E-commerce product image acquisition methods, systems and devices

Abstract

The application is suitable for the technical field of commodity graph acquisition, and particularly relates to an e-commerce commodity graph acquisition method, system and device, which comprises the following steps: obtaining commodity information database data, user behavior data and current market demand data, and analyzing commodity characteristics in multiple dimensions; classifying the commodity information database data, the user behavior data and the current market demand data to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence, so as to realize preliminary screening of commodities, accurately locate commodities meeting user demand and market trends; screening the first recommended commodity sequence and the commodity candidate sequence through the user behavior data to obtain a second recommended commodity sequence, and screening commodities more suitable for personalized user demand; constructing a commodity graph based on the second recommended commodity sequence, and performing personalized commodity recommendation to users, so as to improve the accuracy and matching degree of commodity recommendation, and improve the click rate and purchase conversion rate of recommended commodities by users.

Description

Technical Field

[0001] This application belongs to the field of product map acquisition technology, and in particular relates to a method, system and device for acquiring e-commerce product maps. Background Technology

[0002] In recent years, the e-commerce industry has experienced rapid development and has become an important part of the global economy. Numerous e-commerce platforms have emerged, and the variety and quantity of goods have grown exponentially. From large-scale comprehensive e-commerce platforms to specialized e-commerce platforms in vertical fields, they cover almost all product categories, such as clothing, electronics, food, and household goods.

[0003] Most existing e-commerce recommendation systems are based on simple algorithms and models, relying mainly on the text descriptions of products. They struggle to deeply understand the semantics of products and the real needs of users, resulting in a mismatch between recommended products and users' interests and purchasing intentions, leading to poor accuracy and effectiveness of recommendations. Summary of the Invention

[0004] This application provides a method, system, and apparatus for obtaining e-commerce product graphs, which can solve the problem that the recommended products do not match the user's interests and purchasing intentions, resulting in poor accuracy and effectiveness of recommendations, due to the difficulty in deeply understanding the semantics of the products and the user's real needs during the process of obtaining e-commerce product graphs.

[0005] In a first aspect, embodiments of this application provide a method for obtaining e-commerce product maps, including:

[0006] Acquire data on products that users are interested in within specific areas, user behavior data, and current market demand data;

[0007] Based on the product information database data, the user behavior data, and the current market demand data, a first recommended product sequence and a product alternative sequence adjacent to the first recommended product sequence are obtained.

[0008] The first recommended product sequence and the product candidate sequence are filtered using the user behavior data to obtain the second recommended product sequence;

[0009] Based on the second recommended product sequence, an e-commerce product graph is constructed, and personalized product recommendations are made to the user.

[0010] The technical solutions described in this application embodiment have at least the following technical effects:

[0011] The e-commerce product graph acquisition method provided in this application comprehensively analyzes product characteristics from multiple dimensions by acquiring product information database data, user behavior data, and current market demand data. Based on the product information database data, user behavior data, and current market demand data, a first recommended product sequence and adjacent candidate product sequences are obtained, achieving preliminary product screening and accurately identifying products that meet user needs and market trends. The first recommended product sequence and candidate product sequences are then further filtered using user behavior data to obtain a second recommended product sequence, selecting products that better match users' personalized needs. A product graph is constructed based on the second recommended product sequence, and personalized product recommendations are provided to users, significantly improving the accuracy and matching degree of product recommendations, and effectively increasing user click-through rates and purchase conversion rates for recommended products.

[0012] Secondly, embodiments of this application provide an e-commerce product map acquisition system, including:

[0013] The acquisition unit is used to acquire data from a database of products that users are interested in within a specific field, as well as user behavior data and current market demand data.

[0014] The classification unit is used to classify products based on the product information database data, the user behavior data, and the current market demand data to obtain a first recommended product sequence and a product alternative sequence adjacent to the first recommended product sequence.

[0015] A filtering unit is used to filter the first recommended product sequence and the product candidate sequence using the user behavior data to obtain a second recommended product sequence;

[0016] The recommendation unit is used to construct an e-commerce product graph based on the second recommended product sequence and to make personalized product recommendations to the user.

[0017] Thirdly, embodiments of this application provide an e-commerce product map acquisition device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method described in any of the above aspects.

[0018] Fourthly, embodiments of this application provide a computer program product that, when run on an e-commerce product image acquisition device, causes the e-commerce product image acquisition device to execute the method described in any of the above aspects.

