Menu recommendation method and device, computer equipment, storage medium and program product

Abstract

The invention relates to a menu recommendation method and apparatus, a computer device, a storage medium and a program product. The method comprises the steps of obtaining service data of a current service, processing user data of a target operator of the current service, and menu data of each service menu associated with the current service; performing feature extraction on the business data, the user data and the menu data to obtain a target feature vector; wherein the target feature vector is used for reflecting the post adaptation degree, the scene matching degree, the business association degree and the operation habit integrating degree of the target operator; selecting a target recommendation menu from the business menus according to the target feature vector; and displaying the target recommendation menu to a terminal where the target operator is located. According to the scheme, in the process of determining the target recommended menu, the post adaptation degree, the scene matching degree, the business association degree and the operation habit integrating degree of the target operator are considered, so that the determined target recommended menu is more adaptive to the current business, and the adaptability of the recommended menu is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a menu recommendation method, apparatus, computer device, storage medium, and program product. Background Technology

[0002] Currently, when employees conduct business, the business system usually automatically recommends function menus to them to improve business processing efficiency.

[0003] In traditional technologies, business systems typically recommend menus that employees use more frequently based on their past operation records. This results in recommended menus that are not suitable for the current business scenario, leading to a problem of poor adaptability of recommended menus. Summary of the Invention

[0004] Therefore, it is necessary to provide a menu recommendation method, apparatus, computer device, storage medium, and program product to address the aforementioned technical problems, thereby improving the adaptability of the recommended menu.

[0005] Firstly, this application provides a menu recommendation method, including:

[0006] Acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business;

[0007] Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator;

[0008] Based on the target feature vector, select the target recommendation menu from each business menu;

[0009] The target recommended menu is displayed to the terminal where the target operator is located.

[0010] In one embodiment, selecting the target recommendation menu from each business menu based on the target feature vector includes:

[0011] Based on the target feature vector and the candidate feature vectors of each candidate operator, similar operators that are similar to the target operator are selected from the candidate operators.

[0012] Based on the historical operation records of the similar operators, select candidate recommended menus from each business menu;

[0013] Based on the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, the target recommended menu is selected from the candidate recommended menu.

[0014] In one embodiment, the step of selecting similar operators from among the candidate operators that are similar to the target operator based on the target feature vector and the candidate feature vectors of each candidate operator includes:

[0015] For each candidate operator, determine the similarity between the target feature vector and the candidate feature vector of the candidate operator;

[0016] The similarity scores are sorted from highest to lowest to obtain the first sorting order;

[0017] The candidate operators corresponding to the first preset number of similarities in the first sorting order are regarded as similar operators to the target operator.

[0018] In one embodiment, the step of selecting a candidate recommended menu from each business menu based on the historical operation records of the similar operators includes:

[0019] Based on the historical operation records of the similar operators, determine the number of historical operations for each business menu;

[0020] Sort each business menu in descending order according to the historical number of operations for each business menu to obtain the second sorting order;

[0021] The first second preset number of business menus in the second sorting order are selected as candidate recommended menus.

[0022] In one embodiment, the step of filtering the target recommended menu from the candidate recommended menu according to the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data includes:

[0023] For each candidate recommended menu, the arrangement number of the candidate recommended menu in the second arrangement order is adjusted according to the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, so as to obtain a new arrangement number;

[0024] Based on the new arrangement number of each candidate recommended menu, sort the candidate recommended menus in descending order to obtain the third arrangement order;

[0025] The first three preset number of candidate recommended menus in the third sorting order are taken as the target recommended menu.

[0026] In one embodiment, the scenario weight of each candidate recommendation menu in the business scenario corresponding to the business data is determined based on a knowledge graph, which is constructed in the following way:

[0027] Obtain sample data; wherein, the sample data includes full menu semantic tags and business operation logs;

[0028] Data relationships are obtained from the sample data; wherein, the data relationships include the relationship between the operator's position and the menu, and the relationship between different menus in each business scenario;

[0029] The knowledge graph is constructed based on the data relationships.

[0030] Secondly, this application also provides a menu recommendation device, comprising:

[0031] The acquisition module is used to acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business.

[0032] The extraction module is used to extract features from the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator;

[0033] The determination module is used to select the target recommendation menu from each business menu based on the target feature vector;

[0034] The display module is used to display the target recommended menu to the terminal where the target operator is located.

