Method and system for personalized product intelligent recommendation based on user portrait analysis

Abstract

The present application relates to the technical field of user portrait analysis, and particularly relates to a personalized product intelligent recommendation method and system based on user portrait analysis, which comprises the following steps: collecting desensitization data of a recipient customer, performing portrait comprehensive analysis, and generating a customer exclusive portrait; obtaining product procurement information submitted by an enterprise, performing customized demand matching, and outputting product customization demand information; identifying a product resource library, identifying single product attributes of the product resource library, and constructing a product knowledge graph; performing range convergence retrieval and multi-dimensional matching on the product knowledge graph based on the product customization demand information, and constructing a product recommendation set; performing product combination intelligent recommendation on the product recommendation set based on the customer exclusive portrait, and outputting a product combination recommendation scheme. The present application realizes accurate recommendation and efficient product selection of customized products of an enterprise, improves product recommendation accuracy and procurement efficiency, reduces product selection cost of the enterprise, and enhances customer satisfaction.

Description

Technical Field

[0001] This invention relates to the field of user profile analysis, and in particular to a personalized product intelligent recommendation method and system based on user profile analysis. Background Technology

[0002] Traditional enterprise product procurement typically relies on manual experience for product screening and solution development. Procurement personnel need to make manual judgments based on the company's purpose, budget range, and recipient information, which leads to problems such as insufficient needs analysis, low product matching accuracy, and insufficient procurement efficiency. Recipients of corporate donations exhibit significant individual differences. Different customers vary considerably in age, occupation, interests, corporate culture, and consumption tendencies. Traditional product recommendation methods based on uniform standards are insufficient to achieve personalized matching for different recipients.

[0003] Existing product recommendation methods for some clients primarily rely on sales volume, historical purchase records, or simple keyword matching for product filtering. This only achieves basic product retrieval and fails to comprehensively consider the relationships between enterprise purchasing scenarios, recipient customer characteristics, enterprise brand positioning, and the client's product customization capabilities. Current product resource data is typically stored as discrete product information, lacking the ability to perform correlation analysis on multi-dimensional attributes such as product category, material, customization process, price range, applicable industries, and usage scenarios. This results in low product resource utilization efficiency and makes it difficult to support complex enterprise customization needs. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a personalized product intelligent recommendation method and system based on user profile analysis, thereby resolving at least one of the aforementioned technical issues.

[0005] To achieve the above objectives, this invention provides a personalized product intelligent recommendation method based on user profile analysis, comprising the following steps: Step S1: Collect anonymized data from recipient customers, perform comprehensive profiling analysis, and generate a personalized customer profile; Step S2: Obtain the product procurement information submitted by the enterprise, perform customized demand matching, and output product customization demand information; Step S3: Identify the product resource library, identify individual product attributes in the product resource library, and construct a product knowledge graph; Step S4: Based on product customization requirements, perform range convergence retrieval and multi-dimensional matching on the product knowledge graph to construct a product recommendation set; Step S5: Based on the customer's unique profile, perform intelligent product combination recommendations on the product recommendation set and output a product combination recommendation scheme.

[0006] This specification provides a personalized product intelligent recommendation system based on user profile analysis, used to execute the personalized product intelligent recommendation method based on user profile analysis as described above, including: The customer profiling module is used to collect anonymized data from gift recipients, perform comprehensive profiling analysis, and generate personalized customer profiles. The requirements analysis module is used to obtain product procurement information submitted by enterprises, perform customized requirements matching, and output product customization requirements information. The product library construction module is used to identify the product resource library, identify individual product attributes in the product resource library, and construct a product knowledge graph. The matching module is used to perform range convergence retrieval and multi-dimensional matching on the product knowledge graph based on product customization requirements information, and to build a product recommendation set; The intelligent recommendation module is used to make intelligent product combination recommendations based on the customer's unique profile and output the product combination recommendation scheme.

[0007] The specific beneficial effects of this invention are as follows: By extracting identity features, associating with corporate attributes, and tagging and parsing anonymized customer data, a multi-dimensional customer profile is created. This profile integrates customer identity, interests, job attributes, and corporate culture characteristics, improving the completeness of recipient descriptions and providing an accurate data foundation for matching gift needs. Structured analysis of corporate gift product procurement information extracts key demand features such as gift usage scenarios, product types, procurement quantities, and budgets. This information is then combined with the customer profile for matching, enabling a quantitative expression of corporate procurement needs and improving the accuracy of customized gift product identification. Finally, attribute analysis of individual gift products in the gift product resource library, along with tagging and associating basic, customized, and commercial attributes, constructs a gift product knowledge graph. This enables the organization of multi-dimensional attributes of gift products, improving gift information retrieval and feature matching capabilities.

[0008] By performing a scope-converging search on the gift product knowledge graph based on customized gift product demand information, and combining multi-dimensional evaluation with budget matching, preference matching, scenario applicability, and availability index, the system achieves accurate screening and ranking of candidate gift products, improving the matching degree between recommendation results and actual needs. Furthermore, by combining and analyzing the gift product recommendation set based on customer-specific profiles, and optimizing the combination of different gift types, customer characteristics, and gift attributes, the system achieves associative configuration between multiple gift products, improving the completeness and adaptability of the recommendation scheme and enhancing the effectiveness of the intelligent recommendation process. Attached Figure Description

[0009] Figure 1This is a flowchart illustrating the steps of a personalized product intelligent recommendation method based on user profile analysis according to the present invention. Figure 2 A detailed flowchart illustrating the implementation steps for collecting anonymized data from recipient customers, performing comprehensive profiling and analysis, and generating personalized customer profiles. Figure 3 To obtain product procurement information submitted by enterprises, match customized needs, and output a detailed flowchart of the implementation steps for product customization needs; Figure 4 A diagram illustrating the process of collecting anonymized customer data; Figure 5 A schematic diagram illustrating the distribution of basic attribute information for collecting anonymized customer data; Figure 6 An illustration of a personalized client profile. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. It should also be understood that, as used in this specification and the appended claims, the term "and / or" refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0012] Furthermore, in the description of this invention and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0013] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of the invention include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0014] It should be understood that the sequence number of each step in the following embodiments does not imply 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.

