Product flow analysis and competitive product identification method based on ASIN and keywords

Abstract

The invention belongs to the technical field of e-commerce competitive product analysis, and more specifically relates to an ASIN and keyword-based product flow analysis and competitive product identification method, in the invention, a predetermined time window mechanism is introduced, the similarity change trend of a target ASIN and a high-overlapping competitive product in each time window is repeatedly calculated, and the similarity change trend of the target ASIN and the high-overlapping competitive product in each time window is calculated; screening stable competitive products and temporary competitive products in combination with a similarity threshold value of a continuous window; a flow competition network is constructed by taking the ASIN as a node and the competition similarity as an edge, a potential competitive product pool is obtained by clustering through a Louvain community discovery algorithm, then the flow overlapping degree of the ASIN in the competitive product pool and the target ASIN is further calculated, only high-overlapping-degree competitive products higher than a preset threshold value are reserved, weak correlation ASINs with low flow overlapping are effectively eliminated, and the accuracy of the detection result is improved. And the problem of competition group redundancy is solved.

Description

Technical Field

[0001] This application belongs to the field of e-commerce competitor analysis technology, and more specifically, relates to a product traffic analysis and competitor identification method based on ASIN and keywords. Background Technology

[0002] In e-commerce platforms such as Amazon and Tmall Global, ASIN is the unique identifier of a product, while keywords are the core traffic entry point for users to search for and reach the product. Users search for products by entering keywords, and the keyword search volume, ranking, click-through rate, and conversion rate of ASIN directly determine the effective traffic volume it acquires. The traffic volume is directly related to product sales, market share, and profitability. Therefore, accurately analyzing the keyword traffic characteristics of ASIN to identify competitors that are truly vying for the same keyword traffic is the core prerequisite for merchants to formulate operational strategies.

[0003] This year, the industry has begun to experiment with cluster analysis based on keyword data. For example, it calculates competitive similarity by the number of shared keywords among ASINs and then uses the K-means algorithm to divide competitive groups. However, this method has the following limitations: Competitive relationships are not static. Some ASINs may temporarily enter the core keyword traffic pool of a target ASIN due to seasonal promotions or short-term advertising campaigns, but the competitive relationship weakens rapidly after the promotion ends. On the other hand, some ASINs compete with the target ASIN for the same batch of core keywords for a long time. Simple clustering methods divide groups based on static data at a certain moment, which cannot capture the dynamic nature of time. This leads merchants to include temporary competitors in long-term monitoring or mistakenly regard stable competitors as short-term competition, resulting in serious resource misallocation.

[0004] Secondly, a large number of shared keywords does not necessarily mean a high degree of overlap in core traffic. If the shared keywords among ASINs are mostly marginal keywords with low search volume and low traffic contribution, even if there are many shared keywords, the actual traffic threat to the target ASIN is minimal. However, simple clustering methods fail to filter high-value competitors through traffic overlap, resulting in a large number of related ASINs mixed in with the overall competitive output, making it impossible for merchants to focus on the core threat. Summary of the Invention

[0005] This invention provides a product traffic analysis and competitor identification method based on ASIN and keywords, aiming to solve the technical problems of current methods lacking a dynamic time dimension, being unable to distinguish between stable and temporary products, lacking traffic overlap filtering, and having redundant competitors.