[0019] It is understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant descriptions in the above aspects, and will not be repeated here. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating an embodiment of the e-commerce product map acquisition method provided in this application;

[0022] Figure 2 This is a schematic diagram of the structure of an e-commerce product map acquisition system provided in one embodiment of this application;

[0023] Figure 3 This is a schematic diagram of the structure of an e-commerce product map acquisition device provided in one embodiment of this application. Detailed Implementation

[0024] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0025] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0026] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0027] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determination" or "if the described condition or event is detected" may be interpreted, depending on the context, as "once determination," "in response to determination," "once the described condition or event is detected," or "in response to the detection of the described condition or event."

[0028] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0029] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0030] Most existing e-commerce recommendation systems are based on simple algorithms and models, such as collaborative filtering algorithms based on user browsing history and content-based recommendation algorithms. While these methods can provide some recommendations to users, they have significant limitations. For example, content-based recommendation algorithms rely primarily on textual descriptions of products, making it difficult to deeply understand the semantics of products and the user's true needs. This results in a mismatch between recommended products and the user's interests and purchasing intentions, and the accuracy and effectiveness of recommendations need improvement.

[0031] To address the aforementioned issues, this application provides a method, system, and apparatus for acquiring e-commerce product graphs. The method involves acquiring product information database data, user behavior data, and current market demand data to comprehensively analyze product characteristics from multiple dimensions. Based on the product information database data, user behavior data, and current market demand data, a first recommended product sequence and adjacent candidate product sequences are obtained, enabling preliminary product screening and accurately identifying products that meet user needs and market trends. The first recommended product sequence and candidate product sequences are then further filtered using user behavior data to obtain a second recommended product sequence, selecting products that better match users' personalized needs. A product graph is constructed based on the second recommended product sequence, and personalized product recommendations are provided to users, significantly improving the accuracy and matching degree of product recommendations and effectively increasing user click-through rates and purchase conversion rates.

[0032] The e-commerce product map acquisition method provided in this application embodiment can be applied to an e-commerce product map acquisition device. In this case, the e-commerce product map acquisition device is the execution subject of the e-commerce product map acquisition method provided in this application embodiment. This application embodiment does not impose any restrictions on the specific type of e-commerce product map acquisition device.

[0033] For example, e-commerce product mapping acquisition devices can be ultra-mobile personal computers (UMPCs), netbooks, desktop computers, computers, laptops, communication equipment, computing devices, satellite wireless equipment, etc.

[0034] To better understand the e-commerce product map acquisition method provided in this application embodiment, the specific implementation process of the e-commerce product map acquisition method provided in this application embodiment will be described by way of example below.

[0035] Figure 1 This paper illustrates a schematic flowchart of an e-commerce product map acquisition method provided in an embodiment of this application. Figure 2 This illustration shows a schematic diagram of the operation of the e-commerce product map acquisition method provided in an embodiment of this application. The e-commerce product map acquisition method includes:

[0036] S100 acquires database data on products that users are interested in within a specific field, user behavior data, and current market demand data.

[0037] It's understandable that "specific field" refers to specific industry sectors that users are interested in, such as clothing, electronics, and food. The product information database contains detailed information on numerous products within this field, such as product name, brand, model, function, price, and inventory. This data describes the basic attributes of the products. User behavior data records user interactions with products within this field, such as browsing history, search keywords, purchase history, favorited products, and product ratings. It reflects users' interests, preferences, and behavioral tendencies regarding different products. Current market demand data refers to the market demand for various products in this field during the current time period, including the overall market demand size, the demand for different types of products, and changes in consumer demand trends. The product information database data, user behavior data, and current market demand data can be obtained through APIs or directly from e-commerce platform databases.

[0038] S200: Based on the product information database data, user behavior data, and current market demand data, classify the products to obtain the first recommended product sequence and the alternative product sequence adjacent to the first recommended product sequence.

[0039] The first recommended product sequence can be understood as a set of products that highly match user and market needs, aligning with user interests and current market trends, and are prioritized for recommendation. The adjacent product candidate sequence is a set of products that have some relevance and slightly lower matching degree but still possess recommendation value. These products share similar attributes with the first recommended products and can satisfy some potential, related user needs. By classifying products according to their adjacent candidate sequences, a multi-level recommendation system can be constructed, satisfying core user needs while expanding the user's choices and improving the comprehensiveness and diversity of recommendations.

[0040] In one possible implementation, S200 involves classifying products based on product information database data, user behavior data, and current market demand data to obtain a first recommended product sequence and a candidate product sequence adjacent to the first recommended product sequence, including:

[0041] S210, calculate the similarity between the product information database data and the user behavior data to determine the first similarity between each product information in the product information database and the user behavior data.