[0015] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0016] In the embodiments of the present invention, see Figure 1 This is a flowchart illustrating the steps of a personalized product intelligent recommendation method based on user profile analysis according to the present invention. In this embodiment, the steps of the personalized product intelligent recommendation method based on user profile analysis include: Step S1: Collect anonymized data from recipient customers, perform comprehensive profiling analysis, and generate a personalized customer profile; Step S2: Obtain the product procurement information submitted by the enterprise, perform customized demand matching, and output product customization demand information; Step S3: Identify the product resource library, identify individual product attributes in the product resource library, and construct a product knowledge graph; Step S4: Based on product customization requirements, perform range convergence retrieval and multi-dimensional matching on the product knowledge graph to construct a product recommendation set; Step S5: Based on the customer's unique profile, perform intelligent product combination recommendations on the product recommendation set and output a product combination recommendation scheme.

[0017] In a specific embodiment, taking the procurement of business cooperation gifts by an internet technology company as an example, this invention illustrates the intelligent recommendation method for customized corporate gifts based on user profile analysis. This internet technology company is located in East China, has 680 employees, operates in internet technology services, positions itself as a technology innovation company, and its corporate culture includes "innovation, openness, environmental protection, collaboration, and quality," with blue as its primary brand color. The company plans to conduct an annual partner appreciation event, procuring 300 business cooperation gifts, with a budget of 300-500 yuan per gift. The gifts must support customized company logos, be packaged in gift boxes, and be delivered within 7 days.

[0018] Anonymized data was collected from recipient customers, sourced from enterprise CRM systems and customer management platforms. Customer names, phone numbers, and email addresses were anonymized, retaining only business analysis information such as age, gender, interests, company position, job level, and department. In this example, there were 300 recipients, including 198 customers aged 26-35 (66%), 72 customers aged 36-45 (24%), and 30 customers over 46 (10%). There were 210 male customers (70%) and 90 female customers (30%). Positions included 45 company managers, 96 technical managers, 84 purchasing managers, and 75 marketing managers. Interest tags showed 58% in digital technology, 27% in business and office work, 10% in culture and reading, and 5% in sports and health. Subsequently, the region, business type, brand positioning, and corporate culture tags of the customer's company were retrieved, and a correlation analysis was performed between customer identity characteristics and company attributes. When customer interest tags and corporate culture tags share common characteristics, corresponding matching tags are generated. For example, if a customer has a digital technology interest tag and the corporate culture includes innovation attributes, a technology innovation matching tag is generated; if the customer's position is a company leader or purchasing manager, a business decision-making attribute tag is generated. Statistics show that 247 customers (82.3% of all customers) share three or more common characteristics in their interests and corporate culture tags. Therefore, identity attribute tags, interest preference tags, consumption tendency tags, position attribute tags, and corporate culture matching tags are generated to form a personalized customer profile.

[0019] The system obtains gift procurement information submitted by enterprises, performs structured analysis on the purchase orders, and extracts key elements such as gift usage scenarios, gift types, purchase quantities, and purchase budgets. In this example, the gift usage scenario is an annual partner appreciation event, the purchase quantity is 300 pieces, the purchase budget is 300-500 yuan per piece, and the requirements include support for logo laser engraving, gift box packaging, and a supply cycle of no more than 7 days. Based on the characteristics of business cooperation scenarios, the system extracts demand features such as business attributes, high-quality packaging, strong brand display capabilities, high technological feel, and high long-term use value, and matches them with the customer's exclusive profile. When the proportion of business and office interest tags exceeds 20%, technology interest tags exceed 50%, and the proportion of enterprise leaders, technical leaders, and procurement leaders exceeds 50%, the procurement demand is determined to be a business cooperation type. In this embodiment, 27% of the recipients were business and office-related interest tags, and 58% were digital technology interest tags. A total of 225 people, including enterprise leaders, technical leaders, and purchasing leaders, accounted for 75% of the total number of recipients, which met the criteria for determining business cooperation. Therefore, the customized gift requirements were output, including prioritizing digital products and business office supplies, with a price range of 300 to 500 yuan, and support for laser engraving of logos, gift box packaging, and delivery within 7 days.

[0020] Furthermore, the gift resource library is identified, and the attributes of individual gifts are analyzed. In this embodiment, the gift resource library stores a total of 1,200 gifts. Each gift includes basic attributes, customization attributes, and commercial attributes. The basic attributes include gift category, material, color, size, and brand information. The customization attributes include logo customization method, packaging method, craftsmanship type, and customization period. The commercial attributes include price range, inventory quantity, supply period, and applicable industries. For example, Gift A is a wireless charging business gift box, priced between 300 and 500 yuan, supports laser engraving of the logo, comes in a gift box, has a supply cycle of 4 to 7 days, and has an inventory of 860 pieces. It is suitable for the internet, manufacturing, and financial industries. Gift B is a smart thermos cup gift box, priced between 300 and 500 yuan, supports logo customization, comes in a gift box, has a supply cycle of 4 to 7 days, and has an inventory of 420 pieces. Gift C is a business backpack, priced between 100 and 300 yuan. Although it has an inventory of 1260 pieces, it was excluded during the price range convergence stage because its price range did not meet the 300-500 yuan procurement budget requirement. Subsequently, the attributes of the gifts, such as category, material, packaging, craftsmanship, price, inventory, and industry, were tagged, and the relationships between the tags were established to form a gift knowledge graph.

