Product recommendation method and device, equipment and medium

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a product recommendation method and device, equipment and a medium, and aims to accurately capture behavior patterns and preference characteristics of customers by deeply analyzing product keyword search records, multi-dimensional attribute tags and recommendation interaction data of the customers. On the basis, real-time matching degree detection is carried out on a product recommendation data set to be pushed, and it is ensured that the risk income characteristics of financial products or the professional suitability of medical health services are highly matched with customer requirements. And when the recommendation deviation is identified, a dynamic optimization mechanism for the financial / medical health product recommendation content is automatically triggered, and a personalized recommendation scheme accurately matched with the customer portrait is generated. Through pre-positioned intelligent quality inspection, it is ensured that the recommended content accurately conforms to the financial management demand or health management appeal of the customer, so that the conversion rate of a financial scene and the service accuracy of a medical scene are remarkably improved, the purchase willingness of the user is effectively enhanced, and the order conversion effect is optimized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a product recommendation method, device, equipment and medium. Background Art

[0002] In the financial and healthcare sectors, a growing number of institutions are using social media marketing models to serve their customers. Once marketers establish contact with customers, they simultaneously deliver financial product information and health management advice through multiple, multi-channel content engagements, facilitating insurance purchases or financial services contracts. Currently, in various instant messaging social media marketing channels, one or more marketers typically establish contact with customers. These marketers manually orchestrate customer service scripts, utilize the company's CRM / SCRM system, or leverage the client's own content tools, and regularly send these scripts to customers. However, these standardized scripts struggle to meet the needs and preferences of individual customers, resulting in poor product marketing effectiveness. Summary of the Invention

[0003] The present invention provides a product recommendation method, device, computer equipment and medium to solve the technical problem that unified speech content cannot meet the needs and preferences of each customer, resulting in poor product marketing effect.

[0004] In a first aspect, a product recommendation method is provided, comprising:

[0005] Obtaining at least one first keyword and product recommendation data of a product, as well as at least one attribute tag and number of product recommendations of a target customer;

[0006] Determining whether the product recommendation data matches the target customer based on at least one attribute tag, at least one first keyword, and the number of product recommendations;

[0007] If the product recommendation data does not match the target customer, generating target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword, and sending the target product recommendation data to the target customer;

[0008] If the product recommendation data matches the target customer, the product recommendation data will be sent to the target customer.

[0009] In a second aspect, a product recommendation device is provided, comprising:

[0010] An acquisition module, configured to acquire at least one first keyword of a product and product recommendation data, as well as at least one attribute tag of a target customer and the number of times the product is recommended;

[0011] a judgment module, configured to judge whether the product recommendation data matches the target customer based on at least one attribute tag, at least one first keyword, and the number of product recommendations;

[0012] a generating module for generating target product recommendation data for the product based on at least one attribute tag, the number of product recommendations, and at least one first keyword if there is no match between the product recommendation data and the target customer; and a sending module for sending the target product recommendation data to the target customer;

[0013] The sending module is further configured to send the product recommendation data to the target customer if the product recommendation data matches the target customer.

[0014] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned product recommendation method when executing the computer program.

[0015] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned product recommendation method are implemented.

[0016] In the solution implemented by the above-mentioned product recommendation method, device, computer equipment and storage medium, the customer's behavior and preferences are obtained by parsing the obtained product keywords, customer attribute tags and the current product recommendation round, and then the product recommendation data to be pushed is tested to determine whether the product recommendation data meets the customer's needs and preferences. Based on the test results, it is determined whether the text content of the product needs to be refined or expanded to generate new target product recommendation data. This implements the pre-control of product recommendation data and achieves accurate recommendations through quality inspection of product recommendation data, so that the product recommendation content received by the customer meets the customer's individual needs and interests, increases the user's willingness to purchase, and thus improves the order conversion effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] Figure 1 This is a schematic diagram of an application environment of a product recommendation method according to an embodiment of the present invention;

[0019] Figure 2 This is a flowchart of a product recommendation method according to an embodiment of the present invention;

[0020] Figure 3 yes Figure 2 A schematic flow chart of a specific implementation of step S10;

[0021] Figure 4 yes Figure 2 A schematic flow chart of a specific implementation of step S30;

[0022] Figure 5 is a structural diagram of a product recommendation device in one embodiment of the present invention;

[0023] Figure 6 is a structural diagram of a computer device in one embodiment of the present invention;

