Electronic product purchase price analysis method, system, equipment and medium

Abstract

The invention provides an electronic product purchase price analysis method, system and device and a medium, and belongs to the technical field of artificial intelligence and supply chain management. The method comprises the following steps: obtaining a product description, analyzing the product description through a multi-level product feature extraction mechanism, and generating a structured specification feature of a product; based on the structured specification features, analyzing historical data by using a price sensitivity analysis model to determine the weight coefficient of each specification feature on the influence of the price, and adjusting the weight coefficient according to the market event and the time decay factor; a weighted similarity matching algorithm is adopted to search for selected products in a historical product library, and price information of candidate products is acquired; and performing time decay weighted aggregation on the candidate products with the price conflicts, and determining product price information based on the updated price information. According to the method, the product features are extracted in a multi-level manner, the weight of each feature is dynamically generated, and matching is performed based on the multi-dimensional weighted similarity, so that the accuracy of price analysis is effectively improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of artificial intelligence and supply chain management, and more particularly to an electronic product procurement price analysis method, system, device and medium. BACKGROUND

[0002] In the field of electronic product procurement, price analysis is a core technical means to control procurement costs and optimize resource allocation. With the acceleration of electronic product technology iteration and the increasing complexity of scenario demand, traditional price analysis methods have gradually exposed their limitations. The current mainstream solution mainly builds an analysis model based on category-level statistical data (such as the average price of large categories such as servers and storage devices), and realizes price rationality evaluation through preset parameter weights or simple semantic matching. This method has certain applicability in standardized product procurement scenarios, but in the face of the diversification of electronic product specifications, the differentiation of demand scenarios, and the shortening of product iteration cycles, its analysis accuracy and scenario adaptation ability have been difficult to meet the needs of fine-grained procurement management.

[0003] The core implementation path of the prior art solution includes two categories: one adopts a static weight configuration model, which defines the fixed weight proportion of each parameter (such as CPU frequency and memory capacity) in price evaluation by manual definition, and calculates the theoretical price based on the text matching result of the specification parameters; the other is based on natural language processing technology, which extracts technical parameters from product descriptions through semantic vectorization or keyword matching, and then conducts deviation analysis combined with statistical average price. These methods have significant defects in parameter weight allocation and semantic analysis: the static weight model cannot dynamically adjust the parameter priority according to the specific business scenario (such as AI computing and big data analysis), resulting in the price contribution of high-performance GPUs or special storage media being underestimated; the semantic analysis module lacks domain knowledge constraints and is difficult to handle the heterogeneous expression of electronic product parameter descriptions.

[0004] The above technical path leads to three major systemic defects: first, the mapping relationship between specification parameters and price is simplified as a linear weighted model, ignoring the complex influence of non-linear technology premium on price; second, the matching evaluation of scenario demand and product specifications is missing, making it difficult to realize differentiated pricing analysis for products with different technology routes in the same category; third, the compatibility of new product parameter systems is insufficient, and when manufacturers introduce new specification terms or composite parameter descriptions, existing models need to undergo a manual intervention period of several months to complete knowledge base updates, resulting in a time gap in procurement decision-making. These defects directly restrict the fine-grained degree and business adaptation ability of electronic product procurement price analysis. SUMMARY

[0005] In view of the above problems, the present application aims to provide an electronic product procurement price analysis method, system, device and medium, which extracts product features through multiple levels, dynamically generates the weight of each feature, and matches based on multi-dimensional weighted similarity, effectively improving the accuracy of price analysis.

[0006] To achieve the above-mentioned purpose, the present application realizes the following technical solutions: In a first aspect, the present application provides an electronic product procurement price analysis method, comprising: Obtaining product description, analyzing the product description through a multi-level product feature extraction mechanism, and generating the structured specification features of the product; Based on the structured specification features, using a price sensitivity analysis model to analyze historical data to determine the weight coefficient of each specification feature on the price, and adjusting the weight coefficient according to market events and time decay factors; Using a weighted similarity matching algorithm to find the selected product in the historical product library and obtain the price information of the candidate product; Performing time decay weighted aggregation on the candidate products with price conflicts, and determining the product price information based on the updated price information.