[0021] Based on customized gift demand information, a convergent search was performed on the gift knowledge graph, selecting 436 gifts suitable for business cooperation scenarios from 1200 gifts; 215 gifts were retained based on the price range of 300-500 yuan; 128 gifts remained based on logo laser engraving and gift box packaging requirements; 102 gifts were retained based on a supply cycle of no more than 7 days; and finally, 86 candidate gifts were obtained based on the inventory quantity meeting the purchase requirement of 300 pieces. Budget matching degree, recipient preference matching degree, scenario applicability, and availability index were calculated for these 86 candidate gifts. Budget matching degree evaluates the consistency between the gift price range and the purchase budget, with 100% for identical price ranges and 80% for adjacent ranges; recipient preference matching degree is calculated based on the ratio of the number of shared interest tags to the number of matching tags; scenario applicability is determined based on the matching relationship between the gift's applicable scenario and the business cooperation scenario; and the availability index is determined based on inventory quantity and supply cycle. Furthermore, the four indicators are calculated according to the following weights: budget matching degree 30%, recipient preference matching degree 30%, scenario applicability 25%, and availability index 15%, to obtain a single gift matching degree index.

[0022] For example, Gift A has a budget match rate of 100%, 12 identical customer interest preference tags and gift tags, and a total of 13 matching tags. Therefore, the recipient's preference match rate is 12 ÷ 13 × 100% = 92.3%. It also perfectly matches the business cooperation scenario, with a scenario applicability rate of 100%. With 860 units in stock and a supply cycle of 4-7 days, the availability index is 95%. Therefore, Gift A's match rate index is: 100 × 30% + 92.3 × 30% + 100 × 25% + 95 × 15% = 96.94%. Gift B has a budget match rate of 100%, a preference match rate of 84.6%, a scenario applicability rate of 95%, and an availability index of 88%, resulting in a match rate of 91.63%. Based on the match rate index, a gift recommendation set is formed, with the wireless charging business gift box ranking first, the smart thermos cup gift box ranking second, and the multifunctional business office set ranking third.

[0023] The gift recommendation set is intelligently combined based on the customer's personalized profile. First, the gift type is categorized according to the customer profile and the purchase scenario. When the purchase scenario is partner appreciation, customer visit, or business communication, and the proportion of company leaders, purchasing managers, and technical managers exceeds 50%, it is determined to be a business cooperation type. When the purchase target is mainly internal employees, it is determined to be an employee welfare type. When the purchase purpose is brand promotion or exhibition publicity, it is determined to be an event promotion type. When the purchase target is a long-term cooperative customer with a high maintenance attribute, it is determined to be a customer maintenance type. In this example, the purchase scenario is partner appreciation, and the proportion of company leaders, technical managers, and purchasing managers is 75%, therefore the gift type is determined to be a business cooperation type. Based on this type, three gift combination schemes are formed: wireless charging business gift box + branded greeting card; wireless charging business gift box + smart thermos cup + corporate customized notebook; business office set + Bluetooth speaker. The company's purchasing personnel selected the second scheme, completing the purchase of 300 sets, with an actual purchase amount of 126,000 yuan and an average purchase price of 420 yuan per item. After gift delivery, a total of 286 customer reviews were collected, of which 247 were very satisfied, 31 were satisfied, 8 were average, and there were no dissatisfaction reviews, resulting in a satisfaction rating of 4.8. Within three months, 120 sets of the same gift combination were purchased again, with no returns and only two items requiring packaging replacement. The platform calculated the following: 100% purchase completion rate, 100% budget compliance rate, 96 points of satisfaction score, 62 points of repeat purchase score, and 95 points of after-sales service score. A recommendation effectiveness index was then calculated using a weighted average of 30% for purchase completion rate, 15% for budget compliance rate, 30% for satisfaction score, 15% for repeat purchase rate, and 10% for after-sales service: 100×30%+100×15%+96×30%+62×15%+95×10%=92.6 points. Finally, the customer-specific profile, gift combination recommendation plan, purchase results, and recommendation effectiveness index of 92.6 points were used to generate an independent recommendation sample and stored in a cloud database, providing data support for recommendations of similar gift types.

[0024] In this embodiment, see Figure 2 For a detailed flowchart illustrating the steps involved in collecting anonymized data from recipient customers, performing comprehensive profiling analysis, and generating personalized customer profiles, please refer to [link / reference]. Figures 4 to 6 , Figure 4 A diagram illustrating the process of collecting anonymized customer data; Figure 5 A schematic diagram illustrating the distribution of basic attribute information for collecting anonymized customer data; Figure 6 This is a diagram illustrating the creation of a personalized customer profile. The specific steps involved in collecting anonymized data from recipient customers, performing comprehensive profile analysis, and generating a personalized customer profile are as follows: Collect anonymized data from recipient customers; analyze the anonymized data for identity features, extract customer age, gender, interests, company position, job level, and department information, and generate customer identity features; Obtain basic attribute information of the customer's company, including company location, business type, corporate culture tags, brand color tags, and brand positioning; Feature association analysis is performed on the customer identity characteristics and the basic attribute information to generate multiple customer-specific tags; the customer-specific tags include identity attribute tags, interest preference tags, consumption tendency tags, job attribute tags, and corporate culture adaptation tags; Based on the customer-specific tags, a comprehensive analysis of the profile is performed to generate a customer-specific profile.