[0035] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0036] Acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business;

[0037] Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator;

[0038] Based on the target feature vector, select the target recommendation menu from each business menu;

[0039] The target recommended menu is displayed to the terminal where the target operator is located.

[0040] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0041] Acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business;

[0042] Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator;

[0043] Based on the target feature vector, select the target recommendation menu from each business menu;

[0044] The target recommended menu is displayed to the terminal where the target operator is located.

[0045] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0046] Acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business;

[0047] Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator;

[0048] Based on the target feature vector, select the target recommendation menu from each business menu;

[0049] The target recommended menu is displayed to the terminal where the target operator is located.

[0050] The aforementioned menu recommendation method, apparatus, computer equipment, storage medium, and program product acquire business data of the current business, user data of the target operator processing the current business, and menu data of each business menu associated with the current business. Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector. This target feature vector reflects the target operator's job suitability, scenario matching, business relevance, and operational habit fit. Based on the target feature vector, a target recommended menu is selected from each business menu. This target recommended menu is then displayed to the terminal where the target operator is located. This solution, in determining the target recommended menu, takes into account the target operator's job suitability, scenario matching, business relevance, and operational habit fit, making the determined target recommended menu more suitable for the current business, thereby improving the adaptability of the recommended menu. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating a menu recommendation method in one embodiment;

[0053] Figure 2 This is a flowchart illustrating the process of selecting a target recommended menu from various business menus in one embodiment.

[0054] Figure 3 This is a flowchart illustrating the process of determining candidate recommendation menus in one embodiment;

[0055] Figure 4 This is a schematic diagram of the process of constructing a knowledge graph in one embodiment;

[0056] Figure 5 This is a flowchart illustrating the menu recommendation method in another embodiment;

[0057] Figure 6 This is a structural block diagram of a menu recommendation device in one embodiment;

[0058] Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0060] The menu recommendation method provided in this application can be applied to an environment where tellers are recommended service menus during business transactions. This method can be executed by a server or a terminal. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc.

[0061] In one exemplary embodiment, such as Figure 1 As shown, a menu recommendation method is provided. Taking the application of this method to a server as an example, the method includes the following steps:

[0062] S101, obtain the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business.

[0063] The business data for the current business refers to all information related to the specific business being handled by the target operator, including business type (such as corporate account opening, personal loan approval), business progress (such as account opening completed 80%), customer type (corporate / personal), business occurrence time (such as the last 3 days of the month), and unique identifier (Identifier, ID) of the associated business, covering the entire process data of cross-system business handling.

[0064] User data includes the identity data, permission data, and historical behavior data of the target operators; menu data includes the business attribute data of the menus and the relationships between menus.

[0065] The target operator is the person handling the business, such as a bank teller. The user data of the target operator refers to the basic information and permission information of the target operator handling the current business. Taking bank personnel as an example, user data includes static and dynamic data such as job code (e.g., teller, loan reviewer), permission level (level 1-5), years of service, branch type (community branch, flagship branch), and historical operation habits (operation sequence in the past 30 days).

[0066] The menu data of the business menu refers to the menu information of the bank's full range of front-end business systems (counter transactions, credit approval, customer management, financial accounting, etc.). After unified and standardized processing, it includes structured data such as menu unique identifier (menuId), business unified resource locator (URL), system to which it belongs, permission requirements, and business semantic tags (such as "personal customers - account opening - basic business").

[0067] For example, taking banking business as an example, business data can be extracted from the core system where the current business is located (such as the corporate business system); user data can be retrieved from the bank's unified employee information database, which includes real-time synchronized job and permission update data; menu data can be collected by calling the "menu full query API" of each system (main path) and directly connecting to the database to read the system menu data table (sys_menu) (alternative path).

[0068] S102, extract features from business data, user data, and menu data to obtain the target feature vector.

[0069] Among them, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator; job suitability is used to quantify the degree of matching between the target operator's job and the menu; scenario matching is used to quantify the degree of fit between the current business scenario and the menu; business relevance is used to quantify the strength of the connection between the menu and the current business; and operational habit fit is used to quantify the degree of matching between the menu and the target operator's historical operational habits.

[0070] For example, business data, user data, and menu data can be input into a target recommendation model for feature extraction. This target recommendation model can be a financial business target recommendation model (based on Meta AI 3 (Llama3)). For instance, business data, user data, and menu data can be converted into a Prompt format that the target recommendation model can parse. This Prompt data is then input into the target recommendation model, and its attention mechanism focuses on core features, outputting a target feature vector. This target feature vector includes weights for job suitability, scenario matching, business relevance, and operational habit fit.