[0007] In an optional embodiment, the obtaining product description, analyzing the product description through a multi-level product feature extraction mechanism, and generating the structured specification features of the product, comprises: Obtaining the original product description, inputting the BERT model and fusing the preset electronic industry keyword library to generate the product category label; Based on the product description and the category label, the specification structure is analyzed and the specification type field is extracted to generate the specification dictionary; the specification type field includes: model, numerical parameter and interface type; Reasonably checking the specification field, and generating the structured specification features of the product after the check is passed.

[0008] In an optional embodiment, the specification structure is analyzed based on the product description and the category label, the specification type field is extracted, and the specification dictionary is generated, comprising: Based on the brand model knowledge graph, the model of the product is extracted through regular expression; The numerical parameter of the product is obtained by matching the unit keyword and combining the numerical range check; The interface type of the product is determined by using the enumeration value matching method; When there are multiple values of the extracted numerical parameter of the product, the maximum value is taken as the numerical parameter of the product; The model data of the industry official website is updated based on the extracted specification type field.

[0009] In an optional embodiment, the method of analyzing historical data based on structured specification features using a price sensitivity analysis model to determine the weight coefficient of the impact of each specification feature on price includes: Taking structured specification features as input, the price sensitivity analysis model built based on the XGBoost model is used to analyze historical procurement data. The importance score is calculated as the weight coefficient of the impact of each structured specification feature on price. The calculation formula of the importance score is as follows:

[0010] in, is the i-th specification feature The weight coefficient of is the historical price of the specification feature, N 有效样本 is the sample size after removing abnormal data, N 总样本 is the total original sample size.

[0011] In an optional embodiment, adjusting the weight coefficient according to market events and time decay factors includes: By crawling manufacturer official website information and parsing industry organization announcements, we can monitor new product releases, technical standard upgrades, and supply chain disruptions in real time. For new product release events, the initial value of the corresponding model weight coefficient will be increased by 30%; For technical standard upgrade events, increase the weight coefficient of the corresponding interface type; In the event of supply chain disruption, reduce the weight coefficient of the affected specification characteristics; According to the category labels corresponding to the specification characteristics, set the differentiated half-life and use the exponential decay function Adjust weight coefficient; in, is the decay constant, , is the weight coefficient of the current specification feature, is the weight coefficient of the specification feature that decays after time t, is the half-life.

[0012] In an optional embodiment, the method of using a weighted similarity matching algorithm to search for selected products in a historical product library and obtain price information of candidate products includes: Based on the data type of the specification type field, the model and interface type are classified as text features, and the data value parameters are classified as numeric features; For text features, use the formula Calculate the feature similarity with the corresponding features in the historical product library; where A and B are the strings corresponding to the two text features to be matched, is the edit distance between strings A and B, |A| and |B| are the lengths of strings A and B respectively, is the Tanimoto coefficient; For numerical features, use the formula Calculate the feature similarity with the corresponding features in the historical product library; where A and B are the values ​​corresponding to the two numerical features to be matched, The maximum value of this numerical feature in the historical procurement data; Based on the feature similarity of each specification feature of the product, the formula Calculate the total similarity between the product and each product in the historical product library ;in, is the feature similarity of the i-th specification feature of the product, is the weight coefficient of the i-th specification feature, It is the calculation result of the effectiveness indicator function of the i-th specification feature. When the i-th specification feature is valid , when the i-th specification feature is invalid ; The total similarities are sorted in descending order, the top-K total similarities are screened out, and the corresponding historical products are used as candidate products.

[0013] In an optional embodiment, performing time-decay weighted aggregation on the candidate products with price conflicts and determining product price information based on the updated price information includes: Determine whether there are multiple price versions of the candidate product; If not, directly output the information of all candidate products and the price range of candidate products; If yes, there is a price conflict for the candidate product. The price dispersion of the candidate product is calculated based on all price versions, and it is determined whether the price dispersion is greater than a preset threshold. If the product price dispersion is greater than the preset threshold, the median of all price versions is used as the price information of the candidate product; If the product price dispersion is not greater than a preset threshold, the weighted average of the prices is calculated according to the preset time decay weights as the price information of the candidate product.