[0025] In this embodiment, the enterprise product management platform obtains recipient customer information from the enterprise CRM system, human resource management system, and customer relationship database. Before data collection, the SHA-256 desensitization algorithm is used to irreversibly desensitize sensitive fields such as name, contact number, email address, and ID number, retaining only non-sensitive attributes such as age range, gender, job title, department, and interest tags required for business analysis. Data is cleaned according to unified field specifications, removing records with a null value rate exceeding 20%, and a field mapping method is used to unify the data format of different business systems. For example, job titles are uniformly mapped to standard categories such as management, technical, marketing, and administrative levels; age is divided into four age ranges: 18-25, 26-35, 36-45, and 46 and above; and interests are mapped to primary interest tags such as digital technology, business office, sports and health, culture and reading, and home life. For information from multiple data sources on the same customer, the data is merged using the customer's unique identifier, and the consistency of each field is calculated. For missing fields, data retention rules are determined based on historical customer data integrity statistics: when the proportion of null values ​​in a field is between 10% and 30%, it is supplemented and merged using other data sources for the same customer; when the proportion of null values ​​exceeds 30%, the data is not used as a basis for profile analysis. In this embodiment, 20% is selected as the null value judgment threshold, which is based on the enterprise's historical customer data statistics. This ensures a sufficient number of valid samples while avoiding the impact of low-quality data on the profile results. This threshold can be adjusted within the range of 10% to 30% depending on the data integrity of different enterprises.

[0026] The platform accesses the enterprise basic information database and brand management database to obtain basic enterprise attribute information and completes data association based on a unified enterprise code. Enterprises are geographically categorized into seven regions: East China, South China, North China, Central China, Southwest China, Northwest China, and Northeast China. Business types are classified into industry categories such as manufacturing, internet, finance, education, healthcare, retail, and services. Enterprise culture is extracted from official websites, company profiles, and brand promotional materials using high-frequency keywords. A combination of keyword statistics and semantic clustering is used to summarize enterprise culture into cultural tags such as innovation, stability, openness, environmental protection, quality, and responsibility. The top five tags with the highest weights are retained for each enterprise. Brand color tags are extracted based on the enterprise's VI standards, including primary and secondary colors such as blue for a technological image, green for an environmentally friendly image, and red for a vibrant image. These color characteristics are combined with brand positioning analysis to determine whether the enterprise belongs to a high-end business, young and fashionable, technologically innovative, or mass-market consumer brand positioning category. After completing the extraction of all attributes, all basic attributes are coded to form a unified enterprise feature dataset.

[0027] Customer identity characteristics are mapped to basic enterprise attributes using enterprise ID numbers. The correlation between customer job positions and enterprise brand positioning is then analyzed; for example, management positions correspond to high-end business needs, technical positions to technological innovation needs, and marketing positions to brand promotion needs. Next, common attributes between customer interest categories and enterprise culture tags are analyzed. A high fit is indicated when three or more common characteristics exist between interest categories and enterprise culture tags; a medium fit when two common characteristics exist; and a low fit when fewer than two common characteristics exist. Customer consumption tendencies are analyzed by combining customer age range, job level, and enterprise business type. Management positions correspond to a preference for high-value products, technical positions to a preference for technological experience products, and administrative positions to a preference for practical office products. Based on the above analysis results, identity attribute tags, interest preference tags, consumption tendency tags, job attribute tags, and enterprise culture fit tags are generated, with at least five tags for each customer. These multi-dimensional tags complement each other, more comprehensively reflecting customer needs and improving the accuracy of subsequent product matching analysis.

[0028] Customer-specific tags are categorized and statistically analyzed, and comprehensively organized according to five dimensions: identity attributes, interests and preferences, consumption tendencies, job attributes, and corporate culture fit. The frequency and correlation of each tag are statistically analyzed. When multiple tags exist for the same dimension, they are prioritized according to their frequency of occurrence. Consistency verification is performed by combining customer's historical product browsing history, product purchase history, and product review history. When a tag corresponds to historical behavior in a ratio of 80% or higher, it is designated as a stable tag; when the ratio is below 80%, it is marked as a dynamic tag for updating in subsequent recommendation processes. Stable and dynamic tags are integrated to form a customer-specific profile, which is stored in a profile information database according to customer ID.

[0029] In this embodiment, see Figure 3 The specific steps for obtaining product procurement information submitted by enterprises, matching customized needs, and outputting product customization requirement information are as follows: Obtain product procurement information submitted by enterprises; perform structured parsing on the product procurement information to extract key elements; the key elements include product usage scenarios, product types, procurement quantities, and procurement budgets; Based on the aforementioned key elements, a gift-giving scenario demand analysis is conducted to generate multiple gift-giving scenario demand features. Based on the characteristics of multiple gift-giving scenarios, we perform customized demand matching for customer profiles and output customized product demand information.

[0030] In this embodiment, information such as product usage scenarios, product categories, purchase quantities, purchase budgets, delivery times, and customization requirements are extracted and converted according to a unified field format. Product usage scenarios are categorized into six standard scenarios: customer visits, business cooperation, employee benefits, meetings and events, brand promotion, and holiday gifts. Product types are categorized into six categories: office supplies, digital products, home furnishings, cultural and creative products, food gift boxes, and commemorative collectibles. Purchase quantities are classified into four ranges: 1-100 pieces, 101-500 pieces, 501-1000 pieces, and over 1000 pieces. Purchase budgets are categorized into four price ranges: under 100 yuan, 100-300 yuan, 300-500 yuan, and over 500 yuan. After field extraction, a completeness check is performed on missing fields. If the missing rate of key fields exceeds 20%, the user is prompted to re-enter the data, and the remaining orders proceed to the next processing stage.