[0071] For example, see Table 1, which provides an illustrative process for extracting features from business data, user data, and menu data to obtain a target feature vector.

[0072] Table 1:

[0073]

[0074] S103, based on the target feature vector, select the target recommendation menu from each business menu.

[0075] For example, based on the target feature vector, similar operators can be first screened, and then the target recommended menu can be determined according to the historical operation records of similar operators.

[0076] S104, Display the target recommended menu to the terminal where the target operator is located.

[0077] For example, a target recommendation menu can be displayed to the terminal where the target operator is located. Taking banking business as an example, the target recommendation menu can be displayed through the PC client, web portal, or mobile APP of the bank's unified employee channel system. For example, a "dynamic recommendation area" can be set up on the left side of the interface to display the target recommendation menu and natural language recommendation reasons (such as "Based on your current corporate account opening business and month-end scenario, we recommend signing up for online banking, which is a high-frequency and necessary operation after corporate clients open an account"), allowing employees to directly jump to the corresponding business page by clicking the menu (based on the Single Sign-On (SSO) login-free mechanism).

[0078] In some optional implementations, a unified identity authentication and entry function can be set up. For example, a unified visual style and operation logic can be implemented across PC clients, web portals, and mobile apps. Based on the Central Authentication Service (CAS) architecture, it supports "account password + dynamic token" and "fingerprint / facial biometric authentication." Upon successful authentication, a Ticket-Granting Ticket (TGT) (8 hours) and a Service Ticket (ST) (5 minutes) are generated, enabling login-free access across the entire system. Additionally, a unified "employee ID - system roles - permissions" mapping table can be implemented, synchronizing system-wide permissions within one minute of a job reassignment.

[0079] For example, suppose the target operator is a teller at a bank community branch (job code: GY001, access level: level 2), and the current business is "corporate customer account opening registration" (business data: customer type is corporate, business progress is 80%, time is the last 2 days of the month).

[0080] First, obtain the teller's user data (3 years of service, operation sequence in the past 30 days: customer inquiry → account opening → account activation → online banking signing), the current full process data of corporate account opening business, and the bank's full menu data (including standardized information and semantic tags of menus such as "online banking signing", "corporate settlement agreement" and "account reconciliation").

[0081] The target feature vector was extracted using the target recommendation model. Among them, the job suitability (the matching degree between the teller and the corporate business menu) was 0.85, the scenario matching degree (the fit between the corporate scenario and the menu at the end of the month) was 0.70, the business relevance (the correlation strength between account opening and online banking signing) was 0.95, and the operation habit fit degree (the frequency of the teller signing online banking after opening an account in the past) was 0.90.

[0082] Based on the target feature vector, target recommendation menus (such as "Online Banking Sign-up" and "Corporate Settlement Agreement") are selected and displayed in the dynamic recommendation area of ​​the web portal, with the recommendation reasons. Tellers can click "Online Banking Sign-up" to be redirected to the operation page without logging in.

[0083] In the above embodiments, the business data of the current business, the user data of the target operator processing the current business, and the menu data of each business menu associated with the current business are acquired. Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector. This target feature vector reflects the target operator's job suitability, scenario matching, business relevance, and operational habit fit. Based on the target feature vector, a target recommended menu is selected from each business menu. This target recommended menu is then displayed to the terminal where the target operator is located. This solution, in determining the target recommended menu, takes into account the target operator's job suitability, scenario matching, business relevance, and operational habit fit, making the determined target recommended menu more suitable for the current business, thereby improving the adaptability of the recommended menu.

[0084] In some alternative implementations, see [link to relevant documentation]. Figure 2 , Figure 2 A flowchart illustrating the process of selecting a target recommended menu from various business menus is provided, specifically including the following steps:

[0085] S201, based on the target feature vector and the candidate feature vectors of each candidate operator, select similar operators that are similar to the target operator from among the candidate operators.

[0086] Among them, each candidate operator refers to other operators besides the target operator. Taking bank employees as an example, each candidate operator is an employee within the bank other than the target operator. Their user data and historical operation records have been recorded and standardized by the system to form candidate feature vectors that can be used for similarity comparison.

[0087] For example, similar operators that are similar to the target operator can be selected from among the candidate operators based on the similarity between the target feature vector and the candidate feature vectors of each candidate operator.