[0006] A method for product traffic analysis and competitor identification based on ASINs and keywords includes the following steps: S1: Collect the keyword traffic data of the target ASIN and the keyword traffic data of similar ASINs, and preprocess the collected data to obtain a standardized core keyword library, traffic data of the target ASIN and similar ASINs in the core keyword library; S2: Calculate the traffic competitiveness of each combination of ASIN and core keyword, and then construct a keyword-ASIN traffic weighted matrix, taking the core keyword library as the columns of the matrix and the ASINs as the rows of the matrix to form a two-dimensional matrix with the number of ASINs as the number of rows and the number of core keywords as the number of columns, and each element value in the matrix corresponds to the traffic competitiveness value of the ASIN under the corresponding keyword; S3. For the target ASIN and any other ASIN, traverse each keyword in the core keyword library, calculate the traffic competition similarity based on predefined rules, and obtain a list of competition similarity of the target ASIN and all other ASINs of similar products; S4. Form an ASIN traffic competition network by taking ASINs as network nodes and the connection edge weight between nodes as the traffic competition similarity, cluster the competition network using the Louvain community discovery algorithm, obtain a potential competitor pool, and calculate the traffic overlap degree of each ASIN in the potential competitor pool, filter high-overlap competitors with a traffic overlap degree higher than a predetermined threshold to obtain a high-overlap competitor list; S5. Repeat steps S3 to S4 in a predetermined time window to calculate the similarity score trend of the target ASIN and the high-overlap competitors in each window, and distinguish between stable competitors and temporary competitors. For stable competitors, calculate the traffic proportion difference of the two on common keywords, and filter keywords with a traffic proportion difference greater than a predetermined threshold as disadvantageous keywords for the target ASIN.

[0007] In the present application, by introducing a predetermined time window matrix, the similarity trend of the target ASIN and the high-overlap competitors in each time window is repeatedly calculated, and the similarity threshold of consecutive windows is combined to filter stable competitors and temporary competitors. In the present application, an ASIN is taken as a node, and a competition similarity is taken as an edge to construct a traffic competition network, and the Louvain community discovery algorithm is used for clustering to obtain a potential competitor pool, and then the traffic overlap degree of the ASINs in the competitor pool and the target ASIN is calculated, and only the high-overlap competitors with a traffic overlap degree higher than a predetermined threshold are retained, effectively eliminating weakly related ASINs with low traffic overlap, and solving the problem of redundant competition groups.

[0008] Preferably, the keyword traffic data of the target ASIN and the keyword traffic data of similar ASINs each include a unique identifier of the ASIN, a keyword text, a search volume, a click-through rate, a conversion rate, and a timestamp.

[0009] Preferably, the traffic competitiveness calculation step is as follows: The traffic competitiveness of any ASIN and any core keyword is the exposure of the ASIN under the corresponding keyword multiplied by the click rate and then multiplied by the conversion rate; Wherein the arbitrary ASIN belongs to the same category ASIN set, and the arbitrary core keyword belongs to the core keyword library.

[0010] Preferably, the step S3 comprises the following steps: The total search volume of all core keywords is calculated by the total search volume of each core keyword, and then the total search volume of a single keyword is divided by the sum of the total search volume to obtain the normalized global traffic weight; The traffic competitiveness of the target ASIN and other ASINs under the keyword k is extracted from the two-dimensional matrix, and the smaller value is divided by the sum of the maximum value and a coefficient value to avoid zero denominator to obtain the traffic overlap degree; The global traffic weight of the keyword is multiplied by the traffic overlap degree to obtain the contribution of the keyword to the competition similarity of the two; The total competition similarity is obtained by accumulating the contributions of all keywords, and the competition similarity list of the target ASIN and all other ASINs of the same category is obtained based on this.

[0011] Preferably, in the step 3, if there is no traffic of two ASINs under a certain keyword, the contribution of the keyword to the competition similarity of the two ASINs is 0; If only one ASIN has traffic under a certain keyword, the contribution of the keyword to the competition similarity of the two ASINs is 0; If there is an ASIN under a certain keyword, and the global traffic weight is less than a preset threshold, the contribution of the keyword to the competition similarity of the two ASINs is 0.