[0031] Analyze the correlation between product usage scenarios and procurement budgets. For example, customer visits and business cooperation scenarios prioritize mid-to-high budget ranges, employee benefits scenarios prioritize mid-to-low budget ranges, and conferences, events, and brand promotion scenarios prioritize large-volume procurement needs. Analyze the distribution of product types across different usage scenarios. When a product category accounts for more than 60% in a specified scenario, it is identified as a priority product type; a proportion of 40% to 60% indicates it is an optional product type; and less than 40% indicates it is a supplementary product type. Simultaneously, analyze product configuration methods based on procurement quantity. For procurement quantities below 100 units, personalized customization is prioritized; for procurement quantities between 101 and 500 units, a combination of standardization and customization is used; and for procurement quantities exceeding 500 units, bulk customization is prioritized. This results in multiple gift-giving scenario demand characteristics encompassing budget requirements, product type requirements, procurement scale requirements, and application scenario requirements.

[0032] This analysis correlates the demand characteristics of multiple gift-giving scenarios with customer-specific profiles. It retrieves identity attribute tags, interest preference tags, consumption tendency tags, job attribute tags, and corporate culture fit tags from the customer-specific profiles and matches them item by item with the gift-giving scenario demand characteristics. When interest preference tags match product type requirements, the corresponding product is marked as a priority recommendation; when job attributes match the gift-giving scenario, the recommendation priority of the corresponding product is increased; when corporate culture fit tags match product design style, the product is marked as a brand-fit product. Recommended products are price-filtered based on the purchase budget range, retaining only product information within the budget range; then, product supply capacity is checked based on the purchase quantity, and products whose inventory cannot meet the purchase quantity requirements are eliminated. The output includes product customization demand information, which includes recommended product categories, suggested price ranges, recommended customization processes, recommended packaging methods, and recommended purchase quantities.

[0033] In this embodiment, the specific steps for identifying the product resource library, performing individual product attribute identification on the product resource library, and constructing a product knowledge graph are as follows: The product resource library is identified, and individual product attributes are identified within the product resource library. The basic attributes, customized attributes, and commercial attributes of each product are extracted. The basic attributes include product category, material, color, size, and brand information. The customized attributes include logo customization method, packaging method, process type, and customization cycle. The commercial attributes include price range, inventory quantity, supply cycle, and applicable industries.

[0034] The basic attributes, customized attributes, and commercial attributes are tagged to generate multiple product tags; Calculate the price of a single product; associate the price with multiple product tags to construct a product knowledge graph.

[0035] In this embodiment, product information is read from the product resource library through the product management platform, and attributes are identified one by one according to the unique product number. Basic product attributes are extracted, including product category, material, color, size, and brand information. Product categories are uniformly classified into six categories: office supplies, digital products, home furnishings, cultural and creative products, food gift boxes, and commemorative collectibles. Materials are divided into seven categories: metal, wood, glass, ceramic, plastic, fabric, and composite materials. Colors are uniformly converted to standard color codes according to the company's standard color library. Sizes are uniformly converted to millimeters for storage. Brand information is identified using a unified brand code. Customization attributes are extracted, including logo customization method, packaging method, process type, and customization period. Logo customization methods are divided into five categories: laser engraving, screen printing, UV printing, embroidery, and embossing. Packaging methods are divided into three categories: ordinary packaging, gift box packaging, and environmentally friendly packaging. Process types are divided into categories such as spraying, electroplating, hot stamping, and engraving. Customization periods are classified into four ranges: within 3 days, 4-7 days, 8-15 days, and more than 15 days. Finally, commercial attributes are extracted, including price range, inventory quantity, supply cycle, and applicable industries, to establish complete attribute information for each product.

[0036] Product attributes are tagged according to unified rules. Category tags are generated based on product type, such as office supplies, digital products, and home furnishings; material tags are generated based on material, such as metal, wood, glass, and environmentally friendly materials; brand primary color tags and secondary color tags are generated based on color; laser engraving, screen printing, UV printing, and embroidery tags are generated based on logo customization method; standard packaging, gift box packaging, and environmentally friendly packaging tags are generated based on packaging method; and hot stamping, engraving, and electroplating tags are generated based on process type. For commercial attributes, prices are categorized into four price tags according to the aforementioned unified standards: below 100 yuan, 100-300 yuan, 300-500 yuan, and above 500 yuan. Inventory quantities are categorized into four inventory tags: below 100 units, 100-500 units, 501-1000 units, and above 1000 units. Supply cycles are categorized into four supply cycle tags: within 3 days, 4-7 days, 8-15 days, and above 15 days. Applicable industries are categorized into manufacturing, internet, finance, education, healthcare, retail, and service industries. Each product generates no fewer than 10 tags based on its actual attributes, and these tags are bound according to the product's unique number.

[0037] The system reads product procurement costs, processing costs, packaging costs, and logistics costs to calculate the comprehensive price of a single product. This price is then categorized into four price ranges according to the aforementioned unified standard: below 100 yuan, 100-300 yuan, 300-500 yuan, and above 500 yuan, generating corresponding price tags. Using the product's unique ID as the primary key, the system establishes a correspondence between price tags and category tags, material tags, brand tags, packaging tags, process tags, inventory tags, supply cycle tags, and applicable industry tags. For example, if a product belongs to the digital product category, uses metal materials, supports laser engraving, uses gift box packaging, is priced between 300-500 yuan, and is applicable to the internet and manufacturing industries, then all of the above tags will be associated. When multiple products have the same category tag or the same price tag, a relationship is established between the products. When products have the same applicable industry and the same customized process, industry and process relationships are further added. This forms a product knowledge graph with the product as the core, price as the connecting node, and various tags interconnected, and is stored according to the product ID.

