Commodity monitoring system based on big data

Abstract

The invention relates to the technical field of data monitoring management, in particular to a commodity monitoring system based on big data, which comprises a data acquisition module for forming a multi-dimensional data set of each commodity; the space-time analysis module is used for analyzing the circulation data of each commodity and determining the space-time attribute of each commodity; the propagation identification module is used for determining time sensitivity coefficients and circulation sensitivity coefficients of propagation information and commodity categories; the association identification module is used for determining an associated commodity set of each commodity based on the circulation data of each commodity after space-time division; and the circulation monitoring module is used for determining a circulation change coefficient of the corresponding circulation change commodity based on an analysis result of the propagation information, determining circulation change parameters of the circulation change commodity and the corresponding associated commodity according to the circulation change coefficient, and forming a purchase / inventory adjustment message. According to the invention, the inventory turnover rate and the matching rate of commodity sales, consumption and demand can be improved.

Description

Technical Field

[0001] This invention relates to the field of data monitoring and management technology, and in particular to a commodity monitoring system based on big data. Background Technology

[0002] In traditional logistics and distribution, tracking goods relies on intermediaries or third-party platforms. Logistics data undergoes multiple forwarding and processing cycles, leading to increased inaccuracy and reduced tracking timeliness. The isolated storage and lack of standardized procedures for logistics information hinder information sharing among different participants, resulting in a lack of overall visibility and transparency. Traditional logistics tracking is also vulnerable to data tampering and unauthorized access, raising concerns about information security and trustworthiness. With the continuous development of blockchain technology, it can provide a higher level of visibility and transparency in logistics tracking. Blockchain's decentralized nature eliminates dependence on a single intermediary, enabling direct communication and data sharing among participants at each stage. By recording logistics data in real time, blockchain ensures data timeliness and accuracy. Participants can update and view logistics data in real time through the blockchain network, achieving real-time monitoring and traceability throughout the logistics process, reducing information delays and errors.

[0003] For e-commerce logistics platforms, although the traceability of blockchain technology can record data such as date, location, and transportation method of goods in chronological order at each stage from outbound, transportation, warehousing to last-mile delivery, the production and preparation cycle of goods often cannot be accurately matched with the demand cycle of the market due to the influence of procurement time and market feedback. This can easily lead to delayed feedback and reduce the timeliness of obtaining real-time sales and consumption information of goods.

[0004] Therefore, it is evident that the existing technology has the following problems: The inability to quickly match production preparation with changes in market demand leads to insufficient or excessive product inventory. Summary of the Invention

[0005] To address this issue, the present invention provides a big data-based commodity monitoring system to overcome the problem in existing technologies where the inability to quickly match commodity production preparation with changes in market demand leads to insufficient or excessive product inventory.

[0006] To achieve the above objectives, the present invention provides a big data-based commodity monitoring system, comprising: The data acquisition module is used to collect the circulation data of each product to form a multidimensional dataset for each product. The spatiotemporal analysis module is used to determine the spatiotemporal attributes of each commodity, including time attributes and spatial attributes, based on the analysis of the circulation data of each commodity. The propagation identification module is used to acquire some propagation information, cluster them according to the category of each product, and determine the time sensitivity coefficient and circulation sensitivity coefficient of the propagation information and the product category based on the correspondence between the propagation information and the corresponding product circulation data. The association identification module is used to determine the spatiotemporal division of the circulation data of each product based on the spatiotemporal attributes of each product, and to determine the associated product set of each product based on the circulation data of each product after the spatiotemporal division. The circulation monitoring module is used to trigger the circulation analysis of goods based on changes in the dissemination information collected in the current period, determine the circulation change coefficient of the corresponding circulation change goods based on the analysis results of the dissemination information, and determine the circulation change parameters of the circulation change goods and the corresponding related goods based on the circulation change coefficient and generate procurement / inventory adjustment messages.

[0007] Furthermore, the circulation data includes daily consumption, consumption region, and the number of target objects corresponding to the consumption region, collected in time sequence.

[0008] Furthermore, the circulation change parameters include the purchase quantity adjustment ratio, inventory allocation ratio, related product set, and circulation transfer parameters corresponding to each product in the related product set.

[0009] Furthermore, the spatiotemporal analysis module determines the longitudinal fluctuation ratio based on the ratio of the daily consumption of a product to the number of target objects within a preset period in the circulation data of a single product, and determines the time attribute of a single product based on the longitudinal fluctuation ratio, wherein the fluctuation ratio is the ratio of the maximum value to the minimum value of the ratio of the daily consumption of a product to the number of target objects within a preset period. If the longitudinal fluctuation ratio is greater than or equal to the preset longitudinal ratio threshold, the spatiotemporal analysis module determines that the time attribute of a single product is seasonal. If the longitudinal fluctuation ratio is less than the preset longitudinal ratio threshold, the spatiotemporal analysis module determines that the time attribute of a single product is stable.