[0088] In some optional implementations, for each candidate operator, the similarity between the target feature vector and the candidate feature vector of the candidate operator can be determined first. Here, similarity refers to the degree of similarity between the target feature vector and the candidate feature vector, and its value can range from 0 to 1. The closer the similarity is to 1, the more similar the two operators are in terms of their job roles, operating habits, and business-related preferences.

[0089] For example, similarity can be determined using the following formula:

[0090]

[0091] in, V1 represents the similarity between the target feature vector and the candidate feature vector of the candidate operator, where V1 is the target feature vector and V2 is the candidate feature vector of the candidate operator.

[0092] Then, the similarity scores are sorted from largest to smallest to obtain the first sorting order.

[0093] For example, a quicksort algorithm can be used to arrange all candidate operators in descending order of similarity to form the first sort order. For example, candidate operator A has a similarity of 0.92, candidate operator B has a similarity of 0.90, candidate operator C has a similarity of 0.88, and so on.

[0094] Furthermore, the candidate operators corresponding to the first preset number of similarities in the first sorting order can be regarded as similar operators to the target operator.

[0095] The first preset number can be set according to the actual situation, for example, it can be set to 30. The first 30 candidate operators with similarity to the target operator can be selected from the first sorted order.

[0096] S202, based on the historical operation records of similar operators, select candidate recommended menus from each business menu.

[0097] For example, based on the historical operation records of similar operators, a set of menus with high frequency and high business completion rate can be filtered out from the historical operation records, and then candidate recommended menus can be filtered out from the menu set.

[0098] S203, Based on the scenario weight of the candidate recommended menu in the corresponding business scenario of the business data, filter out the target recommended menu from the candidate recommended menu.

[0099] The scenario weight refers to the importance coefficient of the candidate recommended menu in the current business scenario, which can range from 1.0 to 2.0; it can be dynamically generated based on the knowledge graph to reflect the strength of the association between the menu and the current scenario.

[0100] For example, the ranking number of each candidate recommended menu can be determined based on the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, and a preset number of target recommended menus can be selected based on the ranking number.

[0101] In the above embodiments, the three-level screening mechanism of similar operator screening, candidate menu screening, and scenario weight screening not only draws on the mature operational experience of similar operators to ensure the practicality of the recommended menu, but also combines the particularity of the current business scenario to ensure the adaptability of the recommended menu, further improving the accuracy of the recommended menu, avoiding the recommendation of high-frequency but unsuitable menus, and meeting the continuous business operation needs of employees.

[0102] In some alternative implementations, see [link to relevant documentation]. Figure 3 , Figure 3 A flowchart illustrating the process of determining candidate recommendation menus is provided, which includes the following steps:

[0103] S301, determine the number of historical operations for each business menu based on the historical operation records of similar operators.

[0104] Among them, historical operation records refer to the full-link operation data of similar operators in the business system, including login information, menu click time, dwell time, business completion results, cross-system operation sequences, etc., which are stored in association through "operation session ID".

[0105] Historical operation count refers to the total number of times each business menu appears in the historical operation records of similar operators, including the number of valid clicks (the duration of the click is greater than or equal to 10 seconds) and the number of business completions, which is used to quantify the frequency of menu usage and practicality.

[0106] For example, operation logs from each operator can be stored in a Kafka+HBase storage architecture. Operation logs from similar operators can be retrieved from the Kafka+HBase architecture and preprocessed. Then, valid operation records are filtered from the preprocessed logs (removing accidental clicks and records with a dwell time of less than 10 seconds). The number of valid clicks and the number of completed transactions for each business menu are grouped by menuId to obtain the historical operation count. For instance, the sum of the valid clicks and completed transactions for each business menu can be used as the historical operation count, or a weighted average of the valid clicks and completed transactions for each business menu can be used to obtain the historical operation count.

[0107] S302, sort each business menu in descending order according to the historical number of operations of each business menu to obtain the second sorting order.

[0108] For example, a descending sorting algorithm can be used to arrange all business menus in descending order of the number of historical operations, forming a second sorting order.

[0109] S303, select the second-most preset number of business menus in the second sorting order as candidate recommended menus.

[0110] The second preset quantity can be set according to the actual situation, for example, it can be set to 12.

[0111] For example, a second preset number of business menus can be selected from the second sorting order as candidate recommended menus.

[0112] In some alternative implementations, the business menus in the second sorting order can be reordered, and candidate recommended menus can be selected based on the new sorting order.