[0038] In this embodiment, the specific steps for constructing a product recommendation set by performing range convergence retrieval and multi-dimensional matching on the product knowledge graph based on product customization demand information are as follows: Based on product customization requirements, the product knowledge graph is used for range convergence retrieval and multi-dimensional matching to generate a candidate product set; Based on the candidate product set, calculate the budget matching degree, recipient preference matching degree, scenario applicability and availability index matching degree, and generate a single product matching degree index; Intelligent sorting is performed based on individual product matching metrics to construct a product recommendation set.

[0039] In this embodiment, product customization requirements are read, including product category, product usage scenario, procurement budget, procurement quantity, recommended packaging method, recommended customization process, and applicable industry. These requirements are then used as product search criteria to progressively filter the product knowledge graph. First, the scope is narrowed down based on product category, retaining only products with the same category. Further filtering is performed based on procurement budget, retaining only products within the corresponding budget ranges: below 100 yuan, 100-300 yuan, 300-500 yuan, or above 500 yuan. A second filtering is performed based on recommended packaging method, logo customization method, and process type, eliminating products that do not meet the customization requirements. After completing the above search, inventory information is checked based on procurement quantity, retaining only products with sufficient inventory to meet the procurement requirements. Inventory quantity is categorized into four ranges: below 100 units, 100-500 units, 501-1000 units, and above 1000 units. Finally, matching is performed based on applicable industry and product usage scenario, retaining only products that meet both industry and scenario requirements, forming a candidate product set.

[0040] Each product in the candidate product set is analyzed individually for matching. The consistency between the product's price range and the company's purchasing budget range is statistically analyzed. When both belong to the same price range, the budget matching status is marked as a complete match; adjacent price ranges are marked as a partial match; otherwise, it is marked as a mismatch. Interest preference tags, job attribute tags, and corporate culture fit tags from the customer's personalized profile are read and statistically analyzed against the product tags. The more common tags, the higher the degree of matching between the recipient's preferences. The correspondence between the product and six scenarios—customer visits, business cooperation, employee benefits, meetings and events, brand promotion, and holiday gifts—is analyzed based on product usage scenario tags. When the product is applicable to the current gifting scenario, the scenario applicability status is determined to be highly applicable; when it is only partially applicable, it is determined to be moderately applicable. Finally, the product inventory quantity and supply cycle were statistically analyzed. The inventory quantity was classified into four ranges: less than 100 units, 100-500 units, 501-1000 units, and more than 1000 units. The supply cycle was classified into four ranges: less than 3 days, 4-7 days, 8-15 days, and more than 15 days. The availability index matching degree was determined based on the inventory adequacy and the timeliness of supply. The above results were combined to form a matching degree index for individual products.

[0041] The matching degree index of individual products corresponding to the candidate product set is sorted. An initial sort is performed based on budget matching, placing products with a perfect budget match first. A second sort is then performed based on the recipient's preference matching, prioritizing products with more common tags. For products with the same budget and preference matching, further sorting is based on the applicable scenario, prioritizing products suitable for the current gifting scenario. If multiple products still meet the same criteria, further sorting is performed based on availability index, prioritizing products with an inventory of 501-1000 units or more and a supply cycle of 3 days or less or 4-7 days to ensure timely procurement and customization. After all sorting is completed, the results are output according to recommendation priority to form a product recommendation set, simultaneously retaining recommendation information for each product, including category information, price range, packaging method, logo customization method, process type, inventory status, and supply cycle.

[0042] In this embodiment, the specific steps for constructing a product recommendation set by performing range convergence retrieval and multi-dimensional matching on the product knowledge graph based on product customization demand information are as follows: Based on product customization requirements, the product knowledge graph is used for range convergence retrieval and multi-dimensional matching to generate a candidate product set; Based on the candidate product set, calculate the budget matching degree, recipient preference matching degree, scenario applicability and availability index matching degree, and generate a single product matching degree index; Intelligent sorting is performed based on individual product matching metrics to construct a product recommendation set.

[0043] In this embodiment, products are filtered based on product category, such as office supplies, digital products, home furnishings, cultural and creative products, food gift boxes, and commemorative collectibles, retaining only products within the same category. Price filtering is performed based on the procurement budget, retaining only products priced below 100 yuan, between 100 and 300 yuan, between 300 and 500 yuan, or above 500 yuan. Further filtering is conducted based on logo customization method, packaging method, and process type, eliminating products that do not meet the company's customization requirements. Finally, availability is verified based on inventory quantity and supply cycle. Inventory quantity is judged according to four ranges: below 100 units, 100-500 units, 501-1000 units, and above 1000 units. Supply cycle is judged according to four ranges: within 3 days, 4-7 days, 8-15 days, and above 15 days. Only products with sufficient inventory to meet the procurement quantity and a supply cycle that meets delivery requirements are retained. After completing this layer-by-layer filtering, product relationships are matched based on applicable industries and gifting scenarios to form a candidate product set.