[0014] In a second aspect, an embodiment of the present application further provides an electronic product procurement price analysis system, comprising: The feature extraction module is used to obtain product descriptions, parse the product descriptions through a multi-level product feature extraction mechanism, and generate structured specification features of the product; a weight calculation module, configured to analyze historical data by using a price sensitivity analysis model based on the structured specification features, to determine a weight coefficient of an influence of each specification feature on a price, and to adjust the weight coefficient according to a market event and a time decay factor; a candidate product retrieval module, configured to search for the selected product in a historical product library by using a weighted similarity matching algorithm, and to obtain price information of a candidate product; a conflict processing module, configured to perform time decay weighted aggregation on the candidate product with a price conflict, and to determine product price information based on updated price information.

[0015] In a third aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the program to implement the steps of the electronic product procurement price analysis method according to any one of the preceding aspects.

[0016] In a fourth aspect, a storage medium is provided, having a computer program stored thereon, and the computer program is executable by a processor to implement the steps of the electronic product procurement price analysis method according to any one of the preceding aspects.

[0017] As can be seen from the above technical solutions, the present application has the following advantages: In the electronic product procurement price analysis method provided by the present application, first, a multi-level product feature extraction mechanism is used to fuse a BERT model, an industry keyword library, and a knowledge graph, etc., to accurately analyze product descriptions and generate structured specification features, thereby improving the accuracy and structural degree of feature analysis. Then, a price sensitivity analysis model is constructed based on an XGBoost model, and a weight coefficient is dynamically adjusted in combination with market events (such as new product release, technology standard upgrade, supply chain interruption, etc.) and a time decay factor, so that the weight is more suitable for real-time market changes and has timeliness. Further, a weighted similarity matching algorithm is used, and a differentiated similarity calculation method is designed for text and numerical features, and candidate products are selected in combination with the weight and an effectiveness indicator function, to achieve accurate matching. Finally, time decay weighted aggregation is performed on the candidate products with a price conflict, and a median or a weighted average value is flexibly selected according to the price dispersion to determine the price information, to ensure the reliability of the price information. Overall, this method can improve the efficiency, accuracy, and real-time performance of procurement price analysis, help scientific decision-making, reduce costs, and respond to market dynamics.

[0018] By using the multi-level product feature extraction mechanism, the BERT model, and the electronic industry keyword library, the present application can more accurately analyze product descriptions and generate structured specification features, providing a reliable basis for subsequent price analysis, and helping to more accurately grasp product characteristics.

[0019] The application utilizes a price sensitivity analysis model based on an XGBoost model, combines historical procurement data, scientifically calculates the weight coefficient of the influence of each specification feature on the price, and dynamically adjusts through market events and time decay factors, so that the price analysis is more timely and accurate.

[0020] The application adopts a weighted similarity matching algorithm, designs a similarity calculation formula for different types of features (text type, numerical type), and combines the specification feature weight and effectiveness indication function to more accurately filter out candidate products from the historical product library, improving matching efficiency and accuracy.

[0021] The application performs time decay weighted aggregation on the candidate products with price conflicts, reasonably selects the median or weighted average value as the price information according to the size of product price dispersion, considers the historical stability of the price, and also takes into account market changes, so that the price information is more reasonable and has more reference value.

[0022] The application forms a complete and practical electronic product procurement price analysis system through a series of scientific and rigorous steps from product specification analysis to price information determination, which can adapt to different market environments and product characteristics and provide strong decision support for electronic product procurement. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the present application, the drawings needed to be used in the description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0024] Figure 1 The flowchart of the electronic product procurement price analysis method provided by the present application.

[0025] Figure 2 The structure diagram of the electronic product procurement price analysis system provided by the present application.

[0026] Figure 3 The structure diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0027] In the following detailed description of the specific steps of the electronic product procurement price analysis method, various embodiments of the present disclosure will be described more fully. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents and / or alternatives falling within the spirit and scope of various embodiments of the present disclosure.