[0010] Furthermore, the spatiotemporal analysis module determines the spatial attributes of a single product based on the horizontal fluctuation ratio of the average value of several partition data determined according to a preset spatial division in the circulation data of a single product. If the lateral fluctuation ratio is greater than or equal to a preset lateral ratio threshold, the spatial attribute of a single product is determined to be local. If the horizontal fluctuation ratio is less than the preset horizontal ratio threshold, the spatial attribute of a single product is determined to be global.

[0011] Furthermore, the dissemination identification module determines several hot dissemination keywords based on marketing information and current affairs information, so as to correspond to at least one target product and the corresponding product category based on the hot dissemination keywords; The dissemination identification module is also used to determine the first circulation coefficient of the target product based on marketing information, current affairs information, and the spatial attributes of the target product.

[0012] Furthermore, the dissemination identification module constructs a dissemination sensitivity model based on the dissemination hot keywords determined from marketing information and current affairs information in historical data and the circulation data of each product, so as to identify at least one dissemination target product and the corresponding category of the product based on the dissemination hot keywords.

[0013] Furthermore, the association identification module divides each product into several product type sets based on its spatiotemporal attributes. It then performs correlation analysis on the circulation data of each product within a single product type set to determine the associated product sets of a single product within its corresponding product type set. If the correlation coefficient of the circulation data of two products is greater than the preset correlation coefficient, one of the products is determined to be a related product of the other product.

[0014] Furthermore, the propagation identification module determines the time sensitivity coefficient and circulation sensitivity coefficient of a single product based on the propagation sensitivity model; Among them, the time sensitivity coefficient is related to the peak time of information dissemination and the peak time of daily consumption of goods, while the circulation sensitivity coefficient is related to the dissemination range and volume of the information dissemination and the change in total consumption of goods.

[0015] Furthermore, the circulation monitoring module determines whether to trigger the circulation analysis of goods based on the rate of change of the propagation information and the rate of change of the propagation range collected in the current period, and calculates the circulation change coefficient based on the rate of change of the propagation information and the rate of change of the propagation range, the time sensitivity coefficient and the circulation sensitivity coefficient of the propagation sensitivity model. The flow change coefficient is calculated based on the corrected time sensitivity coefficient and the corrected flow sensitivity coefficient.

[0016] Furthermore, it also includes big data-based product monitoring methods, including: Step S1: Collect the circulation data of each product in chronological order to form a multidimensional dataset for each product; The flow data includes daily consumption, consumption region, and the number of target objects corresponding to the consumption region, collected in time sequence. Step S2: Analyze the multidimensional dataset of each product to determine the spatiotemporal attributes of each product, including time attributes and spatial attributes; Step S3: Obtain several pieces of dissemination information, cluster them according to the category of each product, and determine the time sensitivity coefficient and circulation sensitivity coefficient between the dissemination information and the product category based on the correspondence between the dissemination information and the corresponding product circulation data. Step S4: Determine the spatiotemporal division of the circulation data of each commodity based on the spatiotemporal attributes of each commodity; Step S5: Determine the associated product set for each product based on the circulation data of each product after the spatiotemporal division; Step S6: Trigger commodity circulation analysis based on changes in the dissemination information collected within the current period; Step S7: Determine the corresponding flow transformation coefficient of the flow transformation commodity based on the analysis results of the dissemination information; Step S8: Determine the flow change parameters of the flow change products and their corresponding related products based on the flow change coefficient, and generate a purchase / inventory adjustment message.

[0017] Compared with existing technologies, the beneficial effects of this invention are as follows: the big data-based commodity monitoring system proposed in this invention achieves intelligent dynamic monitoring and prediction of commodity circulation through multi-module collaboration. Through the coordinated operation of data acquisition, spatiotemporal analysis, propagation identification, correlation identification, and circulation monitoring modules, a closed-loop process from commodity circulation data collection to procurement / inventory adjustment is constructed. The data acquisition module provides basic data support for the entire system, ensuring a reliable data source for subsequent analysis; the spatiotemporal analysis module accurately classifies the spatiotemporal attributes of commodities, making commodity management more targeted; the propagation identification module effectively captures the correlation between propagation information and commodities, providing a reference for external influencing factors in commodity circulation analysis; the correlation identification module mines the correlations between commodities, facilitating coordinated management; and the circulation monitoring module dynamically adjusts procurement and inventory strategies based on various analysis results. This multi-module collaborative model enables dynamic monitoring and intelligent control of commodity circulation, breaking down the information silos in each link of the traditional supply chain. It allows the supply chain to respond quickly to market changes, adjusting in a timely manner to sudden surges in demand caused by trending topics or seasonal fluctuations in demand. This significantly improves the supply chain's responsiveness to market changes and reduces the risk of inventory backlog or stockouts caused by information lag.