[0024] Figure 7 FIG. 2 is another structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] The product recommendation method provided by the embodiment of the present invention can be applied in Figure 1 In an application environment, the client communicates with the server through a network. The server receives the product recommendation data sent by the client, and based on the product keywords of the product to be recommended, the customer type of the target customer to be recommended, and the number of product recommendations this time, detects the product recommendation data to determine whether the product recommendation data matches the target customer. If the product recommendation data does not match the target customer, it is necessary to regenerate the target product recommendation data of the product to be recommended, and send the newly generated target product recommendation data to the target customer terminal; if the product recommendation data matches the target customer, the product recommendation data is directly sent to the target client. In the above manner, the product recommendation data received by the target customer can meet the customer needs and the needs of the current recommendation round, improve customer satisfaction, and thus increase the customer's purchase probability. Among them, the client can be but is not limited to various personal computers, laptops, smart phones, tablet computers and portable wearable devices. The server can be implemented with an independent server or a server cluster composed of multiple servers. The present invention is described in detail below through specific embodiments.

[0027] See also Figure 2 As shown, Figure 2A flowchart of a product recommendation method provided by an embodiment of the present invention includes the following steps:

[0028] S10: Acquire at least one first keyword and product recommendation data of a product, as well as at least one attribute label and number of product recommendations of a target customer.

[0029] The product recommendation method provided by the present invention can be applied to the intersection of finance and healthcare (such as health insurance, medical consumer finance, health management services, etc.). Specifically, the product can be a company's product to be promoted. Each product is provided with one or more product keywords, i.e., the first keyword. During the product marketing process, after the financial advisor / health financial planner establishes contact with the customer, the pre-written product recommendation data is sent to the potential customer (i.e., the target customer) through the server, so that the target customer can understand and purchase the product through the product recommendation data.

[0030] For example, in the financial and healthcare product promotion scenario, the products may be health insurance products, medical consumer finance products, pension finance products, and the like.

[0031] In actual application scenarios, facing fierce market competition, continuous and accurate reach is the key to maintaining product competitiveness. The current mainstream marketing channel is the digital private domain community built through instant messaging tools such as WeChat for Business, which forms an efficient customer communication network. In this marketing system, one or more professional health financial advisors or medical financial planners establish contact with target customers and systematically provide customers with the medical financial product information they need for decision-making through a multi-dimensional content push strategy (covering text analysis, infographics, explanation videos, electronic documents, smart cards, and exclusive medical financial applets, etc.). Product recommendation data can be compiled by marketers through manual arrangement, using the company's CRM / SCRM system, or using the client's own content tools. Marketers regularly compile customer service scripts and package the scripts to form product recommendation data, which are then sent to customers from time to time. However, from the customer's perspective, when customers repeatedly receive product recommendations that do not match their health status and financial needs (such as pushing elderly medical care financial plans to young people, or recommending products with limited coverage to patients with chronic diseases), it is very easy to cause information fatigue. They may be unable to bear the harassment and cut off contact or block marketers, resulting in deterioration of customer relationships and even interruption of communication channels. Therefore, in order to improve customer satisfaction, this application proposes to extract the characteristics of medical financial products before sending product recommendation data, obtain product keywords of the products to be recommended, combine customer multi-dimensional portrait analysis (including health indicators, financial capabilities, medical consumption habits, etc.) and interactive behavior tracking (historical recommendation records, content interaction depth, etc.), combine customer attribute labels with the number of product recommendations as a detection standard, test product recommendation data, and determine whether the product recommendation data matches the target customers, ensure that each contact meets the customer's real needs and financial affordability, and optimize customer experience while improving conversion rate.

[0032] Through the above method, the product keywords, customer attribute tags and recommendation times of the product are obtained as the detection basis of the product recommendation data to ensure the accuracy and relevance of the product recommendation data.

[0033] In one embodiment of the present application, Figure 3 As shown, a specific solution for obtaining product-related data and customer-related data is provided. In S10, at least one first keyword and product recommendation data of a product, as well as at least one attribute label and the number of product recommendations of a target customer are obtained. The solution specifically includes the following steps S11-S14:

[0034] S11: In response to a product recommendation request, obtaining the customer name, product name, and product recommendation data corresponding to the product name included in the product recommendation request.

[0035] In this step, the marketer sends the target customer's name, the product name to be recommended, and the product recommendation data corresponding to the product name to the server, and initiates a product recommendation request. After receiving the product recommendation request, the server obtains the product recommendation data, product name, and customer name contained in the product recommendation request.