[0028] Hereinafter, the term "include" or "may include" used in various embodiments of the disclosure indicates the presence of the disclosed functions, operations, or elements and does not limit one or more functions, operations, or elements from being added. Also, as used in various embodiments of the disclosure, the terms "include", "have", and their conjugates merely indicate that specific features, numbers, steps, operations, elements, components, or combinations thereof are present and do not exclude the presence or addition of one or more other features, numbers, steps, operations, elements, components, or combinations thereof.

[0029] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0030] Please refer to Figure 1 The method flowchart of the electronic product procurement price analysis method in an embodiment is shown, and the method comprises: S1: Obtain product description, parse the product description through a multi-level product feature extraction mechanism, and generate the structured specification features of the product.

[0031] In the specific implementation, first, the original product description is obtained, input into the BERT model, and the preset electronic industry keyword library is fused to generate the product category label.

[0032] Then, based on the product description and the category label, the specification structure is parsed, the specification type field is extracted, and the specification dictionary is generated; the specification type field includes: model number, numerical parameter, and interface type. Specifically, the numerical parameter of the product is obtained through unit keyword matching combined with numerical range verification; the interface type of the product is determined by enumeration value matching; when there are multiple values of the numerical parameter of the product, the maximum value is taken as the numerical parameter of the product; and the model number data of the official website of the industry is updated based on the extracted specification type field.

[0033] Finally, the specification field is reasonably checked, and after the check is passed, the structured specification features of the product are generated.

[0034] For example, this step specifically includes the following three-level processing flow, which is as follows: First level: rough classification of category.

[0035] Take the product original description (such as "CPU: Tengyun S2500 processor (64C, 2.1GHz) *2 DDR4 128G (32G*4)") as input, use BERT model and integrate electronic industry keyword library (for example, "2U" corresponds to server), complete the output of category labels such as server, GPU server, computing device, storage device, etc., so as to improve the accuracy of fuzzy description classification.

[0036] This method is mainly based on BERT model + electronic industry keyword library. This method integrates industry keyword library (such as "2U"→server), which can well improve the accuracy of fuzzy description classification, and can solve the misjudgment problem of some traditional abbreviations (such as "A800") and different writing methods of model (such as "G_CPU_INTEL_Gold-5118*2" and "G_CPU_INTEL_Gold-5118*2").

[0037] Second level: specification structure analysis.

[0038] Use regular expressions combined with brand model knowledge graph to extract model (such as "NVIDIA H100 80GB" to "H100"); obtain numerical parameters by matching unit keywords combined with numerical range verification (such as "128GB memory" to "memory: 128"); determine the interface type by enumeration value matching (such as "PCIE3.0*1" to "PCIe3.0"). When there is a conflict, such as multiple values for the same specification, automatically select according to the preset strategy (such as prefer the maximum value), and update the industry official model data regularly. Based on the extracted specification type field, generate a specification dictionary.

[0039] Third level: context association verification.

[0040] Reasonably check the specification dictionary obtained by preliminary analysis. If there is a specification combination that does not conform to the hardware technical rules (such as NF8260M5 with GPU card), an error is prompted.

[0041] S2: Based on the structured specification features, use the price sensitivity analysis model to analyze historical data to determine the weight coefficient of each specification feature on the price, and adjust the weight coefficient according to market events and time decay factors.

[0042] In a specific embodiment, the structured specification features are used as input, and the price sensitivity analysis model based on XGBoost model is used to analyze historical procurement data, and the importance score is calculated as the weight coefficient of each structured specification feature on the price. The calculation formula of the importance score is as follows:

[0043] wherein, is the weight coefficient of the i-th specification feature, is the weight coefficient of the i-th specification feature, is the historical price of the specification feature, N 有效样本 is the sample size after removing abnormal data, N 总样本 is the total amount of original samples.

[0044] Then, by crawling the manufacturer's official website information and parsing industry organization announcements, real-time monitoring of new product release events, technical standard upgrade events, and supply chain interruption events is performed. Specifically, for new product release events, the initial value of the corresponding model weight coefficient is increased by 30%; for technical standard upgrade events, the weight coefficient of the corresponding interface type is increased; and for supply chain interruption events, the weight coefficient of the affected specification feature is reduced.