[0018] In particular, by using the spatiotemporal analysis module to calculate the longitudinal fluctuation ratio based on the ratio of daily consumption of goods to the number of target objects within a preset period, and combining this with a preset longitudinal ratio threshold to determine the time attributes of goods, the module achieves precise differentiation between seasonal and stable characteristics of goods, objectively quantifying the degree of sales fluctuation. This precise attribute classification allows companies to develop differentiated strategies for goods with different time attributes: for seasonal goods, they can prepare inventory in advance before the peak season and reduce inventory during the off-season; for stable goods, they can maintain a relatively stable inventory level. This effectively improves inventory turnover efficiency and reduces capital occupation costs. By dividing space by prefecture-level city administrative regions, calculating the horizontal fluctuation ratio, and combining this with a preset threshold to determine the spatial attributes of goods, the module accurately distinguishes between local and global goods, quantifying the sales differences of goods in different regions. For local goods, companies can concentrate resources on promotion and inventory layout in their best-selling areas, improving resource utilization efficiency; for global goods, they can develop nationwide sales and inventory strategies to ensure supply balance in various regions. This precise spatial classification allows businesses to clearly understand the regional sales characteristics of their products, optimize resource allocation, and significantly improve market coverage and sales performance in various regions.

[0019] In particular, by constructing a communication sensitivity model through a communication identification module, and combining correlation rules based on frequency correlation coefficients and geographical overlap, precise matching of communication hotspots with target products and categories is achieved, effectively capturing the intrinsic link between communication and product circulation. With the help of this communication sensitivity model, companies can quickly identify products and categories affected by communication, anticipate changes in market demand, and provide strong support for developing targeted marketing and supply chain strategies. This allows companies to seize opportunities in market competition, adjust their business strategies in a timely manner, and significantly enhance their market competitiveness.

[0020] In particular, by classifying product categories according to their spatiotemporal attributes through the association identification module and determining related product sets based on correlation analysis of circulation data, the accuracy and relevance of related product identification are ensured. Classification by spatiotemporal attributes makes the association analysis more practical; preset thresholds for different correlation coefficients accommodate the characteristics of different product types. The identified related product sets provide a basis for joint marketing, such as bundling sunscreen with sunglasses; they also support bundled inventory management, ensuring that the inventory levels of related products match, preventing stockouts of one product from affecting the sales of other related products, and significantly improving overall operational efficiency.

[0021] In particular, by calculating time sensitivity and circulation sensitivity coefficients through the dissemination identification module, the correlation between disseminated information and commodity circulation is quantified from two dimensions: time matching degree and impact intensity. The time sensitivity coefficient is derived from the overlap between the peak time of dissemination popularity and the peak time of commodity sales growth rate, reflecting the degree of time matching between dissemination and demand. The circulation sensitivity coefficient is calculated by multiplying the dissemination range index, dissemination volume index, and commodity consumption change index, reflecting the intensity of the dissemination's impact on commodity circulation. These two coefficients provide quantitative indicators for evaluating dissemination effectiveness and key basis for predicting changes in commodity demand, enabling enterprises to adjust inventory and procurement plans in advance based on dissemination characteristics, significantly improving the foresight and flexibility of the supply chain.

[0022] In particular, the circulation monitoring module triggers circulation analysis based on the rate of change in propagation and the rate of change in scope, corrects the time and circulation sensitivity coefficients, and calculates the circulation change coefficient to determine the circulation change parameters, thus achieving dynamic adjustment of commodity circulation. A 24-hour monitoring cycle ensures timely response to changes in propagation; when the rate of change in propagation or the rate of change in scope exceeds a preset threshold, analysis is immediately triggered, and the corrected coefficients better reflect the current propagation dynamics; the circulation change coefficient comprehensively measures the degree of impact, and the procurement and inventory adjustment ratios provide specific operational standards. This dynamic adjustment mechanism effectively balances inventory costs and stockout risks, avoiding both the capital tied up and losses caused by excessive inventory and the sales losses caused by stockouts, significantly improving the overall efficiency of the supply chain. Attached Figure Description

[0023] Figure 1 This is a module connection diagram of a commodity monitoring system based on big data, as described in an embodiment of the present invention. Figure 2 This is a module interaction logic diagram of a commodity monitoring system based on big data, as described in an embodiment of the present invention. Figure 3 This is a flowchart illustrating the main process of a big data-based commodity monitoring method according to an embodiment of the present invention. Figure 4 This invention provides a structural block diagram showing how a set of commodity types is formed by dividing each commodity into several commodity type sets based on its spatiotemporal attributes. Detailed Implementation

[0024] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0025] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0026] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0027] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0028] Please see Figures 1-4 As shown, this embodiment of the invention provides a product monitoring system based on big data, including: The data acquisition module is used to collect the circulation data of each product to form a multidimensional dataset for each product. The spatiotemporal analysis module is used to determine the spatiotemporal attributes of each commodity, including time attributes and spatial attributes, based on the analysis of the circulation data of each commodity. The propagation identification module is used to acquire some propagation information, cluster them according to the category of each product, and determine the time sensitivity coefficient and circulation sensitivity coefficient of the propagation information and the product category based on the correspondence between the propagation information and the corresponding product circulation data. The association identification module is used to determine the spatiotemporal division of the circulation data of each product based on the spatiotemporal attributes of each product, and to determine the associated product set of each product based on the circulation data of each product after the spatiotemporal division. The circulation monitoring module is used to trigger the circulation analysis of goods based on changes in the dissemination information collected in the current period, determine the circulation change coefficient of the corresponding circulation change goods based on the analysis results of the dissemination information, and determine the circulation change parameters of the circulation change goods and the corresponding related goods based on the circulation change coefficient and generate procurement / inventory adjustment messages.