[0012] Preferably, the step S4 comprises the following steps: Building an ASIN traffic competition network: regarding each ASIN as an independent node, including the target ASIN and other ASINs of the same category, if the traffic competition similarity of two ASINs is greater than 0, a edge is established between the nodes, and the weight of the edge adopts the value of the traffic competition similarity, and the edges with the top 20% of the edge weights are retained; Initializing community division: regarding each ASIN as an independent community to form an initial community set; Calculating the modularity: determining the total sum of all edge weights in the ASIN traffic competition network, then calculating the sum of the competition intensity of each ASIN and all other ASINs, and then combining the Kronecker function to obtain the Q value of the current community division in the modularity calculation formula; Local optimization: moving each ASIN from the current community to the community where the adjacent ASIN is located one by one, calculating the modularity gain after each movement, retaining the movement operation with the maximum modularity gain and greater than 0, and repeating the local optimization until all ASINs no longer move; Community aggregation: regarding each community after local optimization as a super node, calculating the edge weight between different super nodes to form a simplified network; Iteration termination: repeating the local optimization to community aggregation process until the lifting amplitude of modularity is less than the predetermined threshold, and outputting the final community division result; Positioning potential competitor pool: based on the final community division result, finding the unique community containing the target ASIN, and all ASINs in the unique community except the target ASIN are the potential competitor pool; Traffic overlap degree screening: calculating the traffic overlap degree of each ASIN in the potential competitor pool, screening high overlap competitors higher than the predetermined threshold, and obtaining a high overlap competitor list.

[0013] Preferably, the traffic overlap degree screening comprises the following steps: Adding the traffic competitiveness values of the target ASIN on all core keywords to obtain the total effective traffic of the target ASIN; For each ASIN in the potential competitor pool, determining the common keywords combination, and then adding the traffic competitiveness of the target ASIN corresponding to all keywords in the set to obtain the total traffic contribution of the common keywords to the target ASIN; Taking the percentage of the total traffic contribution of the common keywords to the target ASIN in the total effective traffic of the target ASIN as the traffic overlap degree; Based on the set traffic overlap degree threshold and the calculated traffic overlap degree, screening competitors higher than the traffic overlap degree threshold to obtain a high overlap competitor list.

[0014] Preferably, the identification of the stable competitor comprises the following steps: If the competitor meets the similarity threshold condition and the fluctuation stability condition at the same time, it is a stable competitor; The similarity threshold condition is that the traffic competitiveness similarity of the high overlap competitor list in the continuous three windows is greater than or equal to the preset similarity threshold; The fluctuation stability condition is that the fluctuation of the traffic competitiveness similarity of the three windows is less than or equal to the predetermined threshold.

[0015] Preferably, obtaining the disadvantageous keywords of the target ASIN comprises the following steps: Extract the shared keyword set of the target ASIN and a single stable competitor, calculate the flow proportion difference of the competitor to the target ASIN for each shared keyword, and based on the set proportion difference threshold, filter out the keywords with flow proportion difference greater than or equal to the proportion difference threshold as disadvantageous keywords.

[0016] Preferably, the temporary competitor identification comprises the following steps: If the competitor satisfies both the sudden increase condition and the seasonal keyword verification, the competitor is a temporary competitor; The sudden increase condition is that when the flow competition similarity of the competitor and the target ASIN in the predetermined window is greater than or equal to the predetermined similarity threshold, and the growth rate is greater than or equal to the predetermined growth rate threshold; The seasonal keyword verification is to extract the keywords that cause the sudden increase of the current window similarity, and verify the existence of nodular or event keywords in the seasonal keyword verification keywords.

[0017] The beneficial effects of the present application include: In the present application, by introducing a predetermined time window matrix, the similarity change trend of the target ASIN and the high overlap competitor in each time window is repeatedly calculated, and the similarity threshold of the continuous window is combined to filter out stable competitors and temporary competitors; and in the present application, ASIN is taken as a node, and competition similarity is taken as an edge to construct a flow competition network, and the Louvain community discovery algorithm is used to cluster to obtain a potential competitor pool, and then the flow overlap degree of the ASIN and the target ASIN in the competitor pool is calculated, and only the high overlap degree competitors higher than the predetermined threshold are reserved, effectively eliminating the weakly related ASIN with low flow overlap, and solving the problem of competition group redundancy. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 The overall step block diagram provided by the embodiments of the present application.