[0045] Finally, according to the category label corresponding to the specification feature, a differentiated half-life is set, and an exponential decay function is used to adjust the weight coefficient; wherein, is the decay constant, , is the current weight coefficient of the specification feature, is the weight coefficient of the specification feature after decay at time t, is the half-life.

[0046] For example, first, the structured specification features are taken as input, and the XGBoost model is used to analyze historical procurement data to calculate the influence coefficient of each specification on the price. At the same time, a sample validity factor is introduced to exclude the interference of abnormal quotes and promotional data, and it is automatically updated every quarter.

[0047] Then, by crawling the manufacturer's official website information and parsing industry organization announcements (such as PCI-SIG announcements), real-time monitoring of new product release, technical standard upgrade (such as PCIe5.0), supply chain interruption, and other market events is performed. For different events, appropriate weight adjustment strategies are adopted, such as resetting the weight of related models (initial value increased by 30%) when new products are released, increasing the weight of related interface types when technical standards are upgraded, and reducing the weight of affected components when supply chains are interrupted.

[0048] Finally, according to the difference of hardware categories, a differentiated half-life is set (such as GPU card half-life of 6 months, server of 12 months), and an exponential decay function is used to adjust the weight of historical data. Data exceeding 3 half-lives will be automatically eliminated.

[0049] S3: Adopt a weighted similarity matching algorithm to find the selected product in the historical product library and obtain the price information of the candidate product. ​

[0050] In the specific implementation, first, based on the data type of the specification type field, the model number and the interface type are classified as text type features, and the data value parameter is classified as a numerical type feature.

[0051] Then, the similarity calculation method as follows is used for the two types of features: For the text type feature, the formula is used to calculate the feature similarity with the corresponding feature in the historical product library; wherein A and B are the strings corresponding to the two text type features to be matched, is the edit distance of the strings A and B, and |A| and |B| are the lengths of the strings A and B, is the Tanimoto coefficient; For the numerical type feature, the formula is used to calculate the feature similarity with the corresponding feature in the historical product library; wherein A and B are the values corresponding to the two numerical type features to be matched, is the maximum value of the numerical type feature in the historical purchase data.

[0052] As can be seen, in the above process, for the text type feature (such as the model number), the composite calculation method of the Levenshtein distance multiplied by the Tanimoto coefficient is used, and a manufacturer alias mapping table is constructed to perform phonetic transcription similarity matching on non-English model numbers; for the numerical type feature (such as the memory capacity), the log scaling or normalized difference value processing is first performed At this time, the specification weight output by the dynamic weight generation module is combined with the validity indication function Ivalid(i) to calculate the total similarity through the weighted aggregation formula, and the Top-K candidate products are found in the historical product library according to the similarity.

[0053] For example, based on the feature similarity of each specification feature of the product, the total similarity of the product with each product in the historical product library is calculated through the formula ; wherein is the feature similarity of the i-th specification feature of the product, is the weight coefficient of the i-th specification feature, is the calculation result of the i-th specification feature through the validity indication function, when the i-th specification feature is valid , and when the i-th specification feature is invalid . ; Finally, the total similarity is arranged in descending order, and the Top-K total similarities are screened out, and the corresponding historical products are taken as candidate products.

[0054] S4: Perform time-decay weighted aggregation on the candidate products with price conflicts, and determine product price information based on the updated price information.

[0055] In this step, first determine whether there are multiple price versions of the candidate product; if not, directly output the information of all candidate products and the price range of the candidate products; if so, there is a price conflict for the candidate product, calculate the product price dispersion of the candidate product based on all price versions, and determine whether the product price dispersion is greater than the preset threshold.

[0056] At this time, if the product price dispersion is greater than the preset threshold, the median of all price versions is used as the price information of the candidate product; if the product price dispersion is not greater than the preset threshold, the weighted average of the prices is calculated according to the preset time decay weight as the price information of the candidate product.