[0029] In this embodiment, a closed-loop system covering the entire process from commodity circulation data collection to procurement / inventory adjustment is constructed through the collaborative operation of a data acquisition module, a spatiotemporal analysis module, a propagation identification module, a correlation identification module, and a circulation monitoring module. The data acquisition module provides basic data support for the entire system, ensuring a reliable data source for subsequent analysis; the spatiotemporal analysis module accurately classifies the spatiotemporal attributes of commodities, making commodity management more targeted; the propagation identification module effectively captures the correlation between propagation information and commodities, providing a reference for external influencing factors in commodity circulation analysis; the correlation identification module mines the relationships between commodities, facilitating coordinated management; and the circulation monitoring module dynamically adjusts procurement and inventory strategies based on various analysis results. This multi-module collaborative model achieves dynamic monitoring and intelligent control of commodity circulation, breaking down information silos in traditional supply chains. It enables the supply chain to respond quickly to market changes, adjusting promptly to sudden surges in demand caused by viral trends or seasonal fluctuations, significantly improving the supply chain's responsiveness to market changes and reducing the risk of inventory backlog or stockouts due to information lag.

[0030] Specifically, the circulation data includes daily consumption, consumption region, and the number of target objects corresponding to the consumption region, collected in time sequence.

[0031] It is understandable that collecting circulation data in a time-series manner is the foundation for the efficient operation of a commodity monitoring system based on big data analysis. This allows for the capture of patterns in changes over time, presenting the circulation trends of commodities at different times and providing a basis for subsequent analysis. Daily consumption is a core indicator reflecting the real-time circulation of commodities. By collecting daily consumption data, one can directly understand the sales speed and market demand for commodities. The consumption region is an important basis for analyzing the spatial attributes of commodities. After determining the consumption region, the number of target objects corresponding to that region can be collected. In this embodiment, the target objects can be individual consumers, retailers, enterprises, or various institutions, which are used to characterize the target endpoint of commodity circulation. The consumption progress equivalent reflects the progress of commodity consumption relative to inventory. It can intuitively display the inventory consumption rate and inventory status of commodities, determined by the ratio of daily commodity consumption to the corresponding total inventory.

[0032] Specifically, the circulation change parameters include the purchase quantity adjustment ratio, inventory allocation ratio, related product set, and circulation transfer parameters corresponding to each product in the related product set.

[0033] It is understood that the purchase quantity adjustment ratio is used to clarify the range of purchase quantity adjustments required based on market changes; the inventory allocation ratio is used to guide the reasonable allocation of inventory between different regions; the associated product set is used to identify several products associated with the circulation change product; and the circulation transfer parameters are used to quantify the transmission effect of the circulation change of the product that triggers circulation analysis among the associated products.

[0034] Specifically, the spatiotemporal analysis module determines the longitudinal fluctuation ratio based on the ratio of the daily consumption of a product to the number of target objects within a preset period in the circulation data of a single product, and determines the time attribute of a single product based on the longitudinal fluctuation ratio. The fluctuation ratio is the ratio of the maximum value to the minimum value of the ratio of the daily consumption of a product to the number of target objects within a preset period. If the longitudinal fluctuation ratio is greater than or equal to the preset longitudinal ratio threshold, the spatiotemporal analysis module determines that the time attribute of a single product is seasonal. If the longitudinal fluctuation ratio is less than the preset longitudinal ratio threshold, the spatiotemporal analysis module determines that the time attribute of a single product is stable.

[0035] In this embodiment, the preset period is set to three to four months, which can capture the sales trend of seasonal products within a season more completely.

[0036] In this embodiment, the process of determining the time attribute of the corresponding product based on the circulation data analysis of the individual product includes: Based on the circulation dataset of the single commodity, the daily consumption of the commodity within a preset period and the number of target objects in the corresponding consumption area are extracted. Preferably, the target objects are individual consumers, and the number of target objects in the corresponding consumption area is the population of the consumption area. Calculate the ratio of daily product consumption to the number of target objects within a preset period; Extract the maximum and minimum values ​​from the ratios, calculate the ratio of the maximum and minimum values, and determine it as the longitudinal fluctuation ratio; If the longitudinal fluctuation ratio is greater than or equal to the preset longitudinal ratio threshold, then the time attribute of a single product is determined to be seasonal. If the longitudinal fluctuation ratio is less than the preset longitudinal ratio threshold, then the time attribute of a single product is determined to be stable.