[0036] S12: Based on the product name, determine at least one first keyword of the product.

[0037] In this step, the core keywords of each product are constructed in advance according to the main functions and characteristics of each product and stored in a preset database. According to the product name of the currently recommended product, one or more first keywords of the product are retrieved from the preset database.

[0038] For example, in finance and healthcare scenarios, product keywords can accurately reflect the core medical attributes and financial characteristics of the product. For example, keywords for banking products may include "interest," "credit limit," and "repayment method," while keywords for insurance products may include "disease coverage," "premium," and "coverage."

[0039] S13: Based on the customer name, determine the customer profile of the target customer and the number of product recommendations to the target customer.

[0040] In this step, based on the customer name of the potential customer (i.e., target customer), the pre-stored customer profile of the target customer is retrieved. The customer profile includes the target customer's basic information (including age, gender, region, occupation, income level, etc.), consumption data (including purchase history, consumption frequency, consumption amount, etc.), behavior data (including customer click behavior and browsing history), feedback data (such as likes, dislikes, neutrality, opinions, etc.). At the same time, if this recommendation is not the first, the number of times the product has been recommended to the target customer is also required. For example, this is the third time that Product A has been recommended to the target customer.

[0041] For example, the target customer groups are diversified and can be divided into two main categories: individual customer groups (such as ordinary working class, high net worth individuals, young customers, elderly customers, etc.) and corporate customer groups (such as small and medium-sized enterprises, large enterprises, medical institutions, etc.).

[0042] In practical applications, historical recommendation records from customer interactions can be stored through methods like conversation and voice archiving. This data can then be combined with basic customer information, consumption data, behavioral data, and feedback data to create a comprehensive customer profile. When marketers initiate product recommendations, they can access this comprehensive profile based on the customer's identity information, helping to identify their needs and preferences, significantly improving the accuracy and conversion efficiency of healthcare and financial product recommendations.

[0043] S14: Based on the customer portrait, determine at least one attribute label of the target customer.

[0044] In this step, different types of customers have varying needs and preferences for product recommendations. For example, young professionals are often time-constrained and busy. This group of customers prefers brief, concise recommendations to quickly obtain key information and facilitate quick decision-making. Older, middle-aged / retired customers, on the other hand, typically have more life experience and consumer knowledge. They prefer detailed, comprehensive analysis and comparison, and prefer more detailed recommendations to make more reliable purchasing decisions. To improve the relevance and effectiveness of product recommendations, we need to identify at least one attribute tag for the target customer based on their profile in order to understand their needs and preferences.

[0045] In practical application scenarios, in the healthcare and finance sector, to gain a deeper understanding of customer needs and preferences, a refined customer tagging system is constructed. Multi-dimensional attribute tags are pre-defined, and the corresponding relationship between each attribute tag and the customer profile is determined. Attribute tags include the following dimensions: basic characteristic tags, occupational characteristic tags, behavioral preference tags, and healthcare-specific tags. For example, a 35-year-old married female customer would be tagged as "Middle-aged + Married + Professional + Pregnancy and Childbirth Health Needs"; a 55-year-old male customer would be tagged as "Middle-aged + Chronic Disease Management + Simple Preferences." By deeply integrating structured tags with customer consumption behavior and health management data, a dynamic tagging system is formed, providing intelligent decision support for the precise recommendation of healthcare and financial products.

[0046] S20: Based on at least one attribute tag, at least one first keyword and the number of product recommendations, determine whether the product recommendation data matches the target customer. If so, proceed to step S40; if not, proceed to step S30.

[0047] In this step, during the product recommendation process, it is usually necessary to send multiple rounds of recommendation data so that customers can gradually understand the product and generate a willingness to buy. However, as the number of recommendations increases, customers may become impatient. At this time, if the content of the recommendation data is too cumbersome or cannot allow customers to understand the product at first glance, customers may block marketers. Therefore, in order to improve customer experience, when testing product recommendation data, based on the analysis of customer needs based on customer portraits, it is also necessary to test the product recommendation data in combination with the current number of recommendations to take into account customer emotions. Specifically, the target customer's attribute label, product recommendation number and product keywords are used as the detection basis to test the product recommendation data to determine whether the product recommendation data matches the target customer. Only when the product recommendation data matches the target customer can the product recommendation data be sent to the target customer to ensure that the product recommendation data received by the target customer meets the customer's needs.

[0048] Through the above methods, we can understand the preferences and needs of target customers through attribute labels and recommendation times, and use them as a basis for judgment to test the content of product recommendation data, thereby providing more personalized recommendations and improving the acceptance and satisfaction of target customers.

[0049] In one embodiment of the present application, a specific product recommendation data detection solution is provided. In S20, based on at least one attribute tag, at least one first keyword, and the number of product recommendations, it is determined whether the product recommendation data matches the target customer. The solution specifically includes the following steps S21-S25:

[0050] S21: Based on a preset mapping relationship, determine a target data standard corresponding to at least one attribute label and the number of product recommendations, wherein the preset mapping relationship includes data standards for product recommendation data corresponding to different attribute labels and product recommendation numbers.

[0051] In this step, data standards corresponding to different attribute tags and product recommendation times are pre-set. Based on the correspondence between these different attribute tags, product recommendation times, and data standards, a pre-set matching mapping relationship is constructed. During the product recommendation process, the target data standard is retrieved from the pre-set mapping relationship based on at least one attribute tag of the target customer and the current number of recommendations.

[0052] For example, in the precise marketing scenarios of medical, health and financial products, differentiated content strategies are implemented according to the characteristics of different customer groups. For example, the third recommendation of Product A to young professionals (25-35 years old) needs to adopt an efficient and concise communication method, with the content controlled within 5 sentences, and the first three sentences must contain core keywords (such as "million-level medical insurance" and "0 deductible for outpatient treatment"). Example of sales talk: "Mr. Zhang, the [million-level medical insurance] Product A you are concerned about has been upgraded again (keyword 1)! Now you can enjoy the [0 deductible for outpatient treatment] privilege (keyword 2). Quick insurance in 30 seconds, monthly payment is only 85 yuan (keyword 3). Click to view your exclusive rate." The third recommendation of Product A to retired aunts (over 55 years old) needs to adopt a detailed comparative communication method, with the content controlled within 5-10 sentences, focusing on highlighting the three major competitive advantages. Sample sales pitch: "Auntie Wang, the [Elderly Cancer Prevention Insurance] Product A you inquired about previously has three major advantages over other products: 1) Direct payment to Grade A hospitals without advance payment (competitive products require reimbursement afterwards); 2) Reimbursement of Traditional Chinese Medicine treatment is included (competitive products do not cover this); 3) Renewal is possible at age 70 (competitive products have a maximum age of 65). Last time you said you were worried about the cost of chemotherapy, and our product can reimburse 90% of the cost of targeted drugs. Do you need me to help you compare the coverage in detail?"

[0053] Optionally, based on a large amount of historical recommendation data, as well as customer behavior data and feedback data, data standards for different attribute labels and corresponding data standards for different product recommendation times can be constructed. For example, in the early stages of product recommendation, when customers are more interested in new products, the number of sentences in the recommendation text can be set to be larger, such as the number of sentences in the first recommendation can be set to no more than 20. As the number of recommendations increases, the number of sentences in the recommendation text can be gradually reduced to avoid customers becoming bored with the increased reading volume. For example, in four recommendation data, the number of sentences in the text can be set to no more than five.

[0054] S22: Obtain text content included in the product recommendation data.

[0055] In this step, the product recommendation data can be of various types, such as text, pictures, videos, files, links, etc. By extracting the text content in the product recommendation data, the text content is tested to determine whether the text content meets the customer's needs.

[0056] In actual application scenarios, in the marketing of healthcare / financial products, intelligent content extraction technology is used to achieve structured processing of multimodal recommendation data, including graphic and text material processing (such as using optical character recognition technology to extract core data from healthcare posters), video material parsing (such as using FFmpeg tools combined with deep learning recognition technology to extract video subtitles for medical insurance product introductions), special document processing (such as using PDF parsing tools to extract the terms and conditions of electronic insurance policies), etc. S23: Based on at least one first keyword and text content, determine whether the product recommendation data meets the target data standard. If so, proceed to step S24; if not, proceed to step S25.

[0057] In this step, the text content is parsed to obtain multiple sentences in the text content. Based on the content of each sentence and the first keyword of the product, a test is performed to determine whether the text content in the product recommendation data meets the target data standard. For example, for the second recommendation of young professionals, the corresponding target data standard is that the product recommendation data contains no more than five text sentences, and at least one product keyword must appear in the first three sentences. At this time, the text content is divided into sentences, each sentence and the number of sentences are obtained, and at least one first keyword is matched with each sentence in turn. It is detected whether the first three sentences contain at least one first keyword and whether the number of sentences is less than or equal to five.

[0058] S24: Determine the match between product recommendation data and target customers;

[0059] S25: Determine whether there is a mismatch between product recommendation data and target customers.

[0060] For steps S24-S25, when it is detected that the text content in the product recommendation data meets the target data standard, it is determined that the product recommendation data matches the target customer, that is, the product recommendation data meets the target customer's needs in the current recommendation round; when it is detected that the text content in the product recommendation data does not meet the target data standard, it is determined that the product recommendation data does not match the target customer, that is, the product recommendation data does not meet the target customer's needs in the current recommendation round.

[0061] S30: Generate target product recommendation data of the product based on at least one attribute tag, the number of product recommendations and at least one first keyword, and send the target product recommendation data to the target customer.

[0062] In this step, if a mismatch between the product recommendation data and the target customer is detected, the target data criteria of at least one attribute tag and the number of product recommendations are used as the writing criteria. In combination with at least one first keyword of the product, the product text content is refined or expanded to form target text content that meets the customer's needs and preferences, and new product recommendation data is generated. Finally, the re-edited target product recommendation data is sent to the target customer's terminal.

[0063] Through the above method, combined with attribute labels and recommendation times, we can more accurately understand the needs and preferences of target customers, and then use the target data standards of attribute labels and recommendation times as the writing standard, and rewrite the product text content through product keywords, including refining or expanding the product text content, so that the final target recommendation data can be more in line with customer expectations, thereby effectively improving customers' purchasing intentions.

[0064] In one embodiment of the present application, Figure 4 As shown, a specific target product recommendation data generation solution is provided. In S30, target product recommendation data of a product is generated based on at least one attribute tag, the number of product recommendations, and at least one first keyword. The solution specifically includes the following steps S31-S33:

[0065] S31: Generate a target writing template based on the target data standard and at least one first keyword.

[0066] S32: Import the target writing template into the pre-trained product recommendation content generation model to obtain the target text content.

[0067] S33: Generate target product recommendation data based on the target text content.

[0068] In steps S31-S33, during the generation of target product recommendation data, a large language model is pre-trained to generate a trained product recommendation content generation model. This model then generates a target writing template based on the target data criteria corresponding to the attribute label and the number of product recommendations, combined with at least one first keyword of the product. Subsequently, the target writing template is imported into the trained product recommendation content generation model to generate target text content that meets customer needs and preferences. Finally, target product recommendation data of the same type is generated based on the target text content, according to the data type of the product recommendation data.

[0069] In actual application scenarios, the Transformer's large language model is trained based on the company's service introduction and product data of all the company's products, and the trained model is used as the product recommendation content generation model. The edited target writing template is imported into the product recommendation content generation model to output the target text content. For example, if the customer is a young professional, it is detected that there is too much text content in the product recommendation data, the content is cumbersome and does not highlight keywords. At this time, it is necessary to use the trained large model to refine the marketing content, automatically screen and reach the content of young professionals; if the customer is a retiree, it is detected that there is less text content in the product recommendation data. At this time, it is necessary to use the large model to expand the marketing content. In addition, the data format of product recommendation data matches the customer's business attributes, and types include text, images, videos, files, link cards, mini-programs, etc. In order to support customers' clear operations and rapid conversion and increase customer interest, after generating new target text content, the target text content is converted into target product recommendation data with a unified interactive style that matches the customer's business attributes according to the customer's business attribute format. For example, if the product recommendation data type is page card type, the target text content is automatically divided into title, introduction, introduction details, action button, and next steps. Subsequently, through the codeless programmatic CMS, the divided target text content is converted into standardized page cards with sharing titles, descriptions, and style images that are consistent with the customer's business attributes, supporting rapid conversion of customer operations. Furthermore, in the process of multiple rounds of recommendations, in order to avoid duplication of recommended content, when it is not the first recommendation, the current recommended data needs to be tested based on the previous recommendation data to avoid content duplication. At the same time, when generating the target text content, it is necessary to exclude the recommended keywords from this recommendation to ensure the freshness of the recommended content.

[0070] For example, the goal writing template is as follows:

[0071] Title: Product 1

[0072] Product keywords: Keyword A, Keyword B, Keyword C;

[0073] Writing requirements: Write a product recommendation text containing five sentences, and the first three sentences must contain at least one product keyword.

[0074] In one embodiment of the present application, Figure 4 As shown, a specific modification reminder solution is provided. After S30, that is, after generating target product recommendation data of the product based on at least one attribute tag, the number of product recommendations and at least one first keyword, the following steps S34-S35 are also included:

[0075] S34: Generate modification prompt information based on the target product recommendation data.

[0076] S35: Send the modification prompt information to the first business person.

[0077] In steps S34-S35, the server regenerates the target product recommendation data using the pre-trained large language model and sends the new target product recommendation data to the marketer responsible for product marketing (i.e., the first business person). Because the target product recommendation data is generated based on the analysis of different customer attributes, it is more closely aligned with customer needs. Therefore, by analyzing the target product recommendation data, marketers can promptly identify deficiencies in the initial product recommendation data, adjust subsequent recommendation strategies, and improve the accuracy and effectiveness of product recommendations.

[0078] In actual application scenarios, the conclusion can be modified based on the differences between the target product recommendation data and the product recommendation data. This modified conclusion can be sent to marketers, allowing them to quickly and accurately identify any issues with the initial recommendation data. For example, if the initial product recommendation data contains too much content, while the target product recommendation data is streamlined, a modified conclusion can be generated: excessive ineffective communication. Marketers can use this modified conclusion to adjust the recommended data content in subsequent recommendations.

[0079] In one embodiment of the present application, a specific product recommendation data detection solution is provided. Before S20, that is, before determining whether the product recommendation data matches the target customer based on at least one attribute tag, at least one first keyword, and the number of product recommendations, the following steps S110-S120 are also included:

[0080] S110: Based on multiple second keywords, detect the product recommendation data, (determine whether the product recommendation data contains at least one second keyword, if so, proceed to step S120, if not, proceed to step S20;

[0081] S120: Determine that there is a violation in the product recommendation data and stop product recommendation.

[0082] In steps S110-S120, the multiple second keywords are pre-set sensitive words (e.g., pornography, terrorism, violence, and other prohibited words). During the product recommendation process, the product recommendation data is first preliminarily tested based on the pre-set sensitive words. If any of the second keywords are detected in the product recommendation data, it indicates that the product recommendation data contains inappropriate words or sensitive content. To maintain the brand image and demonstrate the company's professionalism, the product recommendation is suspended to prevent further spread of violations. If no second keywords are found in the product recommendation data, the product recommendation data is tested for match with the target customer.

[0083] In actual application scenarios, text verification tools (such as sensitive word filtering software and content review tools) are used to audit compliance and private domain risk control. Compliance audits include: using a sensitive word library specifically for the financial industry (including banned words such as "guaranteed returns" and "rigid redemption"); loading medical and health-specific terminology filters (to identify inappropriate expressions such as "cure rate" and "efficacy promise"); and real-time connection to the latest regulatory keyword list of the State Administration of Financial Supervision and Administration. Private domain risk control audits include: direct contact information (such as mobile phone numbers, WeChat accounts, and QR codes); variant expressions (such as "Weiwo" and "VX" homophonic deformations); and subtle guidance (such as "Private chat for more" and "Add consultation"). This intercepts data recommending illegal products, ensuring the compliance of medical and health financial marketing content while effectively preventing regulatory risks brought about by the disorderly expansion of private domain traffic.

[0084] Through the above method, before recommending products to customers, the product recommendation data is checked for violations to avoid the possibility of product recommendation content violating regulations and harassing customers, thereby ensuring the brand image.

[0085] In one embodiment of the present application, after S120, that is, after determining that the product recommendation data has violated regulations and stopping product recommendations, the following steps S130-S140 are also included:

[0086] S130: Generate violation prompt information based on at least one second keyword included in the product recommendation data;

[0087] S140: Send the violation prompt information to the second business personnel.

[0088] In steps S130-S140, based on the violation-sensitive words contained in the product recommendation data, violation warning information is generated. This generated violation warning information is sent to relevant personnel, namely the second business personnel, who then investigate the product recommendation data and the responsible marketing personnel based on the violation warning information, trace the source of the violation data, reduce risks, and protect the brand image.

[0089] S40: Send product recommendation data to target customers.

[0090] In this step, when a match is detected between the product recommendation data and the target customer, it means that the product recommendation data meets the requirements of the customer and this round, and the product recommendation data is sent to the terminal of the target customer.

[0091] As can be seen, in the above solution, by parsing the obtained product keywords, customer attribute tags, and the current product recommendation round, the customer's behavior and preferences are obtained, and then the product recommendation data to be pushed is tested to determine whether the product recommendation data meets the customer's needs and preferences. Based on the test results, it is then determined whether the product text content needs to be refined or expanded to generate new target product recommendation data. This achieves pre-emptive control of product recommendation data. Through quality inspection of product recommendation data, accurate recommendations are achieved, ensuring that the product recommendations received by customers meet their individual needs and interests, increasing users' willingness to purchase and thus improving order conversion effects.

[0092] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0093] In one embodiment, a product recommendation device is provided, which corresponds to the product recommendation method in the above embodiment. Figure 5 As shown, the product recommendation device includes: an acquisition module 101, a judgment module 102, a generation module 103 and a sending module 104. The functional modules are described in detail as follows:

[0094] An acquisition module 101 is configured to acquire at least one first keyword of a product and product recommendation data, as well as at least one attribute tag of a target customer and the number of times the product is recommended;

[0095] A judgment module 102 is configured to judge whether the product recommendation data matches the target customer based on at least one attribute tag, at least one first keyword, and the number of product recommendations;

[0096] The generating module 103 is configured to generate target product recommendation data for the product based on at least one attribute tag, the number of product recommendations, and at least one first keyword if there is no match between the product recommendation data and the target customer; the sending module 104 is configured to send the target product recommendation data to the target customer;

[0097] The sending module 104 is further configured to send the product recommendation data to the target customer if the product recommendation data matches the target customer.

[0098] In one embodiment, the acquisition module 101 specifically includes:

[0099] A first acquiring unit is configured to, in response to a product recommendation request, acquire a customer name, a product name, and product recommendation data corresponding to the product name included in the product recommendation request;

[0100] a first determining unit, configured to determine at least one first keyword of the product based on the product name;

[0101] The second determining unit is used to determine the customer profile of the target customer and the number of product recommendations to the target customer based on the customer name;

[0102] The third determining unit is configured to determine at least one attribute tag of a target customer based on the customer portrait.

[0103] In one embodiment, the determination module 102 specifically includes:

[0104] A fourth determining unit is configured to determine a target data standard corresponding to at least one attribute label and the number of product recommendation times based on a preset mapping relationship, wherein the preset mapping relationship includes data standards for product recommendation data corresponding to different attribute labels and product recommendation times;

[0105] A second acquiring unit, configured to acquire text content contained in the product recommendation data;

[0106] a judgment unit, configured to judge whether the product recommendation data meets a target data standard based on at least one first keyword and text content;

[0107] a fifth determining unit, configured to determine a match between the product recommendation data and the target customer if the product recommendation data meets the target data criteria;

[0108] The sixth determining unit is configured to determine that the product recommendation data does not match the target customer if the product recommendation data does not meet the target data standard.

[0109] In one embodiment, the generation module 103 specifically includes:

[0110] A first generating unit, configured to generate a target writing template based on a target data standard and at least one first keyword;

[0111] The second generation unit is used to import the target writing template into the pre-trained product recommendation content generation model to obtain the target text content;

[0112] The third generating unit is used to generate target product recommendation data based on the target text content.

[0113] In one embodiment, the determination module 102 further includes:

[0114] a fourth generating unit, configured to generate modification prompt information based on the target product recommendation data;

[0115] The first sending unit is used to send the modification prompt information to the first business personnel.

[0116] In one embodiment, the apparatus further comprises:

[0117] A detection module, configured to detect product recommendation data based on a plurality of second keywords;

[0118] In one embodiment, the generating module 103 is further configured to, if the product recommendation data contains at least one second keyword, determine that the product recommendation data has violated regulations and stop recommending the product.

[0119] In one embodiment, the generating module 103 is further configured to generate violation prompt information based on at least one second keyword included in the product recommendation data;

[0120] The sending module 104 is further configured to send the violation prompt information to the second business personnel.

[0121] The present invention provides a product recommendation device that parses acquired product keywords, customer attribute tags, and the current product recommendation round to obtain customer behavior and preferences, and then detects the product recommendation data to be pushed to determine whether the product recommendation data meets customer needs and preferences. Based on the test results, it is determined whether the text content of the product needs to be refined or expanded to generate new target product recommendation data. This implements pre-emptive control of product recommendation data, and through quality inspection of product recommendation data, achieves accurate recommendations, ensuring that the product recommendation content received by customers meets their individual needs and interests, increasing users' willingness to purchase and thus improving order conversion results.

[0122] For the specific definition of the product recommendation device, please refer to the definition of the product recommendation method above and will not be repeated here. Each module in the above-mentioned product recommendation device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0123] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the server side of a product recommendation method.

[0124] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the client side of a product recommendation method.

[0125] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0126] Obtaining at least one first keyword and product recommendation data of a product, as well as at least one attribute tag and number of product recommendations of a target customer;

[0127] Determining whether the product recommendation data matches the target customer based on at least one attribute tag, at least one first keyword, and the number of product recommendations;

[0128] If the product recommendation data does not match the target customer, generating target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword, and sending the target product recommendation data to the target customer;

[0129] If the product recommendation data matches the target customer, the product recommendation data will be sent to the target customer.

[0130] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0131] Obtaining at least one first keyword and product recommendation data of a product, as well as at least one attribute tag and number of product recommendations of a target customer;

[0132] Determining whether the product recommendation data matches the target customer based on at least one attribute tag, at least one first keyword, and the number of product recommendations;

[0133] If the product recommendation data does not match the target customer, generating target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword, and sending the target product recommendation data to the target customer;

[0134] If the product recommendation data matches the target customer, the product recommendation data will be sent to the target customer.

[0135] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0136] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0137] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0138] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A product recommendation method, characterized in that: include: Obtain at least one first keyword and product recommendation data of a product, and obtain at least one attribute tag and number of product recommendations of a target customer; Based on the at least one attribute tag, the at least one first keyword, and the number of product recommendations, determining whether the product recommendation data matches the target customer; If the product recommendation data does not match the target customer, generating target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword, and sending the target product recommendation data to the target customer; If the product recommendation data matches the target customer, the product recommendation data is sent to the target customer.

2. The method according to claim 1, characterized in that The step of obtaining at least one first keyword of a product, product recommendation data, at least one attribute tag of a target customer, and the number of product recommendations specifically includes: In response to a product recommendation request, obtaining a customer name, a product name, and the product recommendation data corresponding to the product name included in the product recommendation request; Determining the at least one first keyword of the product based on the product name; Based on the customer name, determine the customer profile of the target customer and the number of product recommendations to the target customer; Based on the customer portrait, the at least one attribute label of the target customer is determined.

3. The method according to claim 1, characterized in that The step of determining whether the product recommendation data matches the target customer based on the at least one attribute tag, the at least one first keyword, and the number of product recommendations specifically includes: Determining a target data standard corresponding to the at least one attribute tag and the number of product recommendations based on a preset mapping relationship, wherein the preset mapping relationship includes data standards for product recommendation data corresponding to different attribute tags and product recommendation numbers; Obtaining text content contained in the product recommendation data; determining, based on the at least one first keyword and the text content, whether the product recommendation data meets the target data standard; If the product recommendation data meets the target data standard, determining that the product recommendation data matches the target customer; If the product recommendation data does not meet the target data standard, it is determined that the product recommendation data does not match the target customer.

4. The method according to claim 3, characterized in that The step of generating target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword specifically includes: generating a target writing template based on the target data standard and the at least one first keyword; Importing the target writing template into a pre-trained product recommendation content generation model to obtain target text content; Based on the target text content, the target product recommendation data is generated.

5. The method according to claim 4, characterized in that After generating target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword, the method further includes: Generate modification prompt information based on the target product recommendation data; The modification prompt information is sent to the first business personnel.

6. The method according to any one of claims 1 to 5, characterized in that Before determining whether the product recommendation data matches the target customer based on the at least one attribute tag, the at least one first keyword, and the number of product recommendations, the method further includes: Based on multiple second keywords, test product recommendation data; If the product recommendation data contains at least one second keyword, it is determined that the product recommendation data violates regulations and the product recommendation is stopped.

7. The method according to claim 6, characterized in that After determining that the product recommendation data has violated regulations and stopping product recommendations, the method further includes: generating violation prompt information based on the at least one second keyword included in the product recommendation data; The violation prompt information is sent to the second business personnel.

8. A product recommendation device, characterized in that: include: An acquisition module, configured to acquire at least one first keyword of a product and product recommendation data, as well as at least one attribute tag of a target customer and the number of times the product is recommended; a judgment module, configured to judge whether the product recommendation data matches the target customer based on the at least one attribute tag, the at least one first keyword, and the number of product recommendations; a generating module configured to generate target product recommendation data for the product based on the at least one attribute tag, the number of product recommendations, and the at least one first keyword if there is no match between the product recommendation data and the target customer; and a sending module configured to send the target product recommendation data to the target customer; The sending module is further configured to send the product recommendation data to the target customer if the product recommendation data matches the target customer.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the product recommendation method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the product recommendation method according to any one of claims 1 to 7 are implemented.

Images