[0044] The product price range is read and matched with the company's procurement budget range. The budget matching degree is highest when the product price and procurement budget are in the same range; second-best when the product price is in an adjacent budget range; otherwise, the budget matching degree is reduced. Identity attribute tags, interest preference tags, job attribute tags, and corporate culture fit tags from the customer's personalized profile are read and matched with product tags item by item. The number of common tags and the tag coverage ratio are counted. The more common tags, the higher the recipient's preference matching degree. Based on the product usage scenario tags, they are matched with six gifting scenarios: customer visits, business cooperation, employee benefits, meetings and events, brand promotion, and holiday greetings. The scenario applicability is high when the product is suitable for the current gifting scenario; medium when suitable for related scenarios; and low otherwise. The availability index is determined based on inventory quantity and supply cycle. Products with inventory of 501-1000 units or more and a supply cycle of 3 days or less or 4-7 days are considered highly available; products with low inventory or long supply cycles have lower availability. After completing the above analysis, budget matching, preference matching, scenario applicability, and available status are combined to form a single product matching index.

[0045] The system prioritizes products whose budgets perfectly align with the company's purchasing range. For products with identical budgets, a second ranking is based on the recipient's preferences, prioritizing those with high relevance to interest tags, job titles, and company culture. If multiple products still rank identically, a third ranking is conducted based on the applicable scenario, placing products directly suited to the current gifting context first. Finally, a final ranking is based on availability, prioritizing products with inventory quantities between 501 and 1000 units or more and a supply cycle of 3 days or less or 4 to 7 days to ensure timely production, customization, and delivery. After ranking, a product recommendation set is output according to priority, along with information for each product including price range, recommended packaging, logo customization, manufacturing process, inventory status, supply cycle, and applicable scenarios.

[0046] In this embodiment, the step of intelligently recommending product combinations based on a customer-specific profile and outputting a product combination recommendation scheme further includes: The product combination recommendation scheme is output to the user, and user recommendation feedback data is collected; Based on the user recommendation feedback data, the actual purchase results, purchase amount, customer evaluation information, satisfaction score, repeat purchase behavior and product return and exchange status are calculated to generate a recommendation effect index; Record the product portfolio recommendation scheme and recommendation effect index, generate independent recommendation samples, and store them in the cloud.

[0047] In this embodiment, recommended product combinations are presented to procurement personnel through the enterprise procurement platform. Each recommended solution includes information such as product category, quantity, price range, logo customization method, packaging method, process type, supply cycle, and applicable scenarios, and is displayed in order of recommendation priority. Procurement personnel can perform operations such as browsing, clicking, saving, comparing solutions, consulting suppliers, and submitting purchase requests. The platform records all interactive behaviors in real time, forming user recommendation feedback data, including recommendation clicks, browsing time, number of saves, solution adoption status, and purchase request information. After procurement personnel complete the purchase, the platform continues to collect order execution status, product delivery status, and receipt status, and associates the purchase order number to complete the entire process recording. All feedback data is uniformly associated and stored according to the recommended solution number and enterprise number, ensuring that each recommendation corresponds to complete user feedback information.

[0048] Read user recommendation feedback data and statistically analyze the actual procurement situation corresponding to each recommended solution. Determine whether a purchase order has been placed for the recommended products, and statistically analyze the actual quantity and amount of products purchased. The purchase amount is still divided into four budget ranges: below 100 yuan, 100-300 yuan, 300-500 yuan, and above 500 yuan. Analyze the consistency between the recommended solutions and the company's budget. Read customer evaluation information and classify the evaluation content into "very satisfied," "satisfied," "neutral," and "dissatisfied." Calculate the satisfaction rating using integer scales from 1 to 5 points, where 5 points represents "very satisfied" and 1 point represents "very dissatisfied." Continue to statistically analyze whether the company repurchases the same product combination within a preset statistical period to identify duplicate purchasing behavior. Simultaneously, read after-sales data to statistically analyze product returns, exchanges, and quality anomalies. When the purchase completion rate is high, the satisfaction rating is high, the number of repeat purchases is high, and the number of returns and exchanges is low, the recommendation effectiveness evaluation of the corresponding solution is improved; conversely, the evaluation result is lowered. A recommendation effectiveness index is generated for each product combination solution, and a correspondence is established with the recommendation solution number.

[0049] The platform comprehensively organizes the product combination recommendation schemes, product composition information, recommendation time, company industry, gifting scenario, procurement budget range, procurement quantity range, customer-specific tags, recommendation effectiveness index, and user feedback information from this recommendation process, and generates a unique recommendation sample number based on the recommendation task number. The recommendation samples undergo completeness verification to ensure that the recommendation scheme information, procurement feedback information, and recommendation effectiveness index all have corresponding records before being packaged into independent recommendation samples. To facilitate the retrieval of historical recommendation data, the platform establishes a categorized index according to the company industry, product category, gifting scenario, and budget range, and uploads this index to a cloud database for long-term storage, while also retaining the sample creation time, update time, and version information. When a company subsequently initiates product procurement, it can quickly retrieve similar historical recommendation samples for comparative analysis, improving product recommendation efficiency.

[0050] In this embodiment, a personalized product intelligent recommendation system based on user profile analysis is provided, used to execute the personalized product intelligent recommendation method based on user profile analysis as described above, including: The customer profiling module is used to collect anonymized data from gift recipients, perform comprehensive profiling analysis, and generate personalized customer profiles. The requirements analysis module is used to obtain product procurement information submitted by enterprises, perform customized requirements matching, and output product customization requirements information. The product library construction module is used to identify the product resource library, identify individual product attributes in the product resource library, and construct a product knowledge graph. The matching module is used to perform range convergence retrieval and multi-dimensional matching on the product knowledge graph based on product customization requirements information, and to build a product recommendation set; The intelligent recommendation module is used to make intelligent product combination recommendations based on the customer's unique profile and output the product combination recommendation scheme.