[0037] The longitudinal ratio threshold is obtained in advance and is set to 0.95 to 1.05 times the average longitudinal fluctuation ratio of historical cycle seasonal commodities or 1.05 times the maximum longitudinal fluctuation ratio of historical cycle stable commodities.

[0038] In another embodiment, the spatiotemporal analysis module can use wavelet transform instead of simple fluctuation ratio calculation to capture multi-timescale features and obtain more accurate timescale features. It can be implemented using any of the existing methods for analyzing time series data. If the timescale feature or periodicity is greater than or equal to 1 / 2 of the preset period, the time attribute of a single commodity is determined to be seasonal; otherwise, it is determined to be stable.

[0039] Specifically, the spatiotemporal analysis module determines the spatial attributes of a single product based on the horizontal fluctuation ratio of the average value of several partition data determined according to a preset spatial division in the circulation data of a single product. If the lateral fluctuation ratio is greater than or equal to a preset lateral ratio threshold, the spatial attribute of a single product is determined to be local. If the horizontal fluctuation ratio is less than the preset horizontal ratio threshold, the spatial attribute of a single product is determined to be global.

[0040] In this embodiment, the preset space division is based on the administrative districts of prefecture-level cities. It can also be customized based on the sales area. This will not be elaborated here, as long as it can represent the characteristics of the product in the spatial area.

[0041] In this embodiment, the process of determining the spatial attributes of a corresponding product based on the circulation data analysis of the individual product includes: Based on the circulation dataset of the individual product, the daily consumption of each region within a month is extracted; The average value for each region is calculated based on the daily consumption of each region within a month. Calculate the overall average value based on the average value of each divided region; Calculate the standard deviation based on the average value of each region and the overall average value; Calculate the ratio of the standard deviation to the overall mean to determine the lateral fluctuation ratio; If the lateral fluctuation ratio is greater than or equal to a preset lateral ratio threshold, the spatial attribute of a single product is determined to be local. If the horizontal fluctuation ratio is less than the preset horizontal ratio threshold, the spatial attribute of a single product is determined to be global.

[0042] The horizontal ratio threshold is obtained in advance and is set to 1.05 times the minimum horizontal fluctuation ratio of historical periodic local commodities or 1.05 times the maximum horizontal fluctuation ratio of historical global commodities.

[0043] Specifically, the dissemination identification module determines several hot keywords for dissemination based on marketing information and current affairs information, and identifies at least one target product and its corresponding category based on the hot keywords for dissemination. The dissemination identification module is also used to determine the first circulation coefficient of the target product based on marketing information, current affairs information, and the spatial attributes of the target product.

[0044] In this embodiment, marketing information (such as e-commerce platform promotional copy and brand advertisements) and current affairs information (such as news reports, social media topics, and policy announcements) within a 7-day period are collected to generate a text corpus. Long texts are broken down into independent words using a word segmentation tool (such as jieba) (e.g., "summer air conditioner promotion" is broken down into "summer," "air conditioner," "promotion," and "promotion"). The word frequency and inverse document frequency (IVF) of each independent word are calculated, and the product of these two frequencies is used to determine the word's importance. It can be understood that the word frequency is the ratio of the number of times a word appears in a single piece of information to the total number of words in that information, reflecting the concentration of the word in a single piece of information; the IVF is the logarithm of the ratio of the total number of information entries to the number of information entries containing the word, reflecting the uniqueness of the word in the overall corpus. Based on the importance of each word, they are sorted from highest to lowest, and the words corresponding to the top 20% of the data are extracted and identified as trending keywords.

[0045] In one specific embodiment, each product is classified into different categories (snacks, fresh produce, daily necessities, or more specific sales categories) according to sales habits. Marketing information and current affairs information within a preset collection period (1 to 3 months before the peak) corresponding to the peak sales volume of a single product are extracted from its historical sales records. The daily corresponding hot keywords are determined through the above process. The historical keywords of each product are determined based on the Pearson correlation coefficient (preferably ≥0.45) between the frequency of each hot keyword and the sales volume, forming a mapping directory of historical keywords and corresponding products. Then, when the change of the hot keyword with the highest frequency in a daily or preset period is determined based on the real-time monitored marketing information and current affairs information, the products corresponding to the one or several historical keywords with the highest semantic relevance to the hot keyword with the highest frequency are determined as the target products for communication.

[0046] In this embodiment, the first circulation coefficient is used to quantify the impact of dissemination information on the circulation of goods in different spatial ranges. The process of calculating the first circulation coefficient includes: collecting the dissemination areas of marketing information and current affairs information; obtaining the spatial attributes, best-selling regions, and daily consumption within a historical period of the target product; presetting weight coefficients corresponding to spatial attributes, wherein the weight coefficient for local attributes is set to 1.0, and the weight coefficient for global attributes is set to 0.6; multiplying the ratio of the number of overlapping regions between the dissemination area and the best-selling region to the total number of best-selling regions by the corresponding weight coefficient of the spatial attribute to determine the regional overlap; calculating the average daily consumption during the dissemination period and the average daily consumption without dissemination influence within a historical period, and calculating the growth rate; the product of the regional overlap and the growth rate is determined as the first circulation coefficient. It can be understood that the higher the value of the first circulation coefficient, the more significant the positive effect of the dissemination within the core consumption space of the target product.

[0047] Specifically, the dissemination identification module constructs a dissemination sensitivity model based on the dissemination hot keywords determined from marketing information and current affairs information in historical data and the circulation data of each product, so as to identify at least one dissemination target product and the corresponding category of the product based on the dissemination hot keywords.

[0048] In this embodiment, marketing and current affairs information from three years of historical dissemination information are collected to generate a dissemination information dataset. Daily consumption and consumption regions from the circulation data of each commodity are also collected to generate a circulation dataset for each commodity, with all timestamps set to Beijing time. A dissemination sensitivity model is constructed by using the "dissemination cycle of the dissemination information (from the start of dissemination to the date when the dissemination volume drops to 50% of its peak)" as a time window and associating the circulation data of the corresponding commodities within this window. The association rules are preset to a frequency correlation coefficient greater than 0.7 and a regional overlap greater than 0.6 between the dissemination information and commodity circulation data during the dissemination cycle. The frequency correlation coefficient is calculated using the Pearson correlation coefficient formula based on the frequency of keyword occurrences in the dissemination information and the frequency of changes in commodity sales.

[0049] In one specific implementation, the commodities are categorized into 20 subcategories, such as flood control supplies and heatstroke prevention products, and each commodity is associated with 3-5 core sales regions (e.g., "flood control sandbags" are associated with 8 prefecture-level cities in the Yangtze River basin). Historical dissemination information and commodity circulation data from July to August of the past 3 years are extracted to construct a training sample. Data preprocessing: For the dissemination of information related to "rainstorm warning", the dissemination period was determined to be from July 10 to July 20, 2023 (it took 10 days for the dissemination volume to drop from the peak to 50%). At the same time, the daily consumption of "flood control sandbags" (an average of 2,000 pieces per day) and the consumption area (covering 8 cities in the Yangtze River Basin) were extracted during this period. The coverage area of ​​the dissemination of "rainstorm warning" information is 10 cities in the Yangtze River Basin.

[0050] Frequency correlation coefficient calculation: The correlation between the daily frequency of the keyword "rainstorm warning" and the daily sales volume of flood control sandbags was calculated using the Pearson formula. Propagation frequency sequence X: [50,80,120,150,130,90,70,60,40,30] Sales volume sequence Y: [1200, 1500, 1800, 2200, 2000, 1700, 1600, 1400, 1100, 900] The calculated covariance cov(X,Y) is 6840, the standard deviation σX is 42.19, σY is 415.69, and the final correlation coefficient r is 6840 / (42.19×415.69)=0.79 (≥0.7 threshold).

[0051] Regional overlap calculation: According to the intersection-union formula, the intersection of the information coverage area and the hot-selling area of ​​the product is 8 cities, and the union is 10 cities. Therefore, the regional overlap = 8 / 10 = 0.8 (≥ 0.6 threshold).

[0052] Model Training and Matching: Samples meeting both thresholds were labeled as associated samples (e.g., "rainstorm warning" - "flood control sandbags"). After splitting the training and validation sets into an 8:2 ratio, the model was trained using a dual-branch model to achieve an F1 score of 0.89. When the hot keyword "typhoon approaching" was detected in real time, the model calculated its semantic similarity with the historical keyword "rainstorm warning" to be 0.82, and output the top 3 products with the highest association probability: flood control sandbags (0.91), life jackets (0.87), and emergency lights (0.76). The overall association probability of the corresponding category "flood control supplies" was 0.85.

[0053] Specifically, the association identification module divides each product into several product type sets based on its spatiotemporal attributes. Then, it performs correlation analysis on the circulation data of each product within a single product type set to determine the associated product sets of a single product within its corresponding product type set. If the correlation coefficient of the circulation data of two products is greater than the preset correlation coefficient, one of the products is determined to be a related product of the other product.

[0054] In this embodiment, based on the temporal and spatial attributes of each commodity, the commodities can be divided into four commodity type sets: seasonal and local commodity type sets, seasonal and global commodity type sets, stable and local commodity type sets, and stable and global commodity type sets. The circulation data of the target commodity within a historical 12-month period is extracted, and the corresponding commodity type set is determined. The correlation coefficient between the target commodity and any commodity within a single commodity type set is calculated using the Pearson correlation coefficient formula. If the correlation coefficient is greater than a preset correlation coefficient, one commodity is determined to be a related commodity to the other. The preset correlation coefficient is determined based on the characteristics of the commodity type set. For example, seasonal commodity type sets require a higher correlation due to strong seasonal fluctuations, and the preset correlation coefficient is 0.7; stable commodity type sets have smaller daily circulation fluctuations, requiring a lower correlation, and the preset correlation coefficient is 0.5.

[0055] Specifically, the propagation identification module determines the time sensitivity coefficient and circulation sensitivity coefficient of a single product based on the propagation sensitivity model; Among them, the time sensitivity coefficient is related to the peak time of information dissemination and the peak time of daily consumption of goods, while the circulation sensitivity coefficient is related to the dissemination range and volume of the information dissemination and the change in total consumption of goods.

[0056] In this embodiment, daily popularity data of the disseminated information is collected during the dissemination period, including exposure, search volume, and discussion volume. The peak time of the popularity data is extracted for periods exceeding 80%. Daily consumption of the product is collected during the dissemination period, and the growth rate of daily consumption relative to the average daily consumption outside the dissemination period is calculated. The peak time of the growth rate is extracted for periods exceeding 80%. Based on the peak time of the popularity data and the peak time of the growth rate, the time overlap is calculated and determined as a time sensitivity coefficient. It can be understood that the time sensitivity coefficient quantifies the degree of correlation between the peak time of the disseminated information and the peak time of the product's daily consumption; a larger value indicates a higher degree of time matching between the two.

[0057] In this embodiment, the number of regions where the information is disseminated and the number of regions where the product is sold are collected, and the ratio between the two is calculated to determine the dissemination range index. Exposure volume, search volume, and discussion volume are collected, and the ratios of these figures to the industry averages are calculated. A weighted sum is then used to determine the dissemination volume index. Daily consumption of the product during the dissemination period is collected, and the growth rate of daily consumption relative to the average daily consumption outside the dissemination period is calculated to determine the product consumption change index. The product of the dissemination range index, the dissemination volume index, and the product consumption change index is calculated to determine the circulation sensitivity coefficient. It can be understood that the circulation sensitivity coefficient quantifies the correlation between the intensity of information dissemination and the change in product consumption; a higher value indicates a more significant driving effect of dissemination on circulation.

[0058] In this embodiment, the propagation sensitivity model is proposed to employ a deep learning architecture with an attention mechanism to handle the many-to-many relationship between keywords and products. Association analysis can incorporate a transfer entropy algorithm to identify non-linear causal relationships.

[0059] Specifically, the circulation monitoring module determines whether to trigger the circulation analysis of goods based on the rate of change of the propagation information and the rate of change of the propagation range collected in the current period, and calculates the circulation change coefficient based on the rate of change of the propagation information and the rate of change of the propagation range, the time sensitivity coefficient and the circulation sensitivity coefficient of the propagation sensitivity model. The flow change coefficient is calculated based on the corrected time sensitivity coefficient and the corrected flow sensitivity coefficient.

[0060] In this embodiment, a 24-hour monitoring cycle is set. The propagation volume and propagation range of the current and previous cycles are collected, and the propagation volume change rate and propagation range change rate are calculated. When either the propagation volume change rate or the propagation range change rate exceeds a corresponding preset threshold, a commodity circulation analysis is triggered. The thresholds corresponding to the propagation volume change rate and the propagation range change rate are 0.9 times the average value corresponding to the impact on commodity circulation within a historical 3-month period. The original time sensitivity coefficient is multiplied by 1 and summed with the propagation volume change rate to determine the corrected time sensitivity coefficient; the original circulation sensitivity coefficient is multiplied by 1 and summed with the propagation range change rate to determine the corrected circulation sensitivity coefficient. The product of the corrected time sensitivity coefficient and the corrected circulation sensitivity coefficient is calculated to determine the circulation change coefficient. It can be understood that a higher circulation change coefficient indicates a more significant overall impact of the current propagation change on commodity circulation.

[0061] In this embodiment, the flow change parameters of the identified flow change products and the flow transfer parameters of the corresponding related products are determined based on the flow change coefficient. The flow change parameters include the purchase quantity adjustment ratio and the inventory allocation ratio. The purchase quantity adjustment ratio is calculated by multiplying the flow change coefficient by the purchase sensitivity coefficient, which is set according to the product's time attribute: 0.5 for seasonal products and 0.3 for stable products. The inventory allocation ratio is calculated by multiplying the flow change coefficient by the spatial allocation coefficient, which is set according to the product's spatial attribute: 0.5 for local products and 0.3 for global products. Adjustment instructions are generated for the flow change parameters, and structured instructions are output. Any form of existing technology can be used, which will not be elaborated here. It is understood that the flow transfer parameters are used to quantify the transfer effect of the flow change of the product triggering the flow analysis among related products. The transfer parameters of related products corresponding to the transferred product are calculated by multiplying the transfer parameters of the transferred product by the similarity between the transferred product and the corresponding related product. To avoid the infinite transmission of the propagation effect (such as the transfer of the main product to the first-level related product, the transfer of the first-level related product to the second-level related product, etc.) leading to over-adjustment, a transfer termination rule is set, and the transfer stops when any of the following rules are met: If the transfer parameter of the related product is less than 0.1, it indicates that the influence of the main product on the related product is too small, and continuing the transfer is meaningless, so the transfer stops; the maximum transfer depth is set to 3 (the main product is level 0, the first-level related product is level 1, the second-level is level 2, and the third-level is level 3), and the transfer stops when it exceeds level 3 to prevent the influence from being distorted due to an excessively long transfer chain.

[0062] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A commodity monitoring system based on big data, characterized in that, include: The data acquisition module is used to collect the circulation data of each product to form a multidimensional dataset for each product. The spatiotemporal analysis module is used to determine the spatiotemporal attributes of each commodity, including time attributes and spatial attributes, based on the analysis of the circulation data of each commodity. The propagation identification module is used to acquire some propagation information, cluster them according to the category of each product, and determine the time sensitivity coefficient and circulation sensitivity coefficient of the propagation information and the product category based on the correspondence between the propagation information and the corresponding product circulation data. The association identification module is used to determine the spatiotemporal division of the circulation data of each product based on the spatiotemporal attributes of each product, and to determine the associated product set of each product based on the circulation data of each product after the spatiotemporal division. The circulation monitoring module is used to trigger the circulation analysis of goods based on changes in the dissemination information collected in the current period, determine the circulation change coefficient of the corresponding circulation change goods based on the analysis results of the dissemination information, and determine the circulation change parameters of the circulation change goods and the corresponding related goods based on the circulation change coefficient and generate procurement / inventory adjustment messages.

2. The big data-based commodity monitoring system according to claim 1, characterized in that, The circulation data includes daily consumption, consumption region, and the number of target objects corresponding to the consumption region, collected in time sequence.

3. The big data-based commodity monitoring system according to claim 2, characterized in that, The circulation change parameters include the purchase quantity adjustment ratio, inventory allocation ratio, related product set, and circulation transfer parameters corresponding to each product in the related product set.

4. The big data-based commodity monitoring system according to claim 1, characterized in that, The spatiotemporal analysis module determines the longitudinal fluctuation ratio based on the ratio of the daily consumption of a product to the number of target objects within a preset period in the circulation data of a single product, and determines the time attribute of a single product based on the longitudinal fluctuation ratio. The fluctuation ratio is the ratio of the maximum value to the minimum value of the ratio of the daily consumption of a product to the number of target objects within a preset period. If the longitudinal fluctuation ratio is greater than or equal to the preset longitudinal ratio threshold, the spatiotemporal analysis module determines that the time attribute of a single product is seasonal. If the longitudinal fluctuation ratio is less than the preset longitudinal ratio threshold, the spatiotemporal analysis module determines that the time attribute of a single product is stable.

5. The big data-based commodity monitoring system according to claim 3, characterized in that, The spatiotemporal analysis module determines the spatial attributes of a single product based on the horizontal fluctuation ratio of the average value of several partition data determined according to a preset spatial division in the circulation data of a single product. If the lateral fluctuation ratio is greater than or equal to a preset lateral ratio threshold, the spatial attribute of a single product is determined to be local. If the horizontal fluctuation ratio is less than the preset horizontal ratio threshold, the spatial attribute of a single product is determined to be global.

6. The big data-based commodity monitoring system according to claim 5, characterized in that, The dissemination identification module determines several hot dissemination keywords based on marketing information and current affairs information, and identifies at least one target product and its corresponding category based on the hot dissemination keywords. The dissemination identification module is also used to determine the first circulation coefficient of the target product based on marketing information, current affairs information, and the spatial attributes of the target product.

7. The big data-based commodity monitoring system according to claim 3, characterized in that, The dissemination identification module constructs a dissemination sensitivity model based on the dissemination hot keywords determined from marketing information and current affairs information in historical data and the circulation data of each product, so as to correspond to at least one dissemination target product and the corresponding category of the product based on the dissemination hot keywords.

8. The big data-based commodity monitoring system according to claim 7, characterized in that, The association identification module divides each product into several product type sets based on its spatiotemporal attributes. It then performs correlation analysis on the circulation data of each product within a single product type set to determine the associated product sets of each individual product within its corresponding product type set. If the correlation coefficient of the circulation data of two products is greater than the preset correlation coefficient, one of the products is determined to be a related product of the other product.

9. The big data-based commodity monitoring system according to claim 8, characterized in that, The propagation identification module determines the time sensitivity coefficient and circulation sensitivity coefficient of a single product based on the propagation sensitivity model. Among them, the time sensitivity coefficient is related to the peak time of information dissemination and the peak time of daily consumption of goods, while the circulation sensitivity coefficient is related to the dissemination range and volume of the information dissemination and the change in total consumption of goods.

10. The big data-based commodity monitoring system according to claim 7, characterized in that, The circulation monitoring module determines whether to trigger the circulation analysis of goods based on the rate of change of the propagation information and the rate of change of the propagation range collected in the current period, and calculates the circulation change coefficient based on the rate of change of the propagation information and the rate of change of the propagation range, the time sensitivity coefficient and the circulation sensitivity coefficient of the propagation sensitivity model. The flow change coefficient is calculated based on the corrected time sensitivity coefficient and the corrected flow sensitivity coefficient.

Images