11. 15 . A computer program product comprising computer program instructions, which, when executed on a computer, enable the computer to execute the object recommendation method according to claim 1 .

Images