A book sales feature recognition method, system, electronic device and product

Abstract

The present application belongs to the technical field of book supply chain data processing, and aims to provide a book sales feature identification method, system, electronic device and product. The method comprises: obtaining original sales data of all goods in a specified historical period, and preprocessing the original sales data to obtain preprocessed sales data; performing edition aggregation processing on the preprocessed sales data according to preconfigured edition rules to obtain a plurality of edition product sales data; performing sales feature identification on each group of edition product sales data to obtain a book sales feature label of an edition product corresponding to a category of each group of edition product sales data; wherein the book sales feature label comprises long-tail products, frequently sold products and hot-selling products, and the hot-selling products comprise seasonal products and hot products; and performing supply chain decision-making on each edition product according to the book sales feature label of each edition product. The present application can perform multi-dimensional classification on data of different categories, and provide technical support for supply chain decision-making.

Description

Technical Field

[0001] This invention belongs to the field of book supply chain data processing technology, specifically relating to a book sales feature identification method, system, electronic device, and product. Background Technology

[0002] With the development of e-commerce platforms, book sales have become characterized by a wide variety of categories, rapid update cycles, large fluctuations in demand, and diversified sales models, which has significantly increased the complexity of procurement and inventory management in the book supply chain.

[0003] Existing book supply chain management typically relies on statistical analysis based on historical sales, such as using moving weighted averages or simple trend analysis for procurement and inventory decisions, or on analysis based on human experience. This approach struggles to support sophisticated operations. Specifically, in using existing technologies, the inventors have discovered at least the following problems: a. The identification dimension is singular, relying solely on historical total sales or short-term sales trends for judgment, and it cannot distinguish between different sales categories such as regular-selling products, long-tail products, seasonal products, and popular products; b. Lack of cyclical perception ability, unable to predict peak demand for books with obvious seasonal patterns (such as educational supplementary books), leading to delayed procurement or inventory backlog; c. Outlier interference is severe, which can easily lead to misjudging occasional high-sales events such as promotions or fake orders as sustained hot-selling events, affecting the accuracy of purchasing decisions; d. Long-tail demand is not effectively identified. Low-frequency but stable long-tail books are easily excluded from the procurement plan due to low sales volume, which affects user experience and the breadth of platform services.

[0004] Therefore, there is an urgent need for a new book sales characteristic identification scheme to improve the accuracy and foresight of supply chain decisions. Summary of the Invention

[0005] The present invention aims to solve the above-mentioned technical problems to at least a certain extent, and provides a method, system, electronic device and product for identifying book sales features.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for identifying book sales features, comprising: Obtain the raw sales data of all products in a specified historical period, and preprocess the raw sales data to obtain preprocessed sales data; According to the pre-configured merging rules, the pre-processed sales data is aggregated to obtain multiple sets of merged product sales data; wherein, each set of merged product sales data corresponds to different categories of merged products. Sales feature identification is performed on the sales data of each bundled edition to obtain the book sales feature tags of the bundled edition corresponding to each category of sales data; wherein, the book sales feature tags include long-tail products, regular products and hot-selling products, and the hot-selling products include seasonal products and popular products; Based on the book sales characteristic labels of each joint edition, supply chain decisions are made for each joint edition.

[0007] In one possible design, the raw sales data is preprocessed to obtain preprocessed sales data, including: The original sales data is aligned according to the product ID to obtain aligned sales data; The aligned sales data is then cleaned to obtain cleaned sales data. The cleaned sales data is standardized to obtain preprocessed sales data.

[0008] In one possible design, the merging rule is as follows: based on the specified merging information of the product, the sales data of products with the same specified merging information are merged and aggregated; wherein, the specified merging information is grade, subject, or version.

[0009] In one possible design, the sales data for any bundled edition includes daily sales time-series data of bundled editions in its corresponding category; correspondingly, sales feature identification is performed on the sales data of any bundled edition to obtain book sales feature tags for bundled editions in the corresponding category, including: Based on the daily sales time-series data in the sales data of any combination of products, obtain the daily sales sequence variance and annual sales of the combined products of the corresponding category. Determine whether the variance of the daily sales series is less than or equal to a preset variance threshold; If the variance of the daily sales series is less than or equal to a preset variance threshold, then it is further determined whether the annual sales are less than a preset annual sales threshold. If so, the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a long-tail product. If not, the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a regular-selling product. If the variance of the daily sales series is greater than a preset variance threshold, then a hot-selling threshold is obtained based on the daily sales time series data, and it is further determined whether there is any day in the daily sales time series data where the daily sales volume is greater than the hot-selling threshold. If so, then the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a hot-selling product.

