Price verification method, device and equipment based on identification of commodities of same type, and storage medium

Abstract

The invention discloses a price verification method, device and equipment based on identification of commodities of the same type and a storage medium, and relates to the technical field of commodity price comparison, and the method comprises the steps: obtaining heterogeneous commodity data of a plurality of suppliers; processing the heterogeneous commodity data to obtain standardized commodity data; performing multi-modal feature fusion on the standardized commodity data to generate a comprehensive feature vector corresponding to each commodity; calculating the similarity between the commodities according to the comprehensive feature vector, and clustering the commodities with the similarity greater than a similarity threshold value into a same-type commodity set; performing price comparison analysis on commodities in the same-type commodity set to obtain a reference price of the same-type commodity set; receiving a purchasing request, and determining a same-type commodity set based on the commodity identifier in the purchasing request; comparing the commodity price in the purchase request with the reference price of the same commodity set to obtain a verification result; and responding to a purchase request according to the verification result. According to the method, the accuracy of price comparison of the same commodity is improved.

Description

Technical Field

[0001] This application relates to the field of commodity price comparison technology, and in particular to a price verification method, apparatus, device and storage medium based on the identification of the same commodity. Background Technology

[0002] Currently, the industry commonly uses a traditional technical solution based on keyword matching, manual price comparison, and post-purchase compliance checks. This solution typically includes the following steps: After suppliers voluntarily upload product information, the system performs initial matching using simple product name keywords or fixed SKU codes; purchasing personnel manually verify product images, parameters, and other information to determine if they are identical products, and manually collect and organize quotes from various suppliers for comparison; after a purchase order is generated, the compliance department conducts random checks on completed orders to verify for issues such as over-pricing or improper supplier selection. Structurally, this solution usually consists of independent product upload, keyword matching, order generation, and post-purchase check modules, lacking a unified intelligent analysis and real-time control center.

[0003] However, existing technical solutions are not very accurate in comparing prices of the same product. Summary of the Invention

[0004] This application provides a price verification method, apparatus, device, and storage medium based on the identification of similar products, which can improve the accuracy of price comparison of similar products.

[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, this application provides a price verification method based on the identification of identical products, including: Acquire heterogeneous product data from multiple suppliers; The heterogeneous product data is processed to obtain standardized product data; Multimodal feature fusion is performed on the standardized commodity data to generate a comprehensive feature vector corresponding to each commodity; Based on the comprehensive feature vector, the similarity between products is calculated, and products with similarity greater than the similarity threshold are clustered into a set of similar products; Comparative price analysis is performed on products within the same product set to obtain the benchmark price for the same product set; Upon receiving a purchase request, determine the corresponding set of similar products based on the product identifier in the purchase request; The price of the goods in the purchase request is compared with the benchmark price of the same product set to obtain the verification result; Based on the verification result, respond to the procurement request.

[0006] Optionally, the verification results include pass, warning, and block; the comparison of the commodity price in the purchase request with the benchmark price of the same commodity set to obtain the verification result includes: The verification result is passed when the price of the goods in the purchase request is less than or equal to the benchmark price. When the price of the goods in the purchase request is greater than the benchmark price but less than or equal to a first percentage of the benchmark price, the verification result is a warning. When the price of the goods in the purchase request is greater than a second percentage of the benchmark price, the verification result is blocked; the first percentage is less than the second percentage.

[0007] Optionally, responding to the procurement request based on the verification result includes: The purchase request is allowed if the verification result is satisfactory. When the verification result is a warning, the approval process is triggered, and the procurement request is allowed after the approval is passed. When the verification result is blocked, the purchase request is not allowed.

[0008] Optionally, the step of performing multimodal feature fusion on the standardized commodity data to generate a comprehensive feature vector corresponding to each commodity includes: Feature extraction is performed on the text information in the standardized commodity data to obtain a text feature vector; Feature extraction is performed on the image information in the standardized commodity data to obtain image feature vectors; The text feature vector and the image feature vector are fused to obtain the comprehensive feature vector.

[0009] Optionally, the step of performing a price comparison analysis on products within the same product set to obtain a benchmark price for the same product set includes: Calculate the lowest price, average price, and highest price of each item within the same product set; Based on the lowest price, average price, highest price, and performance rating, calculate the overall cost-effectiveness score for each product; The benchmark price is obtained based on the overall cost-effectiveness score.

[0010] Optionally, the calculation of the overall cost-effectiveness score for each product based on the lowest price, average price, highest price, and fulfillment rating includes:

[0011]

[0012] in, Indicates goods The overall cost-effectiveness score, Indicates the first weight. Indicates the second weight. Represents the transition function. Indicates goods The price Indicates supplier Performance rating The first coefficient representing the price factor, The second coefficient represents the price factor. The third coefficient representing the price factor, Indicates the lowest price. Indicates the highest price. This indicates the average price.

[0013] Optionally, the performance rating is obtained in the following way:

[0014] in, Indicates supplier Performance rating Indicates based on supplier Credit rating adjustment factor Indicates the first The weight of each performance indicator Indicates supplier In the Standardized scores on each performance indicator This indicates the total number of performance targets.