[0042] Similarity calculation is a quantitative analysis method used to measure the degree of similarity between product information and user behavior data. During the calculation, key attributes from product information (such as product function, style, and price range) and key features from user behavior data (such as the types of products the user has browsed, the categories of products purchased, and search keywords) can be compared and analyzed. For example, if a user frequently browses sportswear and has purchased running shoes from a certain brand, when calculating the similarity between sportswear products in the product information database and this user's behavior data, key considerations will be whether the brand and style of the product are similar to the products the user has previously browsed and purchased, and whether the product belongs to the sportswear category. Specific algorithms (such as cosine similarity algorithms and Euclidean distance algorithms) can be used to comprehensively calculate these factors, resulting in a numerical value representing the first similarity between product information and user behavior data. The higher the similarity value, the more well the product matches the user's interests and behavioral preferences, and the more likely it is to be prioritized in subsequent recommendations.

[0043] S220, Based on the similarity calculation between the product information database data and the current market demand data, determine the second similarity between each product information in the product information database data and the current market demand data.

[0044] Understandably, using similarity calculation methods, we can compare product information in the product database with current market demand data. Current market demand data includes information such as market demand for various products and popular trends. For example, the current market demand for smartwatches among smart wearable devices is strong, with a preference for products with health monitoring functions and long battery life. When calculating the similarity between smartwatches in the product database and current market demand data, we can use a second similarity score calculated by relevant algorithms. This score reflects the product's popularity in the current market and its alignment with market trends. A higher second similarity score indicates that the product better meets current market demand and has higher recommendation value.

[0045] S230, classify the product information database data based on the first similarity and the second similarity to obtain the first recommended product sequence and the product candidate sequence adjacent to the first recommended product sequence.

[0046] It is understandable that the first and second similarity scores reflect the degree of matching between products and user and market demands from different perspectives. During the classification process, the values ​​of both scores can be combined to reorder and group products in the product database.

[0047] Optionally, in step S230, the product information database data is classified according to a first similarity and a second similarity to obtain a first recommended product sequence and a candidate product sequence adjacent to the first recommended product sequence, including:

[0048] S231, calculate the complementary value of the current market demand data and user behavior data for each product information in the product information database based on the first similarity and the second similarity.

[0049] The complementarity value is understandable; it's a quantitative indicator that comprehensively considers user behavior and market demand. User behavior data reflects personalized user needs, while market demand data reflects overall market trends, and the two complement each other. The purpose of calculating the complementarity value is to more comprehensively assess the overall matching degree between a product and user and market demand. Using the first and second similarity scores, and through specific calculation methods (such as weighted calculation or difference calculation), the impact of user behavior data and current market demand data on the product can be integrated to obtain a value that reflects the complementary relationship between the two. The larger the complementarity value, the better the product's overall performance in meeting personalized user needs and adapting to overall market demand, and the more advantageous it will be in subsequent product classification and recommendation.

[0050] S232, based on the complementary values ​​of the current market demand data and user behavior data of each product information, classify the product information by the first similarity and the second similarity of each product information to obtain the first recommended product sequence and the product candidate sequence adjacent to the first recommended product sequence.

[0051] Complementary values ​​provide a new dimension for product classification. When classifying products, complementary values ​​can be the primary basis, combined with primary and secondary similarity. By comprehensively considering multiple factors, products can be screened and classified more precisely, making recommendations more aligned with users' actual needs and dynamic market changes.

[0052] For example, S231, calculating the complementary value of the current market demand data and user behavior data for each product information in the product information database based on the first similarity and the second similarity includes:

[0053] S2311, calculate the difference term and adjustment term between the first similarity and the second similarity based on the first similarity and the second similarity.

[0054] It is understandable that nonlinear function nesting and multidimensional mapping methods can be used to calculate difference and adjustment terms, in order to more accurately measure the relationship between a product and the overall market demand in terms of meeting personalized user needs. Difference term calculation: Calculate the first similarity. Similarity to the second The absolute value of the difference is used to obtain the difference term. This difference item reflects the degree of discrepancy between the product's alignment with user behavior data and current market demand data. Adjustment item calculation: Calculate the similarity between item 1 and the first similarity. Second similarity The difference between the products yields the adjustment term. The adjustment terms take into account both the first and second similarities and are used to supplement and correct the difference terms in subsequent calculations.

[0055] S2312, Cartesian product operation is performed based on the difference term and adjustment term of the second similarity between the first similarity to obtain the initial complementary value of the current market demand data and user behavior data of each product information.