[0057] For example, we perform weighted aggregation on candidate product price data based on a pre-set time decay factor, prioritizing recent data. Furthermore, when multiple price variations exist for the same historical product, we first calculate price dispersion. If the dispersion exceeds 30%, the data is considered promotional and the median price is taken. If the dispersion is less than 30%, a weighted average is calculated using the time decay weight to minimize the impact of extreme values ​​on price analysis.

[0058] In this embodiment, through multi-level feature extraction and dynamic weight adjustment mechanism, combined with market event perception and time decay strategy, accurate structured analysis of electronic product specifications and price sensitivity modeling are achieved. At the same time, weighted similarity matching and time decay aggregation algorithms are used to effectively solve the problem of historical price data conflicts, significantly improving the accuracy, timeliness and market adaptability of procurement price analysis. It can dynamically capture the impact of market variables such as technology iteration and supply chain fluctuations on prices, provide enterprises with scientific and reasonable procurement decision-making basis, and reduce procurement cost risks.

[0059] like Figure 2 As shown, the following is an embodiment of the electronic product procurement price analysis system provided by the embodiment of the present disclosure. The system and the electronic product procurement price analysis method of the above-mentioned embodiments belong to the same inventive concept. For details not fully described in the embodiment of the electronic product procurement price analysis system, please refer to the embodiment of the above-mentioned electronic product procurement price analysis method.

[0060] An electronic product purchasing price analysis system includes: a feature extraction module, a weight calculation module, a candidate product retrieval module and a conflict processing module.

[0061] The feature extraction module is used to obtain product descriptions, parse product descriptions through a multi-level product feature extraction mechanism, and generate structured specification features of the product.

[0062] The weight calculation module is configured to analyze historical data based on the structured specification features by using a price sensitivity analysis model to determine a weight coefficient of the influence of each specification feature on the price, and adjust the weight coefficient according to market events and a time decay factor.

[0063] The candidate product retrieval module is configured to search for the selected product in a historical product database by using a weighted similarity matching algorithm, and obtain price information of the candidate product.

[0064] The conflict processing module is configured to perform time decay weighted aggregation on the candidate product with price conflict, and determine product price information based on the updated price information.

[0065] The electronic product procurement price analysis system provided by the embodiment realizes accurate analysis of the procurement price of the electronic product by multi-level product feature extraction, scientific price sensitivity analysis, optimized weighted similarity matching algorithm, and time decay weighted aggregation processing on the candidate product with price conflict, improves the accuracy of product specification analysis and price sensitivity analysis, optimizes the similarity matching effect, makes the price information processing more reasonable, provides a comprehensive, scientific and practical decision basis for electronic product procurement, and significantly enhances the accuracy and efficiency of procurement decision.

[0066] Figure 3 A hardware structure schematic diagram of an electronic device for implementing various embodiments of the present application.

[0067] The electronic product procurement price analysis method provided by the embodiment of the present application can be applied to an electronic device. Those skilled in the art can understand that the electronic device structure involved in the embodiment of the present application does not constitute a limitation on the electronic device, and the electronic device can include more or fewer components than the illustration, or combine certain components, or different component arrangements. In the embodiment of the present application, the electronic device includes but is not limited to a laptop computer, a desktop computer, a workstation, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present application described and / or claimed herein.

[0068] The electronic device can include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, a wireless communication module, an audio module, a speaker, a microphone, a sensor module, a key, a camera, a display screen, and a SIM card interface, and the like.

[0069] The processor can include one or more processing units, such as: the processor can include a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), and the like. Among them, different processing units can be independent devices, or can be integrated in one or more processors.

[0070] Among them, the processor can be the nerve center and command center of the electronic device. The controller can generate operation control signals according to instruction operation codes and timing signals, and complete the control of fetching and executing instructions.

[0071] The memory can also be provided in the processor, used to store instructions and data. In some embodiments, the memory in the processor is a cache memory. The memory can save instructions or data that the processor has just used or repeatedly uses. If the processor needs to use the instructions or data again, it can directly call from the memory. Avoiding repeated access, reducing the waiting time of the processor, thus improving the efficiency of the system.