[0015] Secondly, this application provides a price verification device based on the identification of identical products, comprising: The acquisition module is used to acquire heterogeneous product data from multiple suppliers; The processing module is used to process the heterogeneous product data to obtain standardized product data; perform multimodal feature fusion on the standardized product data to generate a comprehensive feature vector corresponding to each product; calculate the similarity between products based on the comprehensive feature vector, and cluster products with similarity greater than a similarity threshold into a set of similar products; perform price comparison analysis on products within the set of similar products to obtain the benchmark price of the set of similar products. The receiving module is used to receive purchase requests and determine the corresponding set of similar products based on the product identifier in the purchase request. The comparison module is used to compare the price of the goods in the purchase request with the benchmark price of the same goods in the same category to obtain a verification result; and respond to the purchase request based on the verification result.

[0016] Thirdly, this application provides a computing device, including a memory and a processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of the first aspects.

[0017] Fourthly, this application provides a computer-readable storage medium for storing a computer program for performing the method as described in any one of the first aspects.

[0018] As can be seen from the above technical solution, this application has at least the following beneficial effects: In this application, multimodal feature fusion technology is used to jointly analyze and vectorize product text information (title, parameters) and image information (main image), generating a comprehensive feature vector that fully represents product attributes. Based on this, through similarity calculation and clustering algorithms, heterogeneous descriptions of the same product from different suppliers can be automatically and accurately aggregated into a single product set. This process fundamentally solves the matching problem caused by the heterogeneity of product information, ensuring that subsequent price comparison analysis is based on an accurate and complete pool of similar products, providing a reliable data foundation for intelligent decision-making.

[0019] Furthermore, in addition to calculating basic information such as the lowest, average, and highest prices for the same product set, supplier performance ratings were introduced, and a quantitative comprehensive cost-effectiveness score calculation model was established. This model not only considers price factors but also incorporates performance ratings reflecting the supplier's historical performance and creditworthiness, scientifically integrating them through weighting. This results in a final benchmark price or price comparison result that is no longer a single price figure but an intelligent recommendation that integrates price competitiveness and supplier reliability, effectively supporting high-quality, high-price procurement decisions and facilitating a balance between procurement costs and procurement quality.

[0020] Furthermore, by linking benchmark prices with procurement requests in real time, a real-time verification mechanism embedded in the procurement process was constructed. The system automatically triggers verification when a procurement order is placed, instantly generating verification results such as approval, warning, or blocking based on preset rules, such as price ratio thresholds, and automatically executing corresponding control actions (direct approval, forwarding for approval, or blocking). This closed-loop design transforms compliance rules into executable system logic, enabling real-time intervention at key procurement stages. This effectively prevents violations such as high-price procurement, significantly reduces compliance risks and supervision costs, and achieves transparency, standardization, and controllability in the procurement process.

[0021] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0022] Figure 1 A flowchart illustrating a price verification method based on the identification of identical products provided in this application embodiment; Figure 2 A schematic diagram of a price verification device based on the identification of identical products provided in this application embodiment; Figure 3 This is a schematic diagram of a computing device provided in an embodiment of this application. Detailed Implementation

[0023] The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.

[0024] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0025] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the related technologies is given first: A "same product set" refers to a collection of products identified by algorithms that have the same or highly similar product attributes (such as model, specifications, functions, etc.) but are provided by different suppliers.

[0026] In the actual operation of multi-supplier e-procurement marketplaces for central and state-owned enterprises, the main technical challenge lies in how to automatically, accurately, and in real-time identify the same products offered by different suppliers. Existing technologies primarily rely on keyword matching and manual comparison, which is not only inefficient but also suffers from extremely low accuracy and recall rates due to significant heterogeneity in product descriptions (such as differences in text wording, parameter formats, and image display). This makes it impossible to provide a reliable data foundation for subsequent price comparison and decision-making. This fundamental deficiency further hinders the realization of automated intelligent price comparison and real-time compliance management.

[0027] In terms of contract performance compliance, existing technical solutions lack the ability to dynamically monitor and intelligently intervene in the procurement contract execution process, causing compliance risks to spread from the price domain to the delivery and quality domains. For example, in the procurement of central enterprises, the same Lenovo ThinkPad X1 Carbon laptop may affect the project schedule due to delayed delivery by supplier A, or cause safety risks due to batch quality problems in the products provided by supplier C. However, the existing e-commerce system only performs simple price comparison at the order generation stage and cannot call the supplier's historical performance rating (such as on-time delivery rate and quality pass rate) in real time when making procurement decisions, let alone automatically trigger procurement restrictions or warnings when the supplier has performance abnormalities (such as continuous delivery delays).

[0028] The underlying cause of these problems lies in the systemic backwardness of the existing technical architecture. Firstly, at the technical level, the solution relies too heavily on fuzzy matching of single text keywords, failing to integrate multimodal information about the product (such as text and images), resulting in incomplete feature representation and poor recognition accuracy. Secondly, in terms of system architecture, the various functional modules (product upload, matching, order processing, and spot checks) operate independently, forming data silos and lacking a unified intelligent analysis center. Finally, in terms of business processes, compliance control adopts a passive, post-event spot check model, unable to perform real-time verification and intervention at critical stages of procurement and order placement. These three factors collectively lead to the current situation of low procurement efficiency, difficulty in cost control, and high compliance risks.

[0029] In view of this, embodiments of this application provide a price verification method based on the identification of identical products, which can be executed by a processing device. The processing device can be a terminal or a server. Terminals include, but are not limited to, smartphones, tablets, laptops, personal digital assistants, or smart wearable devices. The server can be a cloud server, such as a central server in a central cloud computing cluster or an edge server in an edge cloud computing cluster. Alternatively, the server can be a server in a local data center. A local data center refers to a data center directly controlled by the user.

[0030] To address the problems of low accuracy in identifying identical products, poor price comparison efficiency, and lagging compliance control in existing technologies, this invention proposes an intelligent control scheme based on multimodal feature fusion and a real-time rule engine. First, by fusing product text and image information to construct a comprehensive feature vector, this invention overcomes the limitations of single-text matching, achieving high-precision normalization and identical product identification for heterogeneous products. Second, based on the accurately identified set of identical products, a multi-dimensional cost-effectiveness evaluation model integrating price levels and supplier performance is established, providing intelligent support for procurement decisions. Finally, through real-time integration of a configurable rule engine with the order system, compliance verification is transformed from post-event sampling to real-time in-process control, automatically executing verification and intervention at key procurement stages, forming an automated closed loop from identification and price comparison to control, thereby systematically improving procurement efficiency, reducing costs, and ensuring compliance.

[0031] To make the technical solution of this application clearer and easier to understand, the following describes a price verification method based on the identification of identical products provided by an embodiment of this application, in conjunction with the accompanying drawings. Figure 1 As shown, this figure is a flowchart of a price verification method based on the identification of identical products provided in an embodiment of this application. The method includes: S201. The processing equipment acquires heterogeneous product data from multiple suppliers.

[0032] Multiple suppliers refer to multiple independent product providers that are listed on the same procurement platform (such as the State-owned Enterprises E-commerce Mall), and are the source of product and price data.

[0033] Heterogeneous product data refers to original product information from different suppliers that differs significantly in description, data format, and information completeness for the same or similar products. This includes text information (such as product title, specifications, brand and model), image information (such as main product image and detail image), numerical information (such as price and inventory), and attribute information (such as supplier ID and category label). Its heterogeneity is reflected in the non-standardization of naming conventions, parameter units, image specifications, and data structures.

[0034] The processing equipment first actively collects or receives product listing data from multiple suppliers through the back-end system of the state-owned enterprise's e-commerce platform or via data interfaces. This raw data is inherently heterogeneous: for example, for the same laptop, supplier A might describe it as "Lenovo ThinkPad X1 Carbon 2024 i5 16G 512G SSD," while supplier B might abbreviate it as "Lenovo X1C i5 / 16 / 512," and the accompanying images, resolutions, and price formats may differ. Acquiring this heterogeneous product data is the foundation for all subsequent intelligent processing, aiming to aggregate multi-source information and provide rich raw materials for subsequent normalization, comparison, and decision-making.

[0035] S202. The processing equipment processes heterogeneous commodity data to obtain standardized commodity data.

[0036] Standardized commodity data refers to commodity information that conforms to unified data specifications, generated after the original heterogeneous information is automatically transformed and reconstructed through a pre-defined standardized processing model. This standardized processing model defines the mapping rules and transformation logic from multi-source heterogeneous data to standard data structures, including sub-models such as text normalization, image standardization, numerical normalization, and attribute encoding.

[0037] The processing equipment performs a systematic structural transformation on the input heterogeneous commodity data by calling a pre-built standardized processing model. For text modal information, the model first extracts key descriptive entities based on a pre-trained named entity recognition module, and then uses a rule engine-driven semantic mapping mechanism to parse and normalize the multi-dimensional expressions (e.g., 1kg, 1000g, 1 kilogram) into a standardized format that conforms to international measurement standards (e.g., mass: 1000.0, unit: grams).

[0038] For image modal data, the model implements a multi-stage visual preprocessing pipeline, including size normalization resampling based on bilinear interpolation, color space uniform transformation (such as RGB normalization), and quality enhancement processing based on frequency domain filtering, thereby generating a reference image with consistent resolution, color distribution, and signal-to-noise ratio, providing normalized input for subsequent visual feature extraction.

[0039] In the dimension of numerical information processing, the model relies on a structured unit knowledge graph for automatic identification and conversion. By dynamically matching the semantics and dimensional relationships of units, it achieves accurate conversion and decimal alignment across unit systems (such as length, weight, and volume). Simultaneously, for product attribute information, the model maps unstructured tags such as supplier identifiers, product categories, and specifications to a unified coding space by searching a predefined multi-level classification coding table, thus completing the discretization and structured representation of attributes.

[0040] This model-driven standardized process systematically maps the semantic differences, format inconsistencies, and structural diversity inherent in heterogeneous data sources to a unified, complete, and machine-resolvable data representation space. This ensures the consistency and comparability of inputs for subsequent multimodal fusion, similarity calculation, and intelligent decision-making algorithms at the data level, laying the foundation for automated processing throughout the entire process.

[0041] S203. The processing equipment performs multimodal feature fusion on standardized commodity data to generate comprehensive feature vectors for each commodity.

[0042] Multimodal features refer to the feature data of different forms (modalities) in product information. This application mainly involves text modal features and image modal features. Multimodal feature fusion aims to comprehensively utilize features from different information sources to more comprehensively represent products.

[0043] Specifically, features are extracted from the text information in the standardized commodity data to obtain a text feature vector; features are extracted from the image information in the standardized commodity data to obtain an image feature vector; and the text feature vector and the image feature vector are fused to obtain the comprehensive feature vector.

[0044] The processing device first performs feature extraction on the standardized text information: based on a pre-trained natural language processing model, such as BERT or Word2Vec, it transforms the structured text, such as the product title and key parameters, into a high-dimensional text feature vector rich in semantic information. This vector not only captures literal keywords but also understands the relationships between parameters and the context, thereby achieving a deep semantic representation of the product's functions and specifications.

[0045] Simultaneously, the processing equipment performs feature extraction on the standardized image information: using pre-trained computer vision models, such as ResNet or CNN, it analyzes the product display images and extracts image feature vectors covering color, shape, texture, and key visual elements. This process encodes visual information into a machine-understandable numerical form, enabling it to characterize the product's appearance, model, and even some manufacturing processes.

[0046] Finally, feature fusion techniques are used to integrate the two heterogeneous but complementary feature vectors. Common fusion methods include vector concatenation, weighted addition, or attention-based fusion, the aim of which is to establish a shared feature space, allowing the semantic information of the text and the visual information of the image to mutually enhance and verify each other. The resulting comprehensive feature vector possesses both semantic accuracy and visual recognizability, constituting a unique and comprehensive digital fingerprint for the product, laying the foundation for subsequent calculations of product similarity and reliable clustering of similar products.

[0047] S204. The processing device calculates the similarity between products based on the comprehensive feature vector, and clusters products with similarity greater than the similarity threshold into a set of the same type of products.

[0048] The processing device takes the comprehensive feature vectors generated for each product as input and calculates the similarity between any two product feature vectors using mathematical methods, such as cosine similarity. This similarity measure quantifies the degree of proximity between products in a multi-dimensional space that integrates textual semantics and visual features. The higher the value, the more likely the products are to be the same product in terms of essential attributes such as model, specifications, and appearance.

[0049] The expression for calculating similarity is:

[0050] in, Indicates similarity. Represents the vector dot product. The Euclidean norm of a vector. Indicates goods The comprehensive feature vector, Indicates goods The comprehensive feature vector.

[0051] After calculating the similarity between all pairs of products, the system generates a global similarity matrix. This matrix mathematically characterizes the strength of the feature association between any two products in the entire product pool. Based on this matrix, the processing device uses clustering algorithms, such as the DBSCAN algorithm which can autonomously identify dense regions, or hierarchical clustering algorithms that can form hierarchical cluster structures, to automatically aggregate and classify products with similarity exceeding a preset threshold.

[0052] The similarity threshold is a crucial technical parameter that substantially controls the strictness of clustering discrimination. A high similarity threshold will only classify a small number of products with highly consistent features as the same type, which is beneficial for improving recognition accuracy, but may miss some products with significant description differences. A low threshold can recall more potential products of the same type, improving recall rate, but may also introduce false positives for dissimilar products. Therefore, setting the similarity threshold often requires finding the optimal balance between accuracy and recall. This balancing process can be achieved by optimizing the objective function, with the optimal threshold being... The selection can be expressed in the following form:

[0053] in, It is an evaluation function for clustering performance, and its value comprehensively reflects the results at a threshold. The value indicates the quality of the clustering results; a higher value means a better clustering effect. This indicates the lower limit of the threshold search range. This indicates the upper limit of the threshold search range.

[0054]

[0055] in, Indicates accuracy, within a threshold. Below, the proportion of product pairs correctly clustered as the same type out of all product pairs judged as the same type by the system. Recall rate, at the threshold Below, the proportion of product pairs correctly clustered as the same type out of all truly identical product pairs.

[0056] This mathematical expression describes an optimization process. It involves optimizing within a pre-defined, reasonable range (…). Automatically adjusts and tests different similarity thresholds within the system. And calculate the overall clustering score corresponding to each threshold. Finally, the threshold that maximizes the overall score is selected as the optimal threshold for actual system use. .

[0057] After the algorithm is executed, products that are close to each other in the feature space and have a high degree of similarity will be automatically grouped into the same independent set, forming a set of similar products. At the same time, products that do not show sufficient similarity to any product group are regarded as noise points or unique products and are processed or excluded separately by the algorithm.

[0058] This process enables the automated and structured transformation from raw feature data into a collection of similar products with clear business implications, laying a data foundation for subsequent price comparison and compliance analysis.

[0059] S205. The processing equipment performs a price comparison analysis on the products within the same product set to obtain the benchmark price of the same product set.

[0060] Price comparison analysis refers to the technical process of systematically comparing, calculating, and evaluating the prices and related business conditions of goods offered by different suppliers within the same product set, provided that the goods belong to the same product set.

[0061] A benchmark price is a reference standard determined through price comparison analysis and used to measure the reasonableness of the price level of various products in a group of similar products.

[0062] Specifically, the lowest, average, and highest prices of each product within the same product set are calculated; based on the lowest, average, and highest prices and the fulfillment rating, a comprehensive cost-effectiveness score is calculated for each product; and a benchmark price is obtained based on the comprehensive cost-effectiveness score.

[0063] First, the processing equipment performs basic price statistics calculations within the defined set of similar products, extracting and determining the lowest, average, and highest prices for all products in the set. These three price indicators collectively outline the overall price distribution range and level of the same product in the current market, providing an objective quantitative benchmark for subsequent in-depth analysis.

[0064] Next, the system introduces the performance rating as a business dimension. The performance rating is a quantitative assessment of a supplier's comprehensive capabilities, including historical delivery performance, product quality, and service level. The processing equipment combines the price information of each product (including its relative position to the lowest, average, and highest prices) with its corresponding supplier's performance rating, calculates the overall cost-effectiveness score for each product using a pre-set weighted model, and outputs this score. This overall cost-effectiveness score is no longer limited to price alone, but integrates price and quality, achieving a decision-making upgrade from lowest price to optimal value.

[0065] The formula for calculating the overall cost-effectiveness score is:

[0066]

[0067] in, Indicates goods The overall cost-effectiveness score, Indicates the first weight. Indicates the second weight. Represents the transition function. Indicates goods The price Indicates supplier Performance rating The first coefficient representing the price factor, The second coefficient represents the price factor. The third coefficient representing the price factor, Indicates the lowest price. Indicates the highest price. This indicates the average price.

[0068] Performance ratings are obtained through the following methods:

[0069] in, Indicates supplier Performance rating Indicates based on supplier Credit rating adjustment factor Indicates the first The weight of each performance indicator Indicates supplier In the Standardized scores on each performance indicator This represents the total number of performance indicators, such as on-time delivery rate, quality pass rate, and service satisfaction.

[0070] Taking the purchase of Lenovo ThinkPad X1 Carbon laptops as an example, the system calculates the performance rating for supplier B. This may stem from its performance record over the past 12 months: on-time delivery rate of 92% (weighted). ), 98% of goods delivered with a qualified rate (weighted) After-sales rating: 4.5 / 5.0 (weighted) (and take into account its credit rating adjustment factor) The result was obtained later. If supplier B's performance rating... If the score is higher than that of supplier C (due to its recent multiple delivery delays), then when calculating the overall cost-effectiveness score, even if supplier C's price is slightly lower, supplier B may receive a higher overall score due to its superior performance, and thus be prioritized by the system as a reference for setting the benchmark price or as the preferred supplier.

[0071] Furthermore, performance ratings are not only used for price comparison decisions, but also directly drive real-time compliance management. For example, administrators can preset rules: if a supplier's performance rating falls below a performance rating threshold... Any new purchase request from that supplier will automatically trigger a high-risk approval process. When a purchaser selects goods from such a supplier, the system will forcibly alert the purchaser to performance risks and escalate the approval process, thereby blocking high-risk transactions at the purchase initiation stage. This achieves a fundamental shift in compliance control from post-event accountability to pre-event prevention and in-event interception.

[0072] The resulting benchmark price is an intelligent benchmark that deeply integrates market price signals and supplier performance risk assessment. It not only answers which supplier is cheaper, but also which supplier offers the best overall value while meeting compliance and risk control requirements, providing enterprises with a quantifiable and actionable technical path to implement the principle of "high quality, high price" in procurement.

[0073] Finally, the system determines the final benchmark price based on the calculated overall cost-effectiveness score. The expression for the benchmark price is:

[0074] in, Indicates the benchmark price. It is the total number of products of the same type. It is the first The overall cost-effectiveness score of each product; It is the first The price of each product.

[0075] The weighted approach ensures that products with higher scores have a greater impact on the benchmark price. The resulting benchmark price is an intelligent recommendation benchmark that integrates multi-dimensional evaluations. It not only objectively reflects the price distribution of similar products in the current market but also incorporates a reliability assessment of the supplier's historical performance. This method ensures that the final benchmark price is both market-sensitive and business-reasonable, providing a basis for subsequent real-time compliance verification and procurement decisions that adheres to market principles while meeting the company's quality control and risk prevention requirements.

[0076] S206. The processing equipment receives a purchase request and determines the corresponding set of similar products based on the product identifier in the purchase request.

[0077] When purchasing personnel select products, enter quantities, and submit orders on the online store interface, the system generates a structured purchase request. This request includes at least product identifiers (such as product IDs, SKU codes, and other unique codes), purchase quantities, and desired suppliers.

[0078] Upon receiving the request, the processing device immediately initiates real-time query and matching logic. This operation uses the product identifier as an index to quickly retrieve data from the previously constructed database of similar product sets. By comparing the identifier code, the system locates the specific set of similar products into which the algorithm categorized the product during the preprocessing stage.

[0079] This matching process maps individual purchased goods to the set of similar goods to which they belong. Its technical significance lies in linking discrete purchasing actions initiated on the front end with structured product knowledge generated through intelligent analysis in the back end. Once the set of similar goods to which the product belongs is successfully identified, the system obtains a panoramic reference system for that product across the entire multi-supplier market, including price lists of all similar products, supplier performance ratings, and benchmark prices derived from intelligent analysis. This provides an indispensable context and data foundation for subsequent real-time price comparisons and compliance checks before order generation, serving as a crucial link in achieving in-process control in online procurement.

[0080] S207. The processing equipment compares the price of the goods in the purchase request with the benchmark price of the same type of goods in the same product set, and obtains the verification result.

[0081] Specifically, when the price of the goods in the purchase request is less than or equal to the benchmark price, the verification result is passed; when the price of the goods in the purchase request is greater than the benchmark price but less than or equal to a first percentage of the benchmark price, the verification result is a warning; when the price of the goods in the purchase request is greater than a second percentage of the benchmark price, the verification result is blocked; the first percentage is less than the second percentage.

[0082] The system compares the price of the goods in the purchase request with the benchmark price of the same goods in the same product set in real time. If the price of the goods is less than or equal to the benchmark price, it indicates that the quoted price is not higher than the fair market level recommended by the system, and the verification result is directly judged as passed, and the order can continue to be processed.

[0083] If the price of a product is higher than the benchmark price, but does not exceed the first percentage of the benchmark price (e.g., 110% of the benchmark price), it indicates that the quoted price is high but still within an acceptable fluctuation range. The system will issue a warning. In this case, the process will not be automatically blocked, but an approval mechanism will be triggered, requiring the purchasing personnel to explain the reasons or submit the application for review by their superiors.

[0084] If the price of a product exceeds the second percentage of the benchmark price (e.g., 120% of the benchmark price), it means that the quoted price deviates significantly from the reasonable range and poses a high risk of violation. The system will decisively determine that the order has been blocked, directly preventing the order from being generated, and prompting the purchasing personnel to reselect or file an appeal.

[0085] This rule-based design enables a shift from indiscriminate interception to refined control, ensuring compliance while allowing for flexibility in handling reasonable market fluctuations and special circumstances, making the control logic more aligned with actual business needs.

[0086] S208. The processing equipment responds to the procurement request based on the verification results.

[0087] Specifically, when the verification result is passed, the purchase request is allowed; when the verification result is a warning, the approval process is triggered, and the purchase request is allowed after approval; when the verification result is blocked, the purchase request is not allowed.

[0088] When the verification result is satisfactory, the system determines that the current purchase request fully complies with the preset compliance rules and price standards. At this point, the processing device sends a confirmation instruction to the order system, allowing the purchase request to continue with the subsequent process. The system will automatically generate a purchase order without any manual intervention, achieving efficient and automated processing in compliance scenarios.

[0089] When the verification result is a warning, it indicates that although the purchase request is not seriously non-compliant, it has reached the system's risk warning threshold (e.g., the price is too high). In this case, the system will not directly approve it but will immediately trigger the approval process: the order status is set to suspended or pending review, and the approval task and the reason for the warning are automatically pushed to the pre-defined approvers (such as the purchasing manager or compliance specialist). Only after the approver completes the review and gives a clear instruction to approve will the system lift the suspension and allow the purchase request to continue. This mechanism establishes a buffer between efficiency and risk control, ensuring the controlled handling of exceptions.

[0090] When the verification result is blocked, the system identifies that the request has clearly violated mandatory compliance rules (such as severely exceeding price limits). The processing device will immediately terminate the current order generation process, disallowing the procurement request to continue, and returning a clear reason for the block and a prompt message to the procurement personnel, such as suggesting the selection of a similar product with a higher cost-performance ratio. This result usually means that the procurement attempt has been directly rejected by the system, and it cannot continue unless a specially authorized appeal process is initiated. This rigid control effectively prevents major violations from occurring, reflecting the system's zero-tolerance management principle at key risk points.

[0091] Based on the above description, this application has the following beneficial effects: In this application, multimodal feature fusion technology is used to jointly analyze and vectorize product text information (title, parameters) and image information (main image), generating a comprehensive feature vector that fully represents product attributes. Based on this, through similarity calculation and clustering algorithms, heterogeneous descriptions of the same product from different suppliers can be automatically and accurately aggregated into a single product set. This process fundamentally solves the matching problem caused by the heterogeneity of product information, ensuring that subsequent price comparison analysis is based on an accurate and complete pool of similar products, providing a reliable data foundation for intelligent decision-making.