[0020] Figure 2 The flowchart of step S3 provided by the embodiments of the present application. DETAILED DESCRIPTION

[0021] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0022] See Figure 1 As shown, a method for product traffic analysis and competitor identification based on ASINs and keywords includes the following steps: S1: Collect keyword traffic data of the target ASIN and similar ASINs, and preprocess the collected data to obtain a standardized core keyword library, traffic data of the target ASIN and similar products in the core keyword library. The keyword traffic data for the target ASIN and similar ASINs includes the ASIN's unique identifier, keyword text, search volume, click-through rate, conversion rate, and timestamp.

[0023] In this embodiment, the target ASIN and time granularity parameter are passed into the GetKeywordsForASIN interface of the platform API to obtain keyword traffic data for the past 90 days or 12 weeks. If the interface returns missing data, the average of the previous 3 days is used to fill it. If the CTR / CVR is abnormal (such as CTR>50% or CVR>30%, which is outside the reasonable range of the industry), it is marked as invalid value.

[0024] Invalid keywords are filtered out from the collected data. If a keyword has no search volume within the predetermined time collection period, it is filtered out as an invalid keyword. Then, keywords that are not related to the core attributes of ASIN are filtered out based on similarity calculation. For example, if the similarity is less than 0.3, the keyword is considered to be an irrelevant keyword. Next, semantically similar keywords are merged into the same core keyword. For example, based on word2Vec word vector clustering, the word2Vec model is trained using only one year of search term logs from the platform, with parameters of vector dimension=100, window size=5, minimum word frequency=5, negative sampling=5, and outputting a 100-dimensional word vector for each keyword. The cosine similarity of word vectors is calculated for any two keywords, and the calculated cosine similarity is compared with a similarity threshold. If it is greater than the similarity threshold, the two keywords are merged, and the cluster center word is selected from the keyword with the highest traffic contribution within the cluster.

[0025] The percentage of traffic contribution of keyword k to the target ASIN is the proportion of effective traffic acquisition capability of this keyword to the total effective traffic: ; ; In the formula: This indicates the effective traffic acquisition capability of ASIN a under keyword k; This represents the total effective traffic acquisition capability of target ASIN a across all keywords. This indicates the search volume for ASIN a under keyword k; This indicates the conversion rate of ASIN a under keyword k; This indicates the click-through rate of ASIN a under keyword k.

[0026] reserve The keywords were used to form a core keyword library.

[0027] S2: Calculate the traffic competitiveness for each combination of ASIN and core keywords, and then construct a keyword-ASIN traffic weighted matrix. The core keyword library is used as the columns of the matrix and ASIN is used as the rows of the matrix, forming a two-dimensional matrix with the number of ASINs in the rows and the number of core keywords in the columns. Each element value in the matrix corresponds to the traffic competitiveness value of the ASIN under the corresponding keyword. In this embodiment, before calculating the traffic competitiveness, it is necessary to align the time windows of all ASIN keyword data, and then perform time window alignment on any ASIN. and keywords Calculate traffic competitiveness Its traffic competitiveness is represented by the effective traffic acquisition capability of ASIN a under keyword k in the above formula; If an ASIN has no exposure under a certain keyword, its traffic competitiveness is 0. No exposure means no traffic, which in turn leads to no conversion rate. If an ASIN has exposure for a certain keyword but no clicks, its traffic competitiveness is 0. Without clicks, it cannot enter the conversion stage and has no conversion capability. If an ASIN has exposure and clicks but no conversions under a certain keyword, its traffic competitiveness is set to 0. No conversions mean no orders and no effective traffic acquisition capability.