[0072] The external memory interface can be used to connect an external memory card, such as a MicroSD card, to realize the expansion of the storage capacity of the electronic device. The external memory card communicates with the processor through the external memory interface to realize the data storage function. For example, save music, video and other files in the external memory card.

[0073] The internal memory can be used to store computer executable program codes including instructions. The processor performs various function applications and data processing of the electronic device by running the instructions stored in the internal memory. The internal memory can include a program storage area and a data storage area. The internal memory can include a high-speed random access memory, and can further include a non-volatile memory such as at least one of a magnetic disk storage device, a flash memory device, a universal flash storage (UFS), etc.

[0074] The wireless communication function of the electronic device can be implemented through an antenna, a wireless communication module, a modem processor, and a baseband processor, etc.

[0075] The wireless communication module can provide a wireless communication solution including wireless local area networks (WLAN) (e.g., wireless fidelity (Wi-Fi) network), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc. applied to the electronic device.

[0076] The electronic device can implement an audio function, etc. through an audio module, a speaker, a receiver, a microphone, a headphone interface, and an application processor, etc.

[0077] The electronic device can implement a photographing function through an ISP, a camera, a video codec, a GPU, a display screen, and an application processor, etc.

[0078] The electronic device can implement a display function through a GPU, a display screen, and an application processor, etc.

[0079] The GPU is a microprocessor for image processing, which is connected to the display screen and the application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. The processor can include one or more GPUs, which execute program instructions to generate or change display information.

[0080] The display screen is used to display images, videos, etc. The display screen includes a display panel.

[0081] The electronic device realizes the electronic product procurement price analysis method, and has the beneficial effects of accurately analyzing the electronic product procurement price and improving the scientificity and accuracy of procurement decision-making.

[0082] The storage medium provided in the present application stores a program product capable of realizing the electronic product procurement price analysis method.

[0083] The electronic product procurement price analysis method comprises: Obtaining a product description, analyzing the product description through a multi-level product feature extraction mechanism, and generating structured specification features of the product; Based on the structured specification features, analyzing historical data using a price sensitivity analysis model to determine the weight coefficient of each specification feature on the price influence, and adjusting the weight coefficient according to market events and time decay factors; Using a weighted similarity matching algorithm, finding the selected product in the historical product library, and obtaining the price information of the candidate product; Performing time decay weighted aggregation on the price conflict candidate product, and determining the product price information based on the updated price information.

[0084] In some possible implementation manners, the electronic product procurement price analysis method of the present application can be realized in the form of a program product, which includes program codes for causing a terminal device to execute the steps according to various exemplary embodiments of the present application described in the above “Exemplary Method” section of the present specification when the program product is running on the terminal device.

[0085] The storage medium of the present application can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium may, for example, be but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0086] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the appended claims are intended to cover all such modifications that do not depart from the true spirit and scope of the application. Therefore, the application is not limited to the embodiments shown but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An electronic product purchase price analysis method characterized by comprising: The method comprises the following steps: Obtaining product description, analyzing product description through multi-level product feature extraction mechanism, and generating structured specification features of the product; Based on the structured specification features, using the price sensitivity analysis model to analyze the historical data to determine the weight coefficient of each specification feature on the price influence, and adjusting the weight coefficient according to market events and time decay factor; Using weighted similarity matching algorithm to find the selected product in the historical product library and obtain the price information of the candidate product; Performing time decay weighted aggregation on the price conflict candidate product, and determining the product price information based on the updated price information.

2. The method of claim 1, wherein The method comprises the following steps: Obtaining the original product description, inputting the BERT model and fusing the preset electronic industry keyword library to generate the product category label; Based on the product description and the category label, the specification structure is analyzed, the specification type field is extracted, and the specification dictionary is generated; The specification type field includes: model, numerical parameter and interface type; Reasonable check is performed on the specification field, and the structured specification features of the product are generated after the check is passed.

3. The method of claim 2, wherein The method comprises the following steps: Based on the brand model knowledge graph, the model of the product is extracted by regular expression; The numerical parameter of the product is obtained by matching the unit keyword and checking the numerical range; The interface type of the product is determined by enumeration value matching; When there are multiple values of the numerical parameter of the product, the maximum value is taken as the numerical parameter of the product; The model data of the industry official website is updated regularly based on the extracted specification type field.