[0092] Furthermore, in addition to calculating basic information such as the lowest, average, and highest prices for the same product set, supplier performance ratings were introduced, and a quantitative comprehensive cost-effectiveness score calculation model was established. This model not only considers price factors but also incorporates performance ratings reflecting the supplier's historical performance and creditworthiness, scientifically integrating them through weighting. This results in a final benchmark price or price comparison result that is no longer a single price figure but an intelligent recommendation that integrates price competitiveness and supplier reliability, effectively supporting high-quality, high-price procurement decisions and facilitating a balance between procurement costs and procurement quality.

[0093] Furthermore, by linking benchmark prices with procurement requests in real time, a real-time verification mechanism embedded in the procurement process was constructed. The system automatically triggers verification when a procurement order is placed, instantly generating verification results such as approval, warning, or blocking based on preset rules, such as price ratio thresholds, and automatically executing corresponding control actions (direct approval, forwarding for approval, or blocking). This closed-loop design transforms compliance rules into executable system logic, enabling real-time intervention at key procurement stages. This effectively prevents violations such as high-price procurement, significantly reduces compliance risks and supervision costs, and achieves transparency, standardization, and controllability in the procurement process.

[0094] The above text combined Figure 1The price verification method based on the identification of the same product provided in the embodiments of this application has been described in detail. The apparatus and equipment provided in the embodiments of this application will be described below with reference to the accompanying drawings.

[0095] like Figure 2 As shown in the figure, this is a schematic diagram of a price verification device based on the identification of identical products provided in an embodiment of this application. The device includes: The acquisition module 301 is used to acquire heterogeneous product data from multiple suppliers; The processing module 302 is used to process the heterogeneous commodity data to obtain standardized commodity data; perform multimodal feature fusion on the standardized commodity data to generate a comprehensive feature vector corresponding to each commodity; calculate the similarity between commodities based on the comprehensive feature vector, and cluster commodities with similarity greater than a similarity threshold into a set of similar commodities; perform price comparison analysis on the commodities within the set of similar commodities to obtain the benchmark price of the set of similar commodities. The receiving module 303 is used to receive a purchase request and determine the corresponding set of similar products based on the product identifier in the purchase request. The comparison module 304 is used to compare the price of the goods in the purchase request with the benchmark price of the same goods set to obtain a verification result; and respond to the purchase request based on the verification result.

[0096] Optionally, the comparison module 304 is specifically used to verify that the price of the goods in the purchase request is less than or equal to the benchmark price. When the price of the goods in the purchase request is greater than the benchmark price but less than or equal to a first percentage of the benchmark price, the verification result is a warning. When the price of the goods in the purchase request is greater than a second percentage of the benchmark price, the verification result is blocked; the first percentage is less than the second percentage.

[0097] Optionally, the processing module 302 is specifically used to allow the purchase request when the verification result is passed; When the verification result is a warning, the approval process is triggered, and the procurement request is allowed after the approval is passed. When the verification result is blocked, the purchase request is not allowed.

[0098] Optionally, the processing module 302 is specifically used to extract features from the text information in the standardized commodity data to obtain a text feature vector; Feature extraction is performed on the image information in the standardized commodity data to obtain image feature vectors; The text feature vector and the image feature vector are fused to obtain the comprehensive feature vector.

[0099] Optionally, the processing module 302 is specifically used to calculate the lowest price, average price, and highest price of each product within the same product set; Based on the lowest price, average price, highest price, and performance rating, calculate the overall cost-effectiveness score for each product; The benchmark price is obtained based on the overall cost-effectiveness score.

[0100] Optionally, the processing module 302 is specifically used to calculate the comprehensive cost-effectiveness score of each product based on the lowest price, average price, highest price, and performance rating, including:

[0101]

[0102] in, Indicates goods The overall cost-effectiveness score, Indicates the first weight. Indicates the second weight. Represents the transition function. Indicates goods The price Indicates supplier Performance rating The first coefficient representing the price factor, The second coefficient represents the price factor. The third coefficient representing the price factor, Indicates the lowest price. Indicates the highest price. This indicates the average price.

[0103] Optionally, the processing module 302 is specifically used to calculate the performance rating:

[0104] in, Indicates supplier Performance rating Indicates based on supplier Credit rating adjustment factor Indicates the first The weight of each performance indicator Indicates supplier In the Standardized scores on each performance indicator This indicates the total number of performance targets.

[0105] The price verification device based on the identification of identical products according to the embodiments of this application can correspondingly execute the method described in the embodiments of this application, and the other operations and / or functions of each module / unit of the price verification device based on the identification of identical products are respectively for the purpose of implementing Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiments will not be described in detail here.

[0106] This application also provides a computing device. For example... Figure 3 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 700 includes a bus 701, a processor 702, a communication interface 703, and a memory 704. The processor 702, the memory 704, and the communication interface 703 communicate with each other via the bus 701.

[0107] The 701 bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0108] The processor 702 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0109] The communication interface 703 is used for external communication.

[0110] Memory 704 may include volatile memory, such as random access memory (RAM). Memory 704 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0111] The memory 704 stores executable code, and the processor 702 executes the executable code to perform the aforementioned price verification method based on the identification of the same product.

[0112] Specifically, in achieving Figure 2 In the case of the illustrated embodiment, and Figure 2 When the modules or units of the price verification device based on the identification of the same product described in the embodiment are implemented by software, the following steps are performed: Figure 2 The software or program code required for the functions of each module / unit can be partially or wholly stored in the memory 704. The processor 702 executes the program code corresponding to each unit stored in the memory 704 to execute the aforementioned price verification method based on the identification of the same product.

[0113] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium capable of being stored by a computing device, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to execute the aforementioned price verification method based on the identification of identical products.

[0114] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.

[0115] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0116] When the computer program product is executed by a computer, the computer executes any of the aforementioned price verification methods based on the identification of identical products. The computer program product can be a software installation package; when any of the aforementioned price verification methods based on the identification of identical products needs to be used, the computer program product can be downloaded and executed on the computer.

[0117] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0118] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A price verification method based on the identification of identical products, characterized in that, The method includes: Acquire heterogeneous product data from multiple suppliers; The heterogeneous product data is processed to obtain standardized product data; Multimodal feature fusion is performed on the standardized commodity data to generate a comprehensive feature vector corresponding to each commodity; Based on the comprehensive feature vector, the similarity between products is calculated, and products with similarity greater than the similarity threshold are clustered into a set of similar products; Comparative price analysis is performed on products within the same product set to obtain the benchmark price for the same product set; Upon receiving a purchase request, determine the corresponding set of similar products based on the product identifier in the purchase request; The price of the goods in the purchase request is compared with the benchmark price of the same product set to obtain the verification result; Based on the verification result, respond to the procurement request.

2. The method according to claim 1, characterized in that, The verification results include pass, warning, and block; the comparison of the commodity price in the purchase request with the benchmark price of the same commodity set to obtain the verification results includes: The verification result is passed when the price of the goods in the purchase request is less than or equal to the benchmark price. When the price of the goods in the purchase request is greater than the benchmark price but less than or equal to a first percentage of the benchmark price, the verification result is a warning. When the price of the goods in the purchase request is greater than a second percentage of the benchmark price, the verification result is blocked; the first percentage is less than the second percentage.

3. The method according to claim 1, characterized in that, The step of responding to the procurement request based on the verification result includes: The purchase request is allowed if the verification result is satisfactory. When the verification result is a warning, the approval process is triggered, and the procurement request is allowed after the approval is passed. When the verification result is blocked, the purchase request is not allowed.

4. The method according to claim 1, characterized in that, The step of performing multimodal feature fusion on the standardized commodity data to generate a comprehensive feature vector corresponding to each commodity includes: Feature extraction is performed on the text information in the standardized commodity data to obtain a text feature vector; Feature extraction is performed on the image information in the standardized commodity data to obtain image feature vectors; The text feature vector and the image feature vector are fused to obtain the comprehensive feature vector.

5. The method according to claim 1, characterized in that, The comparative price analysis of products within the same product set to obtain the benchmark price of the same product set includes: Calculate the lowest price, average price, and highest price of each item within the same product set; Based on the lowest price, average price, highest price, and performance rating, calculate the overall cost-effectiveness score for each product; The benchmark price is obtained based on the overall cost-effectiveness score.

6. The method according to claim 5, characterized in that, The calculation of the overall cost-effectiveness score for each product based on the lowest price, average price, highest price, and fulfillment rating includes: in, Indicates goods The overall cost-effectiveness score, Indicates the first weight. Indicates the second weight. Represents the transition function. Indicates goods The price Indicates supplier Performance rating The first coefficient representing the price factor, The second coefficient represents the price factor. The third coefficient representing the price factor, Indicates the lowest price. Indicates the highest price. This indicates the average price.

7. The method according to claim 5, characterized in that, The performance rating is obtained through the following methods: in, Indicates supplier Performance rating Indicates based on supplier Credit rating adjustment factor Indicates the first The weight of each performance indicator Indicates supplier In the Standardized scores on each performance indicator This indicates the total number of performance targets.

8. A price verification device based on the identification of identical products, characterized in that, The device includes: The acquisition module is used to acquire heterogeneous product data from multiple suppliers; The processing module is used to process the heterogeneous product data to obtain standardized product data; perform multimodal feature fusion on the standardized product data to generate a comprehensive feature vector corresponding to each product; calculate the similarity between products based on the comprehensive feature vector, and cluster products with similarity greater than a similarity threshold into a set of similar products; perform price comparison analysis on products within the set of similar products to obtain the benchmark price of the set of similar products. The receiving module is used to receive purchase requests and determine the corresponding set of similar products based on the product identifier in the purchase request. The comparison module is used to compare the price of the goods in the purchase request with the benchmark price of the same goods in the same category to obtain a verification result; and respond to the purchase request based on the verification result.

9. A computing device, characterized in that, Including memory and processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method as described in any one of claims 1 to 7.

Images