[0028] Based on the obtained traffic competitiveness, a keyword-ASIN traffic weighted matrix (two-dimensional matrix) is constructed. The number of rows in the matrix is ​​equal to the number of ASINs in the same category, M. Each row corresponds to one ASIN, and the row index is set to the unique identifier of the ASIN. The first row is fixed as the target ASIN. The number of columns in the matrix is ​​equal to the number of core keywords. Each column corresponds to one core keyword, and the column index is set to the keyword text. The columns are arranged in descending order according to the traffic contribution ratio of the target ASIN to the keyword, highlighting the priority of the core keywords. The calculated traffic competitiveness value is then filled into the cell of the current ASIN row and the current keyword column in the matrix.

[0029] S3. For the target ASIN and any other ASIN, iterate through each keyword in the core keyword library, calculate the traffic competition similarity based on predefined rules, and obtain a list of competition similarities between the target ASIN and all other ASINs of the same product. See Figure 2 As shown, the total search volume of all core keywords is calculated by using the total search volume of each core keyword on the platform. Then, the normalized global traffic weight is obtained by dividing the total search volume of each individual keyword by the sum of the total search volumes. The specific expression is as follows: ; In the formula: Indicates global traffic weight; This represents the total number of searches for the core keyword k on the platform. This represents the sum of platform search volumes for all keywords in the core keyword library; Extract the traffic competitiveness of the target ASIN and other ASINs under keyword k from the two-dimensional matrix. Take the smaller of the two values ​​and divide it by the sum of the larger value and the coefficient to avoid zero in the denominator. The traffic overlap is obtained as follows: ; In the formula: This indicates the traffic competitiveness of ASIN a under keyword k; This indicates the traffic competitiveness of ASIN b for keyword k; To avoid coefficient values ​​with a denominator of zero; Indicates the degree of traffic overlap, if If the traffic competitiveness is exactly the same, then If one side's traffic is 0 or ,but .

[0030] The contribution of a keyword to the competitive similarity between two keywords is obtained by multiplying the global traffic weight of the keyword by the traffic overlap. ; In the formula: This represents the contribution value of a single keyword, i.e., the contribution of keyword k to the similarity between ASINs a and b; Then, the contribution values ​​of all keywords in the core keyword library K are summed to obtain the total traffic competition similarity of ASINs a and b. ; If and ,but ;like or ,but ;like If it is less than 0.005, then ; Based on the above steps, the traffic competition similarity of the target ASIN a with all other ASINs is summarized to form a target-ASIN-competition similarity list.

[0031] In this embodiment, the similarity calculation no longer relies on the number of keywords, but is based on the actual traffic overlap of high-value keywords. For example, the target ASIN shares two high-traffic keywords with competitor A (total weight of 0.6), with an overlap of 0.8 and a total similarity of 0.48. It shares 10 low-traffic keywords with competitor B (total weight of 0.1), with an overlap of 0.5 and a total similarity of 0.05. Based on this, competitor A can be accurately identified as a genuine competitor, while competitor B is a fake competitor. This also avoids the situation where one party has traffic while the other has no traffic, or where low-weight keywords lead to artificially high similarity scores, ensuring that the ASIN's competitive similarity score reflects the true threat level. Secondly, the subsequent Louvain community discovery algorithm relies on the edge nodes between nodes for clustering. If the similarity output by S3 contains a large number of false high values, it will cause the clustered communities to contain a large number of weakly related ASINs, which will increase the subsequent screening cost.

[0032] S4. Using ASIN as the network node and the weight of the connection edge between the nodes as the traffic competition similarity, an ASIN traffic competition network is formed. The Louvain community detection algorithm is used to cluster the competition network to obtain a potential competitor pool. The traffic overlap of each ASIN in the potential competitor pool is calculated, and competitors with high overlap exceeding a predetermined threshold are filtered to obtain a list of high overlap competitors. In this embodiment, ASIN is abstracted as a network node, and the competition intensity between nodes is abstracted as edge weights. An undirected weighted competition network is constructed, and the specific steps are as follows: Node set: ASIN combination ,in For target ASIN; For other ASINs of the same category collected; Edge set: Undirected edges , indicating that there is a competitive relationship between any two different ASINs; where the variable weight is the traffic competition similarity calculated in step S3.