[0010] In one possible design, obtaining the best-selling threshold based on the daily sales time-series data includes: The daily sales volume of each day in the daily sales time series data is used as the candidate best-selling threshold. For any candidate best-selling threshold, the daily sales time series data is divided into low-selling data and high-selling data according to the candidate best-selling threshold, and the mean and percentage of the low-selling data and the mean and percentage of the high-selling data are obtained. Based on the mean and percentage of the low-selling data and the mean and percentage of the high-selling data, the inter-class variance between the low-selling data and the high-selling data is calculated. The maximum value of the inter-class variance corresponding to each candidate best-selling threshold is taken as the best-selling threshold.

[0011] In one possible design, after determining that the book sales characteristic tag of the combined edition corresponding to the sales data of any of the combined editions is a best-selling product, the method further includes: Obtain the hot-selling dates corresponding to the daily sales volume that is greater than the hot-selling threshold from the daily sales time series data, merge all hot-selling dates into a hot-selling interval according to time continuity, and set it as the first hot-selling interval. Before obtaining the specified historical period k The corresponding best-selling range for each cycle is designated as the second best-selling range; among which... k It is a natural number greater than 0; Calculate the overlap between the first best-selling range and the second best-selling range; Determine whether the repetition degree is greater than or equal to a preset repetition coefficient. If yes, determine that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a seasonal product. If no, determine that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a popular product.

[0012] In one possible design, after determining that the book sales characteristic tag of the combined edition corresponding to the sales data of any of the combined editions is a seasonal product, the method further includes: The overlapping intervals of the first and second best-selling intervals, the consecutive days before and after the overlapping intervals, and the intervals between the first and second best-selling intervals and the overlapping intervals within the number of days are merged to obtain the seasonal cycle of the combined product category corresponding to the sales data of any combined product.

[0013] Secondly, the present invention provides a book sales feature recognition system, comprising: The data preprocessing module is used to acquire the raw sales data of all products in a specified historical period, and preprocess it to obtain preprocessed sales data; The combined product aggregation module is communicatively connected to the data preprocessing module and is used to perform combined product aggregation processing on the preprocessed sales data according to the pre-configured combined product rules to obtain multiple combined product sales data; wherein, each combined product sales data corresponds to a different category of combined products; The sales feature recognition module is communicatively connected to the combined edition aggregation module and is used to perform sales feature recognition on the sales data of each combined edition to obtain the book sales feature tags of the combined edition corresponding to the category of each combined edition sales data; wherein, the book sales feature tags include long-tail products, regular-selling products and best-selling products; The supply chain decision module is communicatively connected to the sales feature recognition module and is used to make supply chain decisions for each jointed product based on the book sales feature tags of each jointed product.

[0014] Thirdly, the present invention provides an electronic device, comprising: Memory, used to store computer program instructions; and, A processor is configured to execute the computer program instructions to perform the operation of a book sales feature recognition method as described in any of the preceding claims.

[0015] Fourthly, the present invention provides a computer program product, including a computer program or instructions, wherein the computer program or instructions, when executed by a computer, implement a book sales feature recognition method as described in any of the above.

[0016] The beneficial effects of this invention are as follows: 1) This invention, based on a data aggregation mechanism, can extract stable sales characteristics and perform multi-dimensional classification of data from different product categories, providing technical support for precise and intelligent decision-making in the supply chain. In its implementation, this invention introduces a combined aggregation mechanism to unify multi-SKU granular data to the combined level, significantly reducing data sparsity and random fluctuations, resulting in more stable sales characteristics. Combined with a phased sales characteristic identification strategy, it achieves fine differentiation between long-tail products, regular-selling products, and best-selling products, and further supports the identification of seasonal and trending products. This transforms supply chain decision-making from being driven by single sales volume to being driven by multiple characteristics, significantly improving the accuracy and foresight of procurement plans.

[0017] 2) This invention constructs daily sales time-series data and classifies products based on two dimensions: variance and annual sales. This enables joint modeling of sales stability and sales scale, thereby effectively distinguishing between two types of combined products: "low-sales stable (long tail)" and "high-sales stable (regular)". This invention also introduces a hot-selling threshold determination mechanism when the variance is large, enabling it to identify products with obvious sales peaks. This achieves hierarchical identification of different sales behavior patterns and improves the precision of classification results.

[0018] 3) This invention automatically selects the threshold that maximizes the inter-class difference as the optimal threshold, i.e., the best-selling threshold, by traversing candidate thresholds and calculating inter-class variance. T This achieves adaptive segmentation of sales distribution for different products. A larger inter-class variance indicates a higher degree of differentiation between high-selling and low-selling products, ensuring accuracy based on the best-selling threshold. T The accuracy of identifying best-selling products is higher. Compared with fixed threshold methods, this invention can dynamically adjust the boundary between high-selling and low-selling products based on actual sales data, thereby improving the accuracy of identifying best-selling products and enhancing adaptability to different product categories and sales volumes.

[0019] 4) This invention merges peak-selling dates into peak-selling intervals based on temporal continuity, transforming previously discrete high-sales points into business-meaning time periods. This avoids interference from abnormally high sales on a single day and improves the stability of peak-selling identification. Furthermore, by introducing cross-period peak-selling interval repetition calculation, this invention achieves quantitative analysis of sales cyclicality, thereby distinguishing between cyclically driven seasonal products and non-cyclically driven hot-selling products, thus enhancing the business interpretability and decision-making value of sales feature identification.

[0020] Other beneficial effects of the present invention will be further explained in the specific embodiments. Attached Figure Description

[0021] Figure 1 This is a flowchart of a book sales feature recognition method in one embodiment; Figure 2 This is a block diagram of a book sales feature recognition system in one embodiment; Figure 3 This is a block diagram of an electronic device in one embodiment. Detailed Implementation

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0023] Example 1: This embodiment discloses a book sales feature recognition method, which can be executed by, but is not limited to, a computer device or virtual machine with certain computing resources, such as a personal computer, smartphone, personal digital assistant or wearable device, or by a virtual machine.

[0024] like Figure 1 As shown, a book sales feature recognition method may include, but is not limited to, the following steps: S1. Obtain the raw sales data of all products in a specified historical period, and preprocess the raw sales data to obtain preprocessed sales data. In this embodiment, the specified historical period is set to the most recent year, that is, the subsequent book sales feature identification is based on the raw sales data of the most recent year. Based on this, the latest dynamics of the book market and current reader preferences can be captured to the greatest extent, avoiding historical noise interference. At the same time, it can ensure that the identified seasonal products can truly reflect the current consumption trends and seasonal patterns, thereby improving the timeliness and accuracy of sales feature identification and providing data support for inventory management, precision marketing, and product selection decisions.

[0025] In this embodiment, the original sales data is extracted from business systems such as the procurement system, inventory system, and transaction system. The data includes daily sales time-series data, daily inventory time-series data, daily purchase quantity time-series data, and in-transit status of each book. The daily sales time-series data is used for subsequent book sales feature identification, while the daily inventory time-series data, daily purchase quantity time-series data, and in-transit status data are used for the specification and adjustment of the procurement plan.

[0026] In step S1, the original sales data is preprocessed to obtain preprocessed sales data, including: S101. The original sales data is aligned according to the product ID (Identification number) to obtain aligned sales data; S102. Perform data cleaning processing on the aligned sales data to obtain cleaned sales data; specifically, the data cleaning processing includes removing invalid or abnormal records, such as removing records of negative inventory, invalid orders, and non-book products.

[0027] S103. Standardize the cleaned sales data to obtain preprocessed sales data. Specifically, the standardization process includes standardizing fields such as product codes, channel identifiers, and time formats.

[0028] It should be noted that this embodiment relies on the computing power of a big data platform to integrate raw sales data from sources such as the procurement system, inventory system, and transaction system. The raw sales data undergoes data alignment, cleaning, and standardization to ensure consistency in product identification, time dimension, and field format among the data from the procurement system, inventory system, and transaction system, thereby eliminating structural differences between multi-source data. At the same time, by removing negative inventory, invalid orders, and non-book data, the interference of abnormal data on subsequent analysis is effectively reduced, improving the stability and accuracy of the subsequent sales feature identification process.

[0029] S2. According to the pre-configured merging rules, the pre-processed sales data is aggregated to obtain sales data for multiple merging products; wherein, each merging product sales data corresponds to a different category of merging products. Specifically, in this embodiment, before sales feature identification, the sales data of each product is aggregated upwards to the merging granularity to form sales data in units of "merging products," which can effectively solve the data sparsity problem caused by multiple book versions and fine splitting granularity, and enhance the comparability between different products.

[0030] In step S2, the merge rule is as follows: based on the specified merge information of the product, the sales data of products with the same specified merge information are merged and aggregated; wherein, the specified merge information is grade, subject or version.

[0031] It should be understood that product aggregation can also be performed based on product codes, combined product codes, combined product names, age groups, seasons, series, types, brands, or years.

[0032] It should be noted that in this embodiment, by aggregating products based on specified SKU information (such as grade, subject, or version), sales data that was originally scattered across multiple SKUs can be centrally represented, thereby effectively alleviating the data sparsity problem caused by overly granular SKU splitting. This aggregator approach not only improves the statistical stability of sales data but also enhances the comparability between different products, enabling subsequent book sales feature identification steps to operate on a unified scale, thus improving the consistency and reliability of the overall identification results.

[0033] S3. Sales feature identification is performed on the sales data of each bundled edition to obtain the book sales feature tags of the bundled edition corresponding to each category of sales data; wherein, the book sales feature tags include long-tail products, regular-selling products, and hot-selling products, which are structured and classified according to the book category attributes and market behavior patterns; the hot-selling products include seasonal products and popular products; specifically, among the hot-selling products, those with a cross-year recurring periodic hot-selling pattern are identified as seasonal products, and the remaining non-periodic hot-selling products are identified as popular products.

[0034] In step S3, to avoid interference from feature recognition caused by daily sales outliers, since the ratio of the daily maximum sales volume to the 98th percentile sales volume can effectively distinguish whether there are outliers in the annual daily sales sequence, outlier identification is performed on the combined product sales data before sales feature recognition. If the ratio of the daily maximum sales volume to the 98th percentile... X max / X 98If the threshold 'a' is reached, an outlier is identified, and outlier correction is performed before proceeding with subsequent sales feature identification. This avoids the impact of outliers on the accuracy of book sales feature identification.

[0035] It should be noted that long-tail products, which are those with low sales volume but stable demand (identified by variance coefficient and sales volume) and cover a niche user group, may have a small individual contribution, but together they form an important foundation for the breadth of platform services and user satisfaction. Regularly selling products refer to those with stable sales volume, small fluctuations (identified by variance coefficient and sales volume), and continuous sales throughout the year, which are suitable for regular inventory replenishment strategies. Hot-selling products refer to those with obvious sales peaks. Depending on the cause of their popularity, hot-selling products can be further divided into seasonal products and trending products. Seasonal products have a cyclical hot-selling pattern that repeats across years (repeated year coefficient) (such as educational supplementary books experiencing a concentrated surge during the school season and exam season). The system can identify historical hot-selling windows to support the business in making procurement plans in advance. On the other hand, trending products are non-cyclical high-selling products driven by short-term events (such as new book launches, media exposure, and promotional activities), which require rapid response but are not sustainable.

[0036] In step S3, the sales data for any combined edition includes the daily sales time-series data of combined editions of its corresponding category; correspondingly, in step S3, sales feature identification is performed on the sales data of any combined edition to obtain the book sales feature tags of the combined editions of the corresponding category, including: S301. Based on the daily sales time-series data in the sales data of any combined edition, obtain the daily sales sequence variance and annual sales of the combined edition products of the corresponding category. In this embodiment, the daily sales time-series data corresponding to the combined edition product G of the corresponding category can be represented as: X G ={ X 1, X 2,…, X N},in, X 1, X 2,…, X N Indicates the period from day 1 to day 2. N Daily sales N In this embodiment, N is set to 365 to specify the number of observation days for a historical period, meaning that the daily sales time series data is the daily sales sequence of a certain product over the most recent year. It should be noted that the daily sales series variance is used to measure the fluctuation range of the daily sales time series data of a certain product. Specifically, the daily sales series variance of the combined product G is: ; In the formula,X i Indicates the first i Daily sales X This represents the average value of the daily sales time-series data; The annual sales volume of the combined version G is: .

[0037] S302. Determine whether the variance of the daily sales series is less than or equal to a preset variance threshold. b .

[0038] S303. If the variance of the daily sales series is less than or equal to a preset variance threshold. b Then, it is further determined whether the annual sales volume is less than a preset annual sales volume threshold. If so, the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a long-tail product; otherwise, the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a regular seller. It should be noted that the annual sales volume threshold is set according to the bookstore's size and the average annual sales volume of combined editions, etc. In this embodiment, the annual sales volume threshold is set to 100, that is, when the variance of the daily sales volume sequence is less than or equal to the preset variance threshold. b If the annual sales volume is less than the annual sales volume threshold of 100, the book sales characteristic label of the corresponding joint edition product is determined to be a long-tail product; otherwise, it is determined to be a regular-selling product.

[0039] S304. As stated, the variance of the daily sales series is greater than a preset variance threshold. b Then, the best-selling threshold is obtained based on the daily sales time-series data. T Furthermore, it is determined whether there is any day in the daily sales time series data where the daily sales volume exceeds the hot-selling threshold. T If so, the book sales feature tag for the combined edition corresponding to the sales data of any combined edition is determined to be a best-selling item. It should be understood that the book sales feature tag may also include low-selling items. In this case, when determining whether any day's daily sales volume in the daily sales time-series data exceeds the best-selling threshold, if no, the book sales feature tag for the combined edition corresponding to the sales data of any combined edition is determined to be a low-selling item.

[0040] Specifically, in this embodiment, when it is determined that the variance of the daily sales sequence corresponding to a certain batch of counterfeit products is greater than a preset variance threshold... b Then, based on the best-selling threshold T To determine whether a product is a best-selling item, its daily sales data must show at least one day where the daily sales volume exceeds the best-selling threshold. T At that time, it was determined to be a best-selling product.

[0041] In step S304, the best-selling threshold is obtained based on the daily sales time-series data. T ,include: S3041. The daily sales volume of each day in the daily sales time series data is used as the candidate hot-selling threshold.

[0042] S3042. For any candidate best-selling threshold t According to any of the candidate best-selling thresholds t The daily sales time-series data is divided into low-sales data and high-sales data, and the mean and percentage of the low-sales data and the mean and percentage of the high-sales data are obtained.

[0043] Specifically, in this embodiment, the low-selling data can be represented as: { X i | X i ≤ t}, its mean is expressed as m 1( t The sum of these percentages represents the proportion of the total daily sales data in the time-series data. oh 1( t The high-selling data can be represented as: { X i | X i > t}, its mean is expressed as m 2( t The sum of these percentages represents the proportion of the total daily sales data in the time-series data. oh 2( t ).

[0044] S3043. Based on the mean and percentage of the low-selling data and the mean and percentage of the high-selling data, calculate the inter-class variance between the low-selling data and the high-selling data.

[0045] Specifically, in this embodiment, the inter-class variance is represented as: s ² b ( t )= oh 1( t )· oh 2( t )·[ m 1( t )- m 2( t )]²; S3044. The maximum value of the inter-class variances corresponding to each candidate best-selling threshold is taken as the best-selling threshold. T .

[0046] Specifically, in this embodiment, the best-selling threshold is represented as: T= argmax s ² b ( t ); Here, argmax() is an index function used to find the element with the maximum value in the array.

[0047] It should be noted that this embodiment automatically selects the threshold that maximizes the difference between classes as the optimal threshold, i.e., the best-selling threshold, by traversing the candidate thresholds and calculating the inter-class variance. T This achieves adaptive segmentation of sales distribution for different products. A larger inter-class variance indicates a higher degree of differentiation between high-selling and low-selling products, ensuring accuracy based on the best-selling threshold. T The accuracy of identifying best-selling products is higher. Compared to the fixed threshold method, this embodiment can dynamically adjust the boundary between high-selling and low-selling products based on actual sales data, thereby improving the accuracy of identifying best-selling products and enhancing the adaptability to different product categories and sales volumes.

[0048] In step S304, after determining that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a best-selling product, the method further includes: S305. Obtain the daily sales time-series data that are greater than the hot-selling threshold. T The daily sales volume corresponds to the peak sales dates. All peak sales dates are combined according to their time continuity to form a peak sales interval, which is then set as the first peak sales interval. P It should be noted that single-day hot sales may be caused by accidental factors (such as promotions or restocking after stockouts), and only continuous hot sales periods have stable business significance. In this embodiment, by dividing the hot sales dates into intervals, accidental time points are transformed into time periods with business significance.

[0049] S306. Obtain the specified historical period before k The corresponding best-selling range for each cycle is designated as the second best-selling range; among which... k It is a natural number greater than 0; S307. Calculate the overlap between the first best-selling period and the second best-selling period; specifically, in this embodiment, the overlap between the two time periods is calculated by using the overlap duration / total coverage duration method.

[0050] In this embodiment, there is one or more second hot-selling intervals. When there are multiple second hot-selling intervals, the weights are calculated based on the proximity of the years (the more recent the year, the higher the weight), so that the first hot-selling interval and the second hot-selling interval can be found to have a degree of overlap.

[0051] S308. Determine whether the repetition degree is greater than or equal to a preset repetition coefficient. If yes, it means that the product has been continuously selling well in the same period of multiple years. Determine that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a seasonal product. If no, determine that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a hot-selling product.

[0052] It should be noted that if a product consistently sells well during a fixed period each year, driven by seasonal or cyclical factors, it is a seasonal product; if a product only sells well in a specific year, lacks stable cyclicality, and is driven by chance events or short-term trends, it is more likely to be a short-lived hot product.

[0053] This embodiment combines peak-selling dates into peak-selling intervals based on temporal continuity, transforming previously discrete high-sales points into business-meaning time periods. This avoids interference from abnormally high sales on a single day and improves the stability of peak-selling identification. Furthermore, this embodiment introduces cross-cycle peak-selling interval repetition calculation to achieve quantitative analysis of sales cyclicality. This allows for the differentiation between cyclically driven seasonal products and non-cyclically driven hot-selling products, thereby enhancing the business interpretability and decision-making value of sales feature identification.

[0054] In step S308, after determining that the book sales characteristic tag of the combined edition corresponding to the sales data of any combined edition is a seasonal product, the method further includes: S309. The overlapping interval between the first best-selling interval and the second best-selling interval, and the number of consecutive days C before and after the overlapping interval. c The interval between the first best-selling interval and the overlapping interval in the second best-selling interval is in the interval number C. g The intervals within the range are merged to obtain the seasonal cycle P of the combined products corresponding to the sales data of any combined version. s .

[0055] In this embodiment, by identifying the sales characteristics of each bundled product's sales data, the book sales characteristic tags of the bundled products corresponding to each category of bundled product's sales data are obtained. If the book sales characteristic tag of a certain bundled product is a seasonal product, the seasonal cycle of the bundled product is output simultaneously. It can also output multi-dimensional characteristics such as the average sales volume of the same period, the current available inventory, and the quantity in transit, so as to make subsequent supply chain decisions for seasonal products, assist procurement personnel in demand forecasting, and improve the ability to cover seasonal demand.

[0056] Specifically, the peak sales periods in different years may not align perfectly. For example, if the peak sales period for a certain type of book is November 1st-November 25th, 2025, and November 5th-December 5th, 2024, then it needs to be determined by the consecutive days C.c and the interval days C g Merge the intervals that are "similar but slightly offset" into a unified seasonal cycle P. s This allows multiple peak-selling periods with time shifts to be uniformly grouped into a stable seasonal cycle. Specifically, if the number of days between two overlapping periods on the time axis is ≤ C... g If these are considered as continuations of the same seasonal cycle, they are merged into a larger interval. After obtaining the seasonal cycle P... s Then, the smallest start date can be used as the seasonal period P. s The starting point and the maximum end date are used as the seasonal cycle P. s The system identifies the end point of the peak sales period and outputs both Gregorian and Lunar calendar versions as needed. This ensures that even slight deviations in peak sales periods across different years (due to lunar calendar changes or climate fluctuations) can be reasonably grouped into a unified seasonal cycle, guaranteeing that the identified cycle is both stable and provides practical business guidance.

[0057] It should be noted that in this embodiment, by constructing daily sales time-series data and classifying products based on two dimensions—variance and annual sales—a joint modeling of sales stability and sales scale is achieved, thereby effectively distinguishing between two types of combined products: "low-sales stable (long tail)" and "high-sales stable (regular)". This embodiment also introduces a hot-selling threshold determination mechanism when the variance is large, enabling it to identify products with obvious sales peaks, thereby achieving hierarchical identification of different sales behavior patterns and improving the precision of the classification results.

[0058] S4. Make supply chain decisions for each joint edition based on the book sales characteristic labels of each joint edition.

[0059] Specifically, for frequently sold products, they are included in the regular safety stock model, and the business maintains stable turnover through reports and daily monitoring. For long-tail products, although we do not actively stock up in large quantities, we ensure basic supply through a "low threshold trigger replenishment" mechanism to maintain the integrity of the platform's product categories and user satisfaction. For seasonal products among best-selling items, when the peak sales period is approaching but the inventory level is insufficient, a "high-priority stockout warning" is triggered, prompting emergency allocation or additional orders; For the most popular items among the best-selling products, a "small batch, quick response" strategy is adopted to avoid inventory backlog caused by misjudging them as continuously popular.

[0060] This embodiment, based on a data aggregation mechanism, can extract stable sales characteristics and perform multi-dimensional classification of data from different product categories, providing technical support for precise and intelligent decision-making in the supply chain. During implementation, by introducing a combined aggregation mechanism, data at multiple SKU (Stock Keeping Unit) granularities are unified to the combined level, significantly reducing data sparsity and random fluctuations, resulting in more stable sales characteristics. Combined with a phased sales characteristic identification strategy, it achieves fine differentiation between long-tail products, regular-selling products, and best-selling products, and further supports the identification of seasonal and trending products. This transforms supply chain decision-making from being driven by single sales volume to being driven by multiple characteristics, significantly improving the accuracy and foresight of procurement plans.

[0061] Example 2: This embodiment discloses a book sales feature recognition system for implementing the book sales feature recognition method in Embodiment 1; such as Figure 2 As shown, the book sales feature recognition system includes: The data preprocessing module is used to acquire the raw sales data of all products in a specified historical period, and preprocess it to obtain preprocessed sales data; The combined product aggregation module is communicatively connected to the data preprocessing module and is used to perform combined product aggregation processing on the preprocessed sales data according to the pre-configured combined product rules to obtain multiple combined product sales data; wherein, each combined product sales data corresponds to a different category of combined products; The sales feature recognition module is communicatively connected to the combined edition aggregation module and is used to perform sales feature recognition on the sales data of each combined edition to obtain the book sales feature tags of the combined edition corresponding to the category of each combined edition sales data; wherein, the book sales feature tags include long-tail products, regular-selling products and best-selling products; The supply chain decision module is communicatively connected to the sales feature recognition module and is used to make supply chain decisions for each jointed product based on the book sales feature tags of each jointed product.

[0062] It should be noted that the working process, working details and technical effects of the book sales feature recognition system provided in this embodiment 2 can be found in embodiment 1, and will not be repeated here.

[0063] Example 3: Based on Embodiment 1 or 2, this embodiment discloses an electronic device, which may be a smartphone, tablet computer, laptop computer, or desktop computer, etc. The electronic device may be referred to as a user terminal, portable terminal, desktop terminal, etc. Figure 3 As shown, the electronic device includes: Memory, used to store computer program instructions; and, A processor is used to execute the computer program instructions to perform the operation of a book sales feature recognition method as described in any of Embodiment 1.

[0064] Specifically, processor 301 may include one or more processing cores, such as a quad-core processor or an octa-core processor. Processor 301 may be implemented using at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). Processor 301 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, processor 301 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed on the screen.

[0065] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory 302 are used to store at least one instruction, which is executed by the processor 301 to implement the book sales feature recognition method provided in Embodiment 1 of this application.

[0066] In some embodiments, the terminal may also optionally include a communication interface 303 and at least one peripheral device. The processor 301, memory 302, and communication interface 303 can be connected via a bus or signal line. Each peripheral device can be connected to the communication interface 303 via a bus, signal line, or circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 304, a display screen 305, and a power supply 306.

[0067] The communication interface 303 can be used to connect at least one I / O (Input / Output) related peripheral device to the processor 301 and the memory 302. In some embodiments, the processor 301, the memory 302, and the communication interface 303 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 301, the memory 302, and the communication interface 303 can be implemented on separate chips or circuit boards, which is not limited in this embodiment.

[0068] The radio frequency (RF) circuit 304 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The RF circuit 304 communicates with communication networks and other communication devices via electromagnetic signals.

[0069] Display screen 305 is used to display the UI (User Interface). The UI may include any combination of graphics, text, icons, and video.

[0070] Power supply 306 is used to supply power to various components in electronic devices.

[0071] Example 4: Based on any one of Embodiments 1 to 3, this embodiment discloses a computer program product, including a computer program or instructions, which, when executed by a computer, implements a book sales feature recognition method as described in any one of Embodiments 1. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0072] Obviously, those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps as a single integrated circuit module. Thus, the present invention is not limited to any particular hardware and software combination.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying book sales features, characterized in that, include: Obtain the raw sales data of all products in a specified historical period, and preprocess the raw sales data to obtain preprocessed sales data; According to the pre-configured merging rules, the pre-processed sales data is aggregated to obtain multiple sets of merged product sales data; wherein, each set of merged product sales data corresponds to different categories of merged products. Sales feature identification is performed on the sales data of each bundled edition to obtain the book sales feature tags of the bundled edition corresponding to each category of sales data; wherein, the book sales feature tags include long-tail products, regular products and hot-selling products, and the hot-selling products include seasonal products and popular products; Based on the book sales characteristic labels of each joint edition, supply chain decisions are made for each joint edition.

2. The book sales feature recognition method according to claim 1, characterized in that, The original sales data is preprocessed to obtain preprocessed sales data, including: The original sales data is aligned according to the product ID to obtain aligned sales data; The aligned sales data is then cleaned to obtain cleaned sales data. The cleaned sales data is standardized to obtain preprocessed sales data.

3. The book sales feature identification method according to claim 1, characterized in that, The combined version rule is as follows: based on the specified combined version information of the product, the sales data of products with the same specified combined version information are combined and aggregated; wherein, the specified combined version information is grade, subject or version.

4. The book sales feature recognition method according to claim 1, characterized in that, Sales data for any bundled edition includes daily sales time-series data for bundled editions of its corresponding category; correspondingly, sales feature identification is performed on the sales data of any bundled edition to obtain book sales feature tags for bundled editions of the corresponding category, including: Based on the daily sales time-series data in the sales data of any combination of products, obtain the daily sales sequence variance and annual sales of the combined products of the corresponding category. Determine whether the variance of the daily sales series is less than or equal to a preset variance threshold; If the variance of the daily sales series is less than or equal to a preset variance threshold, then it is further determined whether the annual sales are less than a preset annual sales threshold. If so, the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a long-tail product. If not, the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a regular-selling product. If the variance of the daily sales series is greater than a preset variance threshold, then a hot-selling threshold is obtained based on the daily sales time series data, and it is further determined whether there is any day in the daily sales time series data where the daily sales volume is greater than the hot-selling threshold. If so, then the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is determined to be a hot-selling product.

5. The book sales feature identification method according to claim 4, characterized in that, The best-selling threshold is obtained based on the daily sales time-series data, including: The daily sales volume of each day in the daily sales time series data is used as the candidate best-selling threshold. For any candidate best-selling threshold, the daily sales time series data is divided into low-selling data and high-selling data according to the candidate best-selling threshold, and the mean and percentage of the low-selling data and the mean and percentage of the high-selling data are obtained. Based on the mean and percentage of the low-selling data and the mean and percentage of the high-selling data, the inter-class variance between the low-selling data and the high-selling data is calculated. The maximum value of the inter-class variance corresponding to each candidate best-selling threshold is taken as the best-selling threshold.

6. The book sales feature recognition method according to claim 4, characterized in that, After determining that the book sales feature tag of the combined edition corresponding to the sales data of any of the combined editions is a best-selling product, the method further includes: Obtain the hot-selling dates corresponding to the daily sales volume that is greater than the hot-selling threshold from the daily sales time series data, merge all hot-selling dates into a hot-selling interval according to time continuity, and set it as the first hot-selling interval. Before obtaining the specified historical period k The corresponding best-selling range for each cycle is designated as the second best-selling range; among which... k It is a natural number greater than 0; Calculate the overlap between the first best-selling range and the second best-selling range; Determine whether the repetition degree is greater than or equal to a preset repetition coefficient. If yes, determine that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a seasonal product. If no, determine that the book sales feature tag of the combined edition corresponding to the sales data of any combined edition is a popular product.

7. The book sales feature identification method according to claim 6, characterized in that, After determining that the book sales characteristic tag of the combined edition corresponding to the sales data of any of the combined editions is a seasonal product, the method further includes: The overlapping intervals of the first and second best-selling intervals, the consecutive days before and after the overlapping intervals, and the intervals between the first and second best-selling intervals and the overlapping intervals within the number of days are merged to obtain the seasonal cycle of the combined product category corresponding to the sales data of any combined product.

8. A book sales feature recognition system, characterized in that, include: The data preprocessing module is used to acquire the raw sales data of all products in a specified historical period, and preprocess it to obtain preprocessed sales data; The combined product aggregation module is communicatively connected to the data preprocessing module and is used to perform combined product aggregation processing on the preprocessed sales data according to the pre-configured combined product rules to obtain multiple combined product sales data; wherein, each combined product sales data corresponds to a different category of combined products; The sales feature recognition module is communicatively connected to the combined edition aggregation module and is used to perform sales feature recognition on the sales data of each combined edition to obtain the book sales feature tags of the combined edition corresponding to the category of each combined edition sales data; wherein, the book sales feature tags include long-tail products, regular-selling products and best-selling products; The supply chain decision module is communicatively connected to the sales feature recognition module and is used to make supply chain decisions for each jointed product based on the book sales feature tags of each jointed product.

9. An electronic device, characterized in that, include: Memory is used to store computer program instructions; as well as, A processor is configured to execute the computer program instructions to perform the operation of a book sales feature recognition method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or the instructions are executed by the computer, they implement a book sales feature recognition method as described in any one of claims 1 to 7.

Images