4. The method of claim 2, wherein The method comprises the following steps: The structured specification features are used as input, and the price sensitivity analysis model based on XGBoost model is used to analyze the historical purchase data, and the importance score is calculated as the weight coefficient of each structured specification feature on the price influence; The calculation formula of the importance score is as follows: wherein, is the weight coefficient of the i-th specification feature, is the historical price of the specification feature, N 有效样本 is the sample size after removing abnormal data, N 总样本 is the total amount of original samples.​ 5. The method of claim 4, wherein The method comprises the following steps: By crawling the information of the manufacturer's official website and analyzing the industry organization announcement, the new product release event, the technical standard upgrading event and the supply chain interruption event are monitored in real time; For the new product release event, the initial value of the weight coefficient of the corresponding model is increased by 30%; For the technical standard upgrading event, the weight coefficient of the corresponding interface type is increased; For the supply chain interruption event, the weight coefficient of the affected specification feature is reduced; According to the category label corresponding to the specification feature, a differentiated half-life is set, and an exponential decay function is used Adjust the weight coefficient; wherein, is a decay constant, , is a weight coefficient of the current specification feature, is a weight coefficient of the decayed specification feature at time t, is a half-life.

6. The electronic product purchase price analysis method according to claim 5, wherein The method comprises the following steps: Based on the data type of the specification type field, the model and the interface type are classified as text features, and the data value parameter is classified as numerical features; For text type features, the formula is used to calculate the feature similarity with the corresponding features in the historical product library; wherein A and B are the strings corresponding to the two text type features to be matched, is the edit distance of the strings A and B, and |A| and |B| are the lengths of the strings A and B, is the Tanimoto coefficient; For numerical features, the formula is used to calculate the feature similarity with the corresponding features in the historical product library; wherein A and B are the values corresponding to the two numerical features to be matched, is the maximum value of the numerical feature in the historical procurement data. Based on the feature similarity of each specification feature of the product, the total similarity of the product and each product in the historical product library is calculated by the formula ; wherein, is the feature similarity of the i-th specification feature of the product, is the weight coefficient of the i-th specification feature, is the calculation result of the i-th specification feature through the effectiveness indication function, when the i-th specification feature is valid , when the i-th specification feature is invalid ; and . The total similarity is arranged in descending order, and the Top-K total similarity is screened out, and the corresponding historical product is taken as the candidate product.

7. The electronic product purchase price analysis method according to claim 6, wherein The candidate product with price conflict is executed time-decay weighted aggregation, product price information is determined based on updated price information, including: determining whether the candidate product has multiple price versions; if not, directly outputting information of all candidate products and price interval of the candidate product; if yes, the candidate product has price conflict, product price dispersion of the candidate product is calculated based on all price versions, and it is determined whether the product price dispersion is greater than a preset threshold; if the product price dispersion is greater than the preset threshold, the median of all price versions is taken as the price information of the candidate product; if the product price dispersion is not greater than the preset threshold, a weighted average value of the price is calculated as the price information of the candidate product according to a preset time-decay weight.

8. An electronic product purchasing price analysis system, characterized in that: The system adopts the electronic product purchase price analysis method according to any one of claims 1 to 7; The system comprises: a feature extraction module, configured to acquire product description, analyze the product description through a multi-level product feature extraction mechanism, and generate structured specification features of the product; a weight calculation module, configured to analyze historical data by using a price sensitivity analysis model based on the structured specification features, to determine a weight coefficient of influence of each specification feature on the price, and to adjust the weight coefficient according to market events and a time-decay factor; a candidate product retrieval module, configured to find the selected product in a historical product library by using a weighted similarity matching algorithm, and to acquire price information of the candidate product; a conflict processing module, configured to execute time-decay weighted aggregation on the candidate product with price conflict, and to determine product price information based on updated price information.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps of the electronic product purchase price analysis method according to any one of claims 1 to 7.

10. A storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to implement the steps of the electronic product purchase price analysis method according to any one of claims 1 to 7.

Images