[0056] It is understandable that this can be achieved by adjusting the difference term. Perform a sine function transformation: The sine function can... The range of values ​​is mapped to [-1, 1], and is applied in a non-linear manner. Rescaling is performed to highlight the trend of the difference within a specific interval, making the impact of the difference term under different values ​​more significant. Adjustment terms... Perform an exponential-logarithmic composite transformation: This composite transformation can highlight The gradient of changes under different values ​​makes the influence of the adjustment term smoother and more non-linear, avoiding large fluctuations in the results due to small changes in the value of the adjustment term. and Performing the Cartesian product operation yields a two-dimensional vector. This vector is then mapped onto the unit circle. The angle between this vector and the reference vector (1,0) on the unit circle is calculated. The formula is (when (Time). Angle It is the initial complementary value of current market demand data and user behavior data of product information. The initial complementary value integrates the information of difference items and adjustment items, and represents the relationship between products in meeting user and market demands in a new dimension.

[0057] S2313, the initial complementary values ​​are mapped by arctangent through a preset mapping function to obtain the complementary values ​​of the current market demand data and user behavior data for each product information.

[0058] It's understandable, regarding the included angle Perform arctangent transformation: Furthermore, nonlinear compression and transformation are applied to the angle information to make the differences in angle information under different values ​​more reasonable. Through function Mapping to the [0,1] interval yields the final complementary value. This mapping process ensures that complementary values ​​conform to conventional understanding. The closer the value is to 1, the stronger the complementarity between current market demand data and user behavior data; the closer it is to 0, the weaker the complementarity. Through a series of calculations and transformations, it achieves a more comprehensive and accurate assessment of the degree of complementarity between products in meeting users' personalized needs and adapting to overall market demands, providing a more reliable basis for subsequent product classification and recommendations.

[0059] For example, in S232, based on the complementary values ​​of each product information to current market demand data and user behavior data, the product information is classified according to a first similarity and a second similarity of each product information, resulting in a first recommended product sequence and a candidate product sequence adjacent to the first recommended product sequence, including:

[0060] S2321, Based on each product information, perform a classification interval mapping on the complementary values ​​of current market demand data and user behavior data to determine the mapping interval to which each complementary value belongs; wherein, the mapping interval includes the first interval and the second interval.

[0061] It is understandable that after calculating the complementary values ​​of products, in order to classify products more effectively, the complementary values ​​can be mapped to different intervals. Based on business needs and data characteristics, two mapping intervals can be pre-defined. The first interval is set as the range with higher complementary values, such as ([0.8,1]), indicating that products within this interval perform exceptionally well in meeting personalized user needs and adapting to overall market demands. The second interval is set as the range with relatively lower complementary values ​​but still possessing some value, such as ([0.4,0.8)). By comparing the complementary value of each product with the boundaries of these two intervals, its corresponding interval can be quickly determined, thus initially filtering products and laying the foundation for subsequently constructing product sets and recommendation sequences.

[0062] S2322, construct a product set for the first interval based on all product information in the first interval, and construct a product set for the second interval based on all product information in the second interval.

[0063] Understandably, once the complementary value range for each product is determined, all product information within the first range is integrated to construct the first range product set. Products in the first range product set demonstrate excellent overall matching performance and are high-quality recommendation candidates. Similarly, all product information within the second range is aggregated to construct the second range product set. While the complementary values ​​of products in the second range product set are relatively lower than those in the first range, they still play a role in meeting market demand or some user needs. By constructing these two independent product sets, products with different levels of matching can be clearly distinguished, facilitating further filtering and processing based on the characteristics of the products within each set.

[0064] S2323, compare the first similarity of all product information in the first interval product set to obtain the first recommended product sequence.

[0065] It's understandable that the products in the first interval already possess high complementarity values, indicating good overall matching performance. To further filter these products and identify those with the highest match to the user's personal interests, we can compare the first similarity. The first similarity reflects the degree of similarity between the product and the user's behavioral data; a higher similarity indicates that the product better aligns with the user's personal preferences. By sorting the first similarity scores of all products in the first interval in descending order, and selecting a certain number (e.g., the top 20) to form the first recommended product sequence, the resulting first recommended product sequence not only performs well in overall matching but also closely relates to the user's personal interests, better meeting the user's needs and improving recommendation accuracy.

[0066] S2324, compare the second similarity of all product information in the second interval product set to obtain the product candidate sequence adjacent to the first recommended product sequence.

[0067] It's understandable that the complementary value of products in the second interval product set is relatively low, but they still have some value in meeting market demand or some user needs. To filter out products that better meet market demand from these products, a recommendation sequence adjacent to the first recommended product sequence is constructed by comparing second similarity. Second similarity reflects the degree of matching between products and current market demand data. By sorting the second similarity of all products in the second interval product set in descending order, a certain number of products (e.g., the top 30) are selected to form a candidate product sequence adjacent to the first recommended product sequence. Although this candidate product sequence may not be the most perfectly matched to user needs, it has a certain level of popularity and demand in the market, serving as a supplement to the first recommended product sequence, expanding the user's choice range, and providing richer recommendation results.

[0068] S300: Filter the first recommended product sequence and the alternative product sequence to obtain the second recommended product sequence.

[0069] Understandably, the first recommended product sequence and the alternative product sequence can be filtered to obtain a more accurate second recommended product sequence. The expected second similarity of the first recommended product sequence can be calculated to measure the average level of market demand adaptation of the products in that sequence. This reflects the degree to which the products fit current market demand data and provides a reference for subsequent filtering. The expected first similarity of the alternative product sequence can be calculated to assess the overall ability of the alternative product sequence to meet users' personalized needs, reflecting the similarity between the products and user behavior data. Finally, based on the expected first and second similarity, the first recommended product sequence and the alternative product sequence are cross-filtered. Products in the first recommended product sequence with a first similarity lower than the expected first similarity of the alternative product sequence are adjusted, as are products in the alternative product sequence with a second similarity lower than the expected second similarity of the first recommended product sequence. High-quality products are retained and integrated to form the second recommended product sequence, thereby improving the accuracy and effectiveness of recommendations and better meeting users' personalized needs while adapting to overall market demand.

[0070] In one possible implementation, S300, the first recommended product sequence and the product candidate sequence are filtered based on user behavior data to obtain a second recommended product sequence, including:

[0071] S310, based on the second similarity of all product information in the first recommended product sequence, the expected second similarity of the first recommended product sequence is obtained.

[0072] It's understandable that the first recommended product sequence is a set of products that, after initial screening, have a high degree of matching with user and market needs. The second similarity reflects the degree to which the products in the first recommended product sequence fit current market demand data. Calculating the expected second similarity of the first recommended product sequence is to measure the average level of market demand adaptation among the products in the sequence as a whole. Specifically, the expected second similarity is obtained by summing the second similarity corresponding to each product in the first recommended product sequence and then dividing by the number of products. For example, if there are 5 products in the first recommended product sequence, and their second similarities are 0.7, 0.8, 0.6, 0.75, and 0.85 respectively, then the expected second similarity is (0.7 + 0.8 + 0.6 + 0.75 + 0.85) ÷ 5 = 0.74. The expected second similarity can serve as a reference standard for subsequent screening, helping to determine which products perform better in adapting to market demand and which may require further adjustment.

[0073] S320: Statistically calculate the first similarity of all product information in the product candidate sequence to obtain the expected first similarity of the product candidate sequence.

[0074] It can be understood that a product candidate sequence is a set of products with a certain recommendation potential. The first similarity reflects the degree of similarity between the product and the user's behavioral data, indicating how well the product matches the user's personal interests. Calculating the expected first similarity of the product candidate sequence is to assess the overall ability of the products in the sequence to meet the user's personalized needs. Similar to calculating the expected second similarity of the first recommended product sequence, the expected first similarity is obtained by summing the first similarity of each product in the candidate sequence and then dividing by the total number of products. Assuming there are four products in the candidate sequence with first similarities of 0.65, 0.7, 0.68, and 0.72, the expected first similarity is (0.65 + 0.7 + 0.68 + 0.72) ÷ 4 = 0.6875. The expected first similarity result can serve as an important basis for screening product candidate sequences, determining which products better meet the user's personalized needs, thus enabling more reasonable decisions in subsequent screenings.

[0075] S330, the first recommended product sequence is filtered by the first similarity expectation, and the candidate product sequence is filtered by the second similarity expectation to obtain the second recommended product sequence.

[0076] It's understandable that the expected first and second similarities provide quantitative standards for product selection from the perspectives of personalized user needs and overall market demand, respectively. For the first recommended product sequence, the first similarity of each product is compared with the expected first similarity of the alternative product sequence. If a product's first similarity is lower than the expectation, it indicates that it may be relatively weak in meeting personalized user needs, and it can be considered for removal from the sequence or a reduction in its recommendation priority; while products with a first similarity higher than the expectation are more likely to match user interests and will be retained or have their recommendation priority increased. Similarly, for the alternative product sequence, the second similarity of each product is compared with the expected second similarity of the first recommended product sequence. Products with a second similarity lower than the expectation may be insufficient in adapting to market demand and need adjustment; products with a second similarity higher than the expectation are more valuable for recommendation. Through this two-way screening using the expected first and second similarities, the high-quality products retained from the first recommended product sequence and the alternative product sequence are integrated to obtain the second recommended product sequence. This sequence performs better in meeting personalized user needs and adapting to overall market demand, providing users with more accurate and practical product recommendations, thus improving the quality and effectiveness of recommendations.

[0077] S400 constructs an e-commerce product graph based on the second recommended product sequence and provides personalized product recommendations to users.

[0078] It's understandable that the second recommended product sequence is the result of multiple rounds of screening and optimization, making it more aligned with users' current needs. An e-commerce product graph can be built based on this second recommended product sequence to further explore the relationships between products, providing users with more accurate and personalized product recommendations. Through the e-commerce product graph, not only can direct relationships between products be displayed, but potential relationship patterns can also be discovered, thereby expanding the scope and depth of recommendations and enhancing the user's shopping experience.

[0079] In one possible implementation, S400 constructs an e-commerce product graph based on the second recommended product sequence and provides personalized product recommendations to the user, including:

[0080] S410, Based on the second recommended product sequence, determine the product association information between each product information in the second recommended product sequence.

[0081] It is understandable that the relationships between products in the second recommended product sequence can be determined, providing a basis for constructing an e-commerce product graph and conducting personalized recommendations. Co-occurrence analysis can be used based on user behavior data to determine the frequency of users simultaneously purchasing or browsing products, thus establishing a co-occurrence threshold to filter significant associations. Alternatively, a category hierarchy can be constructed based on product attributes and category associations, determining the degree of association based on the product's position in the category tree; or associations can be clarified by calculating product attribute similarity using algorithms such as cosine similarity.

[0082] S420: Using each product information in the second recommended product sequence as a node and the product association information as an edge, an e-commerce product graph is constructed.

[0083] E-commerce product graphs are understandable as a graph-structured data model that represents product information as nodes and product association information as edges, visually illustrating the relationships between products. Each product in the second recommended product sequence can be abstracted as a node, containing basic product attributes such as name, price, brand, and description. Based on the previously determined product association information, edges are created between the corresponding nodes. Edge attributes can include association strength and association type (e.g., purchase association, browsing association). For example, if the purchase association strength between product A and product B is 0.8, then an edge is created between node A and node B, with the association strength of 0.8 as an attribute of the edge. A graph database (such as Neo4j) is used to store the e-commerce product graph. Graph databases can efficiently process graph-structured data, supporting fast node and edge queries and traversals. By constructing an e-commerce product graph, the complex relationships between products can be clearly seen, providing strong support for subsequent personalized recommendations.

[0084] S430 determines personalized product recommendations based on e-commerce product maps and provides personalized product recommendations to users.

[0085] It's understandable that e-commerce product graphs present the relationships between products, serving as a crucial basis for personalized recommendations. Starting with product nodes where the user has previously interacted (e.g., purchased, browsed), paths closely related to these products can be explored within the graph. Products corresponding to the nodes connected by these paths can then be considered as recommendation candidates. For example, if a user purchased a camera, following the camera-related paths in the graph might lead to lenses, memory cards, camera bags, etc. These products are more likely to be needed by the user next and can be included in the recommendation scope. The centrality of nodes in the graph can also be analyzed, such as degree centrality and betweenness centrality. Nodes with high degree centrality are connected to many other nodes, indicating that the product occupies a core position in the graph and possesses high versatility and relevance. Nodes with high betweenness centrality play a key role in information transmission and connecting different subgraphs. Recommending products corresponding to nodes with high centrality to users increases the likelihood that the recommended products match the user's broad interests. For example, in an electronics product graph, mobile phones are typically nodes with high degree centrality because they are associated with many accessories, software, and other products; popular mobile phones and related accessories can be recommended to users.

[0086] Optionally, S430 determines personalized recommended products based on the e-commerce product map and makes personalized product recommendations to the user, including:

[0087] S431, calculate the area of ​​the polygon formed by each node and its corresponding side length based on the e-commerce product map.

[0088] It's understandable that in an e-commerce product graph, multiple related nodes and their corresponding edges can form polygons. Calculating the area of ​​these polygons can provide a new perspective on the strength and importance of the connections between nodes. To calculate the polygon area, the nodes in the graph need to be coordinated. Nodes can be represented using multi-dimensional vectors, where each dimension of the vector corresponds to a product attribute (such as price, sales volume, rating, etc.). By normalizing these attributes, the nodes are mapped to a multi-dimensional space, yielding their coordinates. For polygons composed of multiple nodes, the shoelace formula is used to calculate their area. Assume the node coordinates of the polygon are... Then the area of ​​the polygon The calculation formula is: By calculating the area of ​​the polygon enclosed by each node and its corresponding side length, the degree of connection between nodes can be quantified. The larger the area, the stronger the connection between nodes, and the higher the importance of the corresponding product combination in the graph.

[0089] S432: Compare the areas of all nodes and the corresponding polygons formed by the side lengths of the nodes, determine the product information corresponding to the five nodes with the largest polygon areas as personalized recommended products, and make personalized product recommendations to the user.

[0090] It's understandable that the node combination with the largest polygon area represents a set of closely related and important products in the graph. These products may have strong intrinsic connections and are more likely to meet the user's potential needs. The product information corresponding to the five nodes with the largest polygon areas can be used as personalized recommendations to users. This recommendation method, starting from the overall structure of the e-commerce product graph, can uncover products with high relevance and importance, providing users with more valuable recommendations and improving user shopping satisfaction and purchase conversion rates.

[0091] Corresponding to the e-commerce product map acquisition method in the above embodiments, this application also provides an e-commerce product map acquisition system, in which each unit can implement each step of the e-commerce product map acquisition method. Figure 2 The diagram shows a structural block diagram of an e-commerce product map acquisition system provided in an embodiment of this application. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0092] Reference Figure 2 The e-commerce product mapping acquisition system includes:

[0093] The acquisition unit is used to acquire data from a database of products that users are interested in within a specific field, as well as user behavior data and current market demand data.

[0094] The classification unit is used to classify products based on the product information database data, the user behavior data, and the current market demand data to obtain a first recommended product sequence and a product alternative sequence adjacent to the first recommended product sequence.

[0095] A filtering unit is used to filter the first recommended product sequence and the product candidate sequence using the user behavior data to obtain a second recommended product sequence;

[0096] The recommendation unit is used to construct an e-commerce product graph based on the second recommended product sequence and to make personalized product recommendations to the user.

[0097] It should be noted that the information interaction and execution process between the above systems / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0098] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit module can exist physically separately, or two or more unit modules can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0099] This application also provides an e-commerce product map acquisition device. Figure 3 This is a schematic diagram of the structure of an e-commerce product map acquisition device provided in an embodiment of this application. Figure 3 As shown, the e-commerce product map acquisition device 6 of this embodiment includes: at least one processor 60 ( Figure 3 Only one is shown in the image), at least one memory 61 ( Figure 3 (Only one is shown in the image) and a computer program 62 stored in the at least one memory 61 and executable on the at least one processor 60. When the processor 60 executes the computer program 62, it causes the e-commerce product map acquisition device 6 to implement the steps in any of the above-described e-commerce product map acquisition method embodiments, or causes the e-commerce product map acquisition device 6 to implement the functions of each unit in the above-described system embodiments.

[0100] For example, the computer program 62 can be divided into one or more units, which are stored in the memory 61 and executed by the processor 60 to complete this application. The one or more units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 62 in the e-commerce product map acquisition device 6.

[0101] The e-commerce product image acquisition device 6 can be a computing device or terminal device such as a desktop computer, laptop, handheld computer, or cloud server. This e-commerce product image acquisition device may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art will understand that... Figure 3This is merely an example of the e-commerce product map acquisition device 6 and does not constitute a limitation on the e-commerce product map acquisition device 6. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, buses, etc.

[0102] The processor 60 can be a Central Processing Unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0103] In some embodiments, the memory 61 may be an internal storage unit of the e-commerce product image acquisition device 6, such as a hard drive or memory of the e-commerce product image acquisition device 6. In other embodiments, the memory 61 may be an external storage device of the e-commerce product image acquisition device 6, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the e-commerce product image acquisition device 6. Further, the memory 61 may include both internal storage units and external storage devices of the e-commerce product image acquisition device 6. The memory 61 is used to store operating systems, applications, bootloaders, data, and other programs, such as the program code of the computer program. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0104] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0105] This application provides a computer program product that, when run on an e-commerce product mapping acquisition device, enables the e-commerce product mapping acquisition device to implement the steps in any of the above method embodiments.

[0106] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to the e-commerce product map acquisition device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0107] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0108] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0109] In the embodiments provided in this application, it should be understood that the disclosed e-commerce product map acquisition system / apparatus and method can be implemented in other ways. For example, the e-commerce product map acquisition system / apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0110] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0111] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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. Such 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 this application, and should all be included within the protection scope of this application.

Claims

1. An e-commerce commodity graph acquisition method, characterized in that, The method comprises: obtaining commodity information database data, user behavior data and current market demand data in which the user is interested in a specific field; classifying the commodity information database data, the user behavior data and the current market demand data to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence; screening the first recommended commodity sequence and the commodity candidate sequence to obtain a second recommended commodity sequence; constructing an e-commerce commodity graph based on the second recommended commodity sequence and making personalized commodity recommendations to the user; the classification of the commodity information database data, the user behavior data and the current market demand data to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence comprises: calculating the similarity between the commodity information database data and the user behavior data to determine the first similarity between each commodity information in the commodity information database data and the user behavior data; calculating the similarity between the commodity information database data and the current market demand data to determine the second similarity between each commodity information in the commodity information database data and the current market demand data; classifying the commodity information database data based on the first similarity and the second similarity to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence. 2.The method of claim 1, wherein, the classification of the commodity information database data based on the first similarity and the second similarity to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence comprises: calculating the complementary value of the current market demand data and the user behavior data of each commodity information in the commodity information database based on the first similarity and the second similarity; classifying the commodity information based on the first similarity and the second similarity of each commodity information according to the complementary value of the current market demand data and the user behavior data of each commodity information to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence. 3.The method of claim 2, wherein, the calculation of the complementary value of the current market demand data and the user behavior data of each commodity information in the commodity information database based on the first similarity and the second similarity comprises: calculating the difference term and the adjustment term between the first similarity and the second similarity based on the first similarity and the second similarity; performing Cartesian product operation based on the difference term and the adjustment term between the first similarity and the second similarity to obtain the initial complementary value of the current market demand data and the user behavior data of each commodity information; performing arctangent mapping of the initial complementary value through a preset mapping function to obtain the complementary value of the current market demand data and the user behavior data of each commodity information.

4. The method of claim 2, wherein the method further comprises: The classification of the complementary values of the current market demand data and the user behavior data according to each of the commodity information based on the first similarity and the second similarity of each of the commodity information obtains a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence, and includes: The classification interval mapping of the complementary values of the current market demand data and the user behavior data according to each of the commodity information determines the mapping interval to which each of the complementary values belongs; wherein the mapping interval includes a first interval and a second interval; The first interval commodity set is constructed based on all the commodity information of the first interval, and the second interval commodity set is constructed based on all the commodity information of the second interval; The first similarity corresponding to all the commodity information in the first interval commodity set is compared to obtain a first recommended commodity sequence; The second similarity corresponding to all the commodity information in the second interval commodity set is compared to obtain a commodity candidate sequence adjacent to the first recommended commodity sequence.

5. The method of claim 1, wherein, The screening of the first recommended commodity sequence and the commodity candidate sequence obtains a second recommended commodity sequence, and includes: The second similarity of all the commodity information in the first recommended commodity sequence is counted to obtain a second similarity expectation of the first recommended commodity sequence; The first similarity of all the commodity information in the commodity candidate sequence is counted to obtain a first similarity expectation of the commodity candidate sequence; The first recommended commodity sequence is screened through the first similarity expectation, and the commodity candidate sequence is screened through the second similarity expectation to obtain a second recommended commodity sequence.

6. The method of claim 1, wherein, The construction of an e-commerce commodity graph based on the second recommended commodity sequence and the personalized commodity recommendation to the user include: Based on the second recommended commodity sequence, the commodity association information between each commodity information in the second recommended commodity sequence is determined; Each commodity information in the second recommended commodity sequence is taken as a node, and the commodity association information is determined as an edge to construct an e-commerce commodity graph; The personalized recommended commodity is determined according to the e-commerce commodity graph, and the personalized commodity recommendation is made to the user.

7. The method of claim 6, wherein the method further comprises: The personalized recommended commodity is determined according to the e-commerce commodity graph, and the personalized commodity recommendation is made to the user, and includes: The corresponding polygon area surrounded by each of the nodes and the corresponding edge length of the node is calculated according to the e-commerce commodity graph; All the nodes and the corresponding polygon area surrounded by the corresponding edge length of the node are compared to determine the five nodes corresponding to the commodity information with the largest polygon area as the personalized recommended commodity and make the personalized commodity recommendation to the user.

8. An e-commerce commodity atlas acquisition system, characterized in that, The e-commerce commodity graph acquisition system for implementing the method of any one of claims 1 to 7 includes: An acquisition unit configured to acquire commodity information library data, user behavior data, and current market demand data that a user is interested in a specific field; The classification unit is configured to classify the commodity information base data, the user behavior data, and the current market demand data to obtain a first recommended commodity sequence and a commodity candidate sequence adjacent to the first recommended commodity sequence. The screening unit is configured to screen the first recommended commodity sequence and the commodity candidate sequence based on the user behavior data to obtain a second recommended commodity sequence. The recommendation unit is configured to construct an e-commerce commodity graph based on the second recommended commodity sequence and perform personalized commodity recommendation for the user.

9. An e-commerce commodity graph acquisition device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the method in any one of claims 1 to 7.

Images