[0113] For example, for each candidate recommended menu, the order of the candidate recommended menu in the second sorting order can be adjusted according to the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, so as to obtain a new sorting order.

[0114] For example, for each candidate recommended menu, the scenario weight of the candidate recommended menu in the current business scenario can be obtained from the knowledge graph (e.g., "Online Banking Sign-up" weight 1.8, "Account Reconciliation" weight 1.9). The new sorting number is calculated according to the formula "New sorting number = Original sorting number × (2 - Scenario weight)". The higher the weight, the smaller the new number and the higher the priority.

[0115] Then, based on the new arrangement number of each candidate recommended menu, the candidate recommended menus are sorted in descending order to obtain the third arrangement order.

[0116] For example, all candidate recommendation menus can be sorted in ascending order according to the new arrangement number to form a third arrangement order.

[0117] Furthermore, the first three preset number of candidate recommended menus in the third sorting order are taken as the target recommended menu.

[0118] The third preset quantity can be set according to actual needs; for example, it can be set to 5.

[0119] For example, the top 5 candidate recommended menus can be selected from the third sort order as the target recommended menu.

[0120] For example, assume the target operator is a teller (job code GY001). First, calculate the similarity between the target operator's feature vector and the candidate feature vectors of 1000 other tellers in the bank, and select the 30 tellers with the highest cosine similarity as similar operators. Analyze the historical operation records of these 30 tellers and find that the top 12 menus (click count + completion count) for menus such as "Online Banking Sign-up," "Corporate Settlement Agreement," "Account Reconciliation," and "Account Activation" are selected as candidate recommended menus. Next, based on the current "end-of-month corporate account opening" scenario, query the knowledge graph to find that "Online Banking Sign-up" has a scenario weight of 1.8, "Corporate Settlement Agreement" has a scenario weight of 1.7, and "Account Reconciliation" has a scenario weight of 1.9. Adjust the ranking of the candidate recommended menus according to these weights, and finally select the top 5 menus as the target recommended menus.

[0121] In the above embodiments, the frequency of menu usage is statistically analyzed based on the effective historical operation records of similar operators to ensure that the candidate recommended menus are high-frequency and practical menus that have been verified in practice, and can meet the operating habits and business needs of the target operators. This provides a high-quality candidate pool for the subsequent selection of target recommended menus and avoids recommending low-frequency and inefficient menus.

[0122] In some optional implementations, the scenario weight of each candidate recommendation menu item within the corresponding business scenario of the business data can be determined based on a knowledge graph, see [link to relevant documentation]. Figure 4 , Figure 4 A flowchart illustrating the process of constructing a knowledge graph is provided, which includes the following steps:

[0123] S401, Obtain sample data.

[0124] The sample data refers to the basic data required to build the knowledge graph, including full menu semantic tags (including three-level classifications of "business domain-functional group-operation item") and business operation logs, and may also include business manuals.

[0125] For example, taking banking operations as an example, a dual-path mechanism of "main path Application Programming Interface (API) collection + backup path cross-collection" can be used to collect full menu data from all banking systems. A three-level classification system of "business domain - function group - operation item" is established, converted into a unified JSON format, and synchronized to a "standardized menu database." Furthermore, business semantic tags are added to each menu (e.g., "account registration" is labeled "individual customer - account opening - basic business"), providing sample data for building a knowledge graph.

[0126] For example, business manuals (such as the "Corporate Business Operation Specifications") can be retrieved from the business management system, and full menu semantic tags can be obtained from the unified menu integration module. Cross-system business operation logs (with abnormal data removed after preprocessing) can be retrieved from the full-domain operation data recording module to form a sample dataset.

[0127] S402, Obtain data correlation from sample data.

[0128] Among them, data association refers to the logical relationship between entities extracted from sample data, including the adaptation relationship between the operator's position and the menu (such as "counter teller - high frequency operation - cash deposit and withdrawal"), and the association relationship between different menus in each business scenario (such as "account opening registration - necessary subsequent operation - online banking signing").

[0129] For example, sample data can be input into a lightweight financial business target recommendation model. By fine-tuning it with instructions (such as Prompt "parse the business logic relationship between 'account opening registration' and 'online banking signing'"), the model can learn business semantics, automatically extract the relationship between entities, and output "entity-relationship-entity" triples, such as <counter teller, high-frequency operation, cash deposit and withdrawal>, <account opening registration, necessary subsequent operation, online banking signing>.

[0130] S403, construct a knowledge graph based on data relationships.

[0131] Among them, knowledge graph refers to a structured semantic network with "menu-scenario-job" as the core entity and business relationship, job matching relationship and scenario dependency relationship as the core relationship. It is stored in a graph database (such as Neo4j) and used to support scenario weight calculation and business logic reasoning.

[0132] For example, the extracted triples can be imported into a graph database (Neo4j), and a graph structure configuration (such as entity attribute definition and relation type annotation) can be generated through the target recommendation model to automatically build a knowledge graph. The update frequency can be set, for example, setting the knowledge graph to be updated once a day based on new business data and feedback data.

[0133] In the above embodiments, a knowledge graph is constructed by parsing sample data using an AI-based target recommendation model, achieving accurate capture of business semantic relationships and solving the problem of lack of business semantic understanding in traditional recommendation schemes. The scenario weights determined based on this knowledge graph accurately reflect the compatibility between the menu and the current business scenario and job position, providing reliable semantic support for the recommendation process and ensuring that the recommended menu has business coherence and scenario adaptability.

[0134] In some alternative implementations, see [link to relevant documentation]. Figure 5 , Figure 5 A flowchart illustrating another menu recommendation method is provided, which includes the following steps:

[0135] S501: Obtain the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business.

[0136] S502 extracts features from business data, user data, and menu data to obtain the target feature vector.

[0137] S503: Based on the target feature vector and the candidate feature vectors of each candidate operator, select similar operators that are similar to the target operator from among the candidate operators.

[0138] S504, determine the number of historical operations for each business menu based on the historical operation records of similar operators.

[0139] S505: Sort each business menu in descending order according to the historical number of operations for each business menu to obtain the second sorting order.

[0140] S506, select the second-most preset number of business menus in the second sorting order as candidate recommended menus.

[0141] S507: Based on the scenario weight of the candidate recommended menu in the corresponding business scenario of the business data, filter out the target recommended menu from the candidate recommended menu.

[0142] The specific processes of S501 to S507 described above can be found in the description of the above method embodiments. Their implementation principles and technical effects are similar, and will not be repeated here.

[0143] Furthermore, the execution order of the above steps is merely illustrative and is not intended to limit the execution steps. The execution order of other steps is within the protection scope of the embodiments of this application.

[0144] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0145] Based on the same inventive concept, this application also provides a menu recommendation apparatus for implementing the menu recommendation method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more menu recommendation apparatus embodiments provided below can be found in the limitations of the menu recommendation method described above, and will not be repeated here.

[0146] In one exemplary embodiment, such as Figure 6 As shown, a menu recommendation device is provided, comprising:

[0147] The acquisition module 10 is used to acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business.

[0148] The extraction module 20 is used to extract features from business data, user data, and menu data to obtain target feature vectors; wherein, the target feature vectors are used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator;

[0149] The determination module 30 is used to select the target recommendation menu from each business menu based on the target feature vector;

[0150] The display module 40 is used to display the target recommended menu to the terminal where the target operator is located.

[0151] The aforementioned menu recommendation device acquires business data for the current business, user data of the target operator processing the current business, and menu data for each business menu associated with the current business. It then extracts features from the business data, user data, and menu data to obtain a target feature vector. This target feature vector reflects the target operator's job suitability, scenario matching, business relevance, and operational habit compatibility. Based on the target feature vector, it selects a target recommended menu from each business menu and displays it to the target operator's terminal. This solution, in determining the target recommended menu, considers the target operator's job suitability, scenario matching, business relevance, and operational habit compatibility, making the determined target recommended menu more suitable for the current business and thus improving the adaptability of the recommended menu.

[0152] In one embodiment, the determining module 30 specifically includes:

[0153] The first screening unit is used to screen out similar operators that are similar to the target operator from among the candidate operators based on the target feature vector and the candidate feature vectors of each candidate operator.

[0154] The selection unit is used to select candidate recommended menus from various business menus based on the historical operation records of similar operators;

[0155] The second filtering unit is used to filter out the target recommended menu from the candidate recommended menu based on the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data.

[0156] In one embodiment, the first screening unit is specifically used for:

[0157] For each candidate operator, determine the similarity between the target feature vector and the candidate feature vector of the candidate operator; sort the similarities in descending order to obtain the first sorting order; and take the candidate operators corresponding to the first preset number of similarities in the first sorting order as similar operators similar to the target operator.

[0158] In one embodiment, the selection unit is specifically used for:

[0159] Based on the historical operation records of similar operators, determine the historical operation count of each business menu; sort each business menu in descending order according to the historical operation count of each business menu to obtain a second sorting order; select the second-most preset number of business menus in the second sorting order as candidate recommended menus.

[0160] In one embodiment, the second filtering unit is specifically used for:

[0161] For each candidate recommended menu, the arrangement number of the candidate recommended menu in the second arrangement order is adjusted according to the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, and a new arrangement number is obtained; according to the new arrangement number of each candidate recommended menu, each candidate recommended menu is sorted in descending order to obtain the third arrangement order; the first three preset number of candidate recommended menus in the third arrangement order are taken as the target recommended menu.

[0162] In one embodiment, the device further includes a building module for:

[0163] Obtain sample data, which includes full menu semantic tags and business operation logs; extract data relationships from the sample data, including the relationships between the operator's job position and the menu, as well as the relationships between different menus in each business scenario; construct a knowledge graph based on the data relationships.

[0164] Each module in the aforementioned menu recommendation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0165] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores menu data and business data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a menu recommendation method.

[0166] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0167] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the menu recommendation method described in any of the above embodiments.

[0168] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the menu recommendation method described in any of the above embodiments.

[0169] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the menu recommendation method described in any of the above embodiments.

[0170] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0171] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0172] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0173] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A menu recommendation method, characterized in that, The method includes: Acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business; Feature extraction is performed on the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator; Based on the target feature vector, select the target recommendation menu from each business menu; The target recommended menu is displayed to the terminal where the target operator is located.

2. The method according to claim 1, characterized in that, The step of selecting a target recommendation menu from various business menus based on the target feature vector includes: Based on the target feature vector and the candidate feature vectors of each candidate operator, similar operators that are similar to the target operator are selected from the candidate operators. Based on the historical operation records of the similar operators, select candidate recommended menus from each business menu; Based on the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, the target recommended menu is selected from the candidate recommended menu.

3. The method according to claim 2, characterized in that, The step of selecting similar operators from among the candidate operators that are similar to the target operator based on the target feature vector and the candidate feature vectors of each candidate operator includes: For each candidate operator, determine the similarity between the target feature vector and the candidate feature vector of the candidate operator; The similarity scores are sorted from highest to lowest to obtain the first sorting order; The candidate operators corresponding to the first preset number of similarities in the first sorting order are regarded as similar operators to the target operator.

4. The method according to claim 2, characterized in that, The step of selecting candidate recommended menus from various business menus based on the historical operation records of the similar operators includes: Based on the historical operation records of the similar operators, determine the number of historical operations for each business menu; Sort each business menu in descending order according to the historical number of operations for each business menu to obtain the second sorting order; The first second preset number of business menus in the second sorting order are selected as candidate recommended menus.

5. The method according to claim 4, characterized in that, The step of filtering the target recommended menu from the candidate recommended menu according to the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data includes: For each candidate recommended menu, the arrangement number of the candidate recommended menu in the second arrangement order is adjusted according to the scenario weight of the candidate recommended menu in the business scenario corresponding to the business data, so as to obtain a new arrangement number; Based on the new arrangement number of each candidate recommended menu, sort the candidate recommended menus in descending order to obtain the third arrangement order; The first three preset number of candidate recommended menus in the third sorting order are taken as the target recommended menu.

6. The method according to claim 2, characterized in that, The scenario weight of each candidate recommendation menu in the business scenario corresponding to the business data is determined based on a knowledge graph, which is constructed in the following way: Obtain sample data; wherein, the sample data includes full menu semantic tags and business operation logs; Data relationships are obtained from the sample data; wherein, the data relationships include the relationship between the operator's position and the menu, and the relationship between different menus in each business scenario; The knowledge graph is constructed based on the data relationships.

7. A menu recommendation device, characterized in that, The device includes: The acquisition module is used to acquire the business data of the current business, the user data of the target operator of the current business, and the menu data of each business menu associated with the current business. The extraction module is used to extract features from the business data, user data, and menu data to obtain a target feature vector; wherein, the target feature vector is used to reflect the job suitability, scenario matching, business relevance, and operational habit fit of the target operator; The determination module is used to select the target recommendation menu from each business menu based on the target feature vector; The display module is used to display the target recommended menu to the terminal where the target operator is located.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images