[0051] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0052] In the embodiments provided in this application, it should be understood that the disclosed apparatus / devices and methods can be implemented in other ways. For example, the apparatus / device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0053] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0054] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0055] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein are implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A personalized product intelligent recommendation method based on user profile analysis, characterized in that, Includes the following steps: Collect anonymized data from recipient customers, perform comprehensive profiling and analysis, and generate personalized customer profiles. Obtain product procurement information submitted by enterprises, match customized needs, and output product customization requirement information; Identify the product resource library, identify individual product attributes in the product resource library, and construct a product knowledge graph; Based on product customization requirements, a product knowledge graph is used for range convergence retrieval and multi-dimensional matching to construct a product recommendation set; Based on a customer's unique profile, the system intelligently recommends product combinations from the product recommendation set and outputs a product combination recommendation solution.

2. The personalized product intelligent recommendation method based on user profile analysis according to claim 1, characterized in that, The specific steps for collecting anonymized data from recipient customers, performing comprehensive profiling and analysis, and generating personalized customer profiles are as follows: Collect anonymized data from recipient customers; analyze the anonymized data for identity features, extract customer age, gender, interests, company position, job level, and department information, and generate customer identity features; Obtain basic attribute information of the customer's company, including company location, business type, corporate culture tags, brand color tags, and brand positioning; Perform feature association analysis on the customer identity characteristics and the basic attribute information to generate multiple customer-specific tags; The customer-specific tags include identity attribute tags, interest preference tags, consumption tendency tags, job attribute tags, and corporate culture fit tags; Based on the customer-specific tags, a comprehensive analysis of the profile is performed to generate a customer-specific profile.

3. The personalized product intelligent recommendation method based on user profile analysis according to claim 2, characterized in that, The specific steps for obtaining product procurement information submitted by enterprises, matching customized needs, and outputting product customization requirement information are as follows: Obtain product procurement information submitted by enterprises; perform structured parsing on the product procurement information to extract key elements; the key elements include product usage scenarios, product types, procurement quantities, and procurement budgets; Based on the aforementioned key elements, a gift-giving scenario demand analysis is conducted to generate multiple gift-giving scenario demand features. Based on the characteristics of multiple gift-giving scenarios, we perform customized demand matching for customer profiles and output customized product demand information.

4. The personalized product intelligent recommendation method based on user profile analysis according to claim 3, characterized in that, The specific steps for identifying the product resource library, performing individual product attribute identification on the product resource library, and constructing a product knowledge graph are as follows: Identify the product resource library, perform individual product attribute identification on the product resource library, and extract the basic attributes, customized attributes, and commercial attributes of each product; The basic attributes, customized attributes, and commercial attributes are tagged to generate multiple product tags; Calculate the price of a single product; The prices are associated with multiple product tags to construct a product knowledge graph.

5. The personalized product intelligent recommendation method based on user profile analysis according to claim 4, characterized in that, The basic attributes include product category, material, color, size, and brand information; the customization attributes include logo customization method, packaging method, process type, and customization cycle; and the commercial attributes include price range, inventory quantity, supply cycle, and applicable industries.

6. The personalized product intelligent recommendation method based on user profile analysis according to claim 4, characterized in that, The specific steps for constructing a product recommendation set by performing range convergence retrieval and multi-dimensional matching on the product knowledge graph based on product customization demand information are as follows: Based on product customization requirements, the product knowledge graph is used for range convergence retrieval and multi-dimensional matching to generate a candidate product set. Based on the candidate product set, calculate the budget matching degree, recipient preference matching degree, scenario applicability and availability index matching degree, and generate a single product matching degree index; Intelligent sorting is performed based on individual product matching metrics to construct a product recommendation set.

7. The personalized product intelligent recommendation method based on user profile analysis according to claim 6, characterized in that, The specific steps for intelligently recommending product combinations based on customer-specific profiles and outputting product combination recommendation schemes are as follows: Based on customer-specific profiles, user types are categorized into multiple gift types, including customer maintenance, employee welfare, event promotion, and business cooperation. Based on multiple gift types, the product recommendation set is intelligently recommended to create product combinations, and a product combination recommendation scheme is output.

8. The personalized product intelligent recommendation method based on user profile analysis according to claim 7, characterized in that, The intelligent product combination recommendation based on the customer-specific profile and the output product combination recommendation solution also include: The product combination recommendation scheme is output to the user, and user recommendation feedback data is collected; Based on the user recommendation feedback data, the actual purchase results, purchase amount, customer evaluation information, satisfaction score, repeat purchase behavior and product return and exchange status are calculated to generate a recommendation effect index; Record the product portfolio recommendation scheme and recommendation effect index, generate independent recommendation samples, and store them in the cloud.

9. A personalized product intelligent recommendation system based on user profile analysis, characterized in that, The method for performing personalized product intelligent recommendation based on user profile analysis as described in claim 1 includes: The customer profiling module is used to collect anonymized data from gift recipients, perform comprehensive profiling analysis, and generate personalized customer profiles. The requirements analysis module is used to obtain product procurement information submitted by enterprises, perform customized requirements matching, and output product customization requirements information. The product library construction module is used to identify the product resource library, identify individual product attributes in the product resource library, and construct a product knowledge graph. The matching module is used to perform range convergence retrieval and multi-dimensional matching on the product knowledge graph based on product customization requirements information, and to build a product recommendation set. The intelligent recommendation module is used to make intelligent recommendations for product combinations based on the customer's unique profile and output the product combination recommendation scheme.

Images