[0033] Then, based on the Louvain community detection algorithm, cluster analysis is performed to divide the competitive group into the core competitive group and the potential competitor pool. The specific steps are as follows: Initialize community partitioning: Treat each ASIN as an independent community, and the community set is as follows: ; Calculate modularity: ; In the formula: , which represents the sum of the weights of all edges in the network; , representing a node The weighting degree; Represents a node The community mentioned above; Denote the Kronecker function, if , representing a node and nodes If they belong to the same community, then Otherwise, it is 0; This represents the modularity, with a value ranging from -1 to 1. A larger value indicates a more tightly connected node network within the community. Represents a node and nodes Edge weights between them; Represents a node The weighting degree; Represents a node The community to which it belongs.

[0034] Local optimization: for each node Move the node from its current community to the community of its neighboring node (a node directly connected by an edge). Calculate the modularity gain after the move. Select the community that maximizes the modularity gain. If the modularity gain is greater than 0, move the node; otherwise, retain the original community. Repeat this process until all nodes are no longer moved.

[0035] Community aggregation: Treat each community as a super node, calculate the edge weights between super nodes (the sum of edge weights of all node pairs within two communities), and form a simplified network.

[0036] Iteration Termination: Repeat the steps from local optimization to community aggregation until the modularity no longer increases significantly. If the increase in modularity is less than a predetermined threshold, the final community division is obtained.

[0037] In the final community segmentation results, the unique community containing the target ASIN is found, and all ASINs in this unique community other than the target ASIN are considered as potential competitors closely related to the target ASIN.

[0038] Next, by calculating the traffic overlap between the target ASIN and potential competitors, highly overlapping competitors vying for core traffic are identified. The specific steps are as follows: First, calculate the total effective traffic for the target ASIN under the core keyword library K: ; In the formula: With the preceding formula The calculation formula is the same; Indicates the target ASIN Total effective traffic under the core keyword library K; Calculate common keyword traffic for each ASIN in the potential competitor pool, where the common keyword set is: The target keywords have traffic for both the ASIN and b. ; In the formula: This represents the total traffic contribution of common keywords to the target ASIN.

[0039] based on and Calculate the traffic overlap between the target ASIN and competitor b: ; In the formula: This indicates the degree of traffic overlap between the target ASIN and competitor b; Set an overlap threshold based on the business scenario. For example, if the overlap threshold is 15%, when the traffic overlap between the target ASIN and competitor b exceeds 15%, competitor b is considered a high overlap competitor, and a set of high overlap competitors is obtained based on this.

[0040] In this embodiment, community partitioning based on network structure can group directly and indirectly competing ASINs into the same potential competitor pool, avoiding the splitting of core competitors. For example, in the competitive chain of ABC, K-means might group A into one group and BC into another, while the Louvain community discovery algorithm groups ABC into the same group. Secondly, it can focus on high-threat competitors. For example, if there are 100 ASINs in the potential competitor pool, only 20 ASINs with an overlap greater than 15% are retained after traffic overlap filtering, so that merchants do not have to waste monitoring resources on 80 weakly related ASINs.

[0041] S5. Repeat steps S3 to S4 within a predetermined time window, calculate the trend of similarity score changes between the target ASIN and highly overlapping competitors within each window, distinguish between stable competitors and temporary competitors. For stable competitors, calculate the difference in traffic share between the two on common keywords, and filter keywords with a difference in share greater than a predetermined threshold as disadvantageous keywords for the target ASIN.

[0042] In this embodiment, the basic time window unit is a week, and the sliding step is 1 week. Steps S3 to S4 are repeated for each window, targeting only the target ASIN and the highly overlapping competitor pool, avoiding full calculation. Stable competitor identification: Set Let ASIN a be the similarity of traffic competition between target ASIN a and competitor b at time window t. For a competitor to be considered a stable competitor, it must meet the following two conditions: Similarity threshold condition: The similarity of traffic competition over three consecutive time windows is greater than or equal to the threshold. ; Fluctuation stability condition: The similarity fluctuation within three consecutive windows does not exceed 10%; the fluctuation calculation formula is as follows: ; In the formula: This represents the average similarity of traffic contention within three consecutive windows. , and The similarity value of traffic contention within three consecutive windows; Temporary competitor identification: A competitor that meets both the sudden increase condition and seasonal keyword verification is a temporary competitor. The surge condition is when the traffic competition similarity between the competitor and the target ASIN within a predetermined window is greater than or equal to a predetermined similarity threshold, and the growth rate is greater than or equal to a predetermined growth rate threshold. The seasonal keyword verification involves extracting keywords that cause a sudden increase in the similarity of the current window, and verifying the presence of nodular or event-related keywords among the seasonal keywords.

[0043] Next, for stable competitors, identify the target ASIN's weak keywords in shared keywords. The specific steps are as follows: For each stable competitor b and target ASIN a, analyze their shared keyword set and calculate the traffic share difference for each shared keyword k. : ; In the formula: This represents the percentage of traffic contribution of keyword k to competitor b, i.e., a traffic competitiveness indicator. This indicates the percentage of traffic contribution of keyword k to target ASIN a, i.e., the traffic competitiveness index. Based on the set percentage difference threshold, the traffic percentage difference is filtered out. Keywords with a percentage difference greater than or equal to the threshold indicate that the competitor has a significant advantage and there is a clear traffic gap with the target ASIN, which are considered disadvantageous keywords. The selected weak keywords are categorized as follows: Sort in descending order. The larger the value, the more obvious the competitor's advantage, and the more likely it is to be included in the optimization scope.

[0044] In this embodiment, steps S3-S4 are repeated within a predetermined time window to observe the trend of similarity changes, thereby identifying stable and temporary competitors. This allows for subsequent monitoring of the corresponding competitors based on the actual situation. For example, temporary competitors do not require long-term monitoring resources, while stable competitors can be analyzed with concentrated resources, reducing operating costs. Furthermore, the output of disadvantageous keywords enables merchants' optimization strategies to achieve precise targeting, improving the effectiveness of strategy implementation.

[0045] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for product traffic analysis and competitor identification based on ASINs and keywords, characterized in that, Includes the following steps: S1: Collect keyword traffic data of the target ASIN and similar ASINs, and preprocess the collected data to obtain a standardized core keyword library, traffic data of the target ASIN and similar products in the core keyword library. S2: Calculate the traffic competitiveness for each combination of ASIN and core keywords, and then construct a keyword-ASIN traffic weighted matrix. The core keyword library is used as the columns of the matrix and ASIN is used as the rows of the matrix, forming a two-dimensional matrix with the number of ASINs in the rows and the number of core keywords in the columns. Each element value in the matrix corresponds to the traffic competitiveness value of the ASIN under the corresponding keyword. S3. For the target ASIN and any other ASIN, iterate through each keyword in the core keyword library, calculate the traffic competition similarity based on predefined rules, and obtain a list of competition similarities between the target ASIN and all other ASINs of the same product. S4. Using ASIN as the network node and the weight of the connection edge between the nodes as the traffic competition similarity, an ASIN traffic competition network is formed. The Louvain community detection algorithm is used to cluster the competition network to obtain a potential competitor pool. The traffic overlap of each ASIN in the potential competitor pool is calculated, and competitors with high overlap exceeding a predetermined threshold are filtered to obtain a list of high overlap competitors. S5. Repeat steps S3 to S4 within a predetermined time window, calculate the trend of similarity score changes between the target ASIN and highly overlapping competitors within each window, distinguish between stable competitors and temporary competitors. For stable competitors, calculate the difference in traffic share between the two on common keywords, and filter keywords with a difference in share greater than a predetermined threshold as disadvantageous keywords for the target ASIN.

2. The method for product traffic analysis and competitor identification based on ASIN and keywords according to claim 1, characterized in that, The keyword traffic data for the target ASIN and similar ASINs includes the ASIN's unique identifier, keyword text, search volume, click-through rate, conversion rate, and timestamp.

3. The method for product traffic analysis and competitor identification based on ASIN and keywords according to claim 1, characterized in that, The calculation steps for the traffic competitiveness are as follows: The traffic competitiveness of any ASIN and any core keyword is the ASIN's exposure under the corresponding keyword multiplied by the click-through rate and then multiplied by the conversion rate. Each of these ASINs belongs to the same category ASIN set, and each core keyword belongs to the core keyword library.

4. The product traffic analysis and competitor identification method based on ASIN and keywords according to claim 1, characterized in that, Step S3 includes the following steps: The total search volume of all core keywords is calculated by taking the total search volume of each core keyword on the platform. Then, the normalized global traffic weight is obtained by dividing the total search volume of each individual keyword by the sum of the total search volumes. Extract the traffic competitiveness of the target ASIN and other ASINs under keyword k from the two-dimensional matrix, and take the smaller value of the two and divide it by the sum of the maximum value and the coefficient value to avoid the denominator being zero to obtain the traffic overlap. The contribution of a keyword to the competitive similarity between the two is obtained by multiplying the global traffic weight of the keyword by the traffic overlap. The total competitive similarity is obtained by summing the contributions of the similarity of all keywords, and based on this, a list of competitive similarities between the target ASIN and all other ASINs of the same product is obtained.

5. The product traffic analysis and competitor identification method based on ASIN and keywords according to claim 4, characterized in that, In step 3, if two ASINs have no traffic under a certain keyword, then the keyword's contribution to the competitive similarity of the two ASINs is 0. If only one ASIN has traffic under a certain keyword, then the keyword contributes 0 to the competitive similarity between the two ASINs. If an ASIN has a global traffic weight less than a preset threshold under a certain keyword, then the keyword's contribution to the competitive similarity between the two ASINs is 0.

6. The product traffic analysis and competitor identification method based on ASIN and keywords according to claim 1, characterized in that, Step S4 includes the following steps: Construct an ASIN traffic competition network: Treat each ASIN as an independent node, including the target ASIN and other ASINs of the same category. If the traffic competition similarity between two ASINs is greater than 0, then establish an edge between the nodes. The weight of the edge adopts the value of the traffic competition similarity, and retain the edges with the top 20% weights. Initialize community partitioning: Treat each ASIN as an independent community to form an initial community set; Calculate modularity: Determine the sum of all edge weights in the ASIN traffic contention network, then calculate the sum of the contention strengths of each ASIN with all other ASINs, and then combine the Kronecker function to obtain the Q value of the current community division in the modularity calculation formula. Local optimization: For each ASIN, move it sequentially from its current community to the community of the adjacent ASIN, calculate the modularity gain after each move, retain the move operation with the maximum modularity gain that is greater than 0, and repeat the local optimization until all ASINs no longer move. Community aggregation: Each community, after local optimization, is treated as a super node, and the edge weights between different super nodes are calculated to form a simplified network; Iteration Termination: Repeat the local optimization to community aggregation process until the improvement in modularity is less than a predetermined threshold, then output the final community partitioning result; Locating the potential competitor pool: Based on the final community segmentation results, find the unique community containing the target ASIN. All ASINs in the unique community other than the target ASIN are the potential competitor pool. Traffic overlap filtering: Calculate the traffic overlap for each ASIN in the potential competitor pool, filter out competitors with high overlap (above a predetermined threshold), and obtain a list of highly overlapping competitors.

7. The product traffic analysis and competitor identification method based on ASIN and keywords according to claim 6, characterized in that, The traffic overlap filtering includes the following steps: The total effective traffic for the target ASIN is obtained by summing the traffic competitiveness values ​​of all core keywords. For each ASIN in the potential competitor pool, identify common keyword combinations, and then add up the traffic competitiveness of the target ASIN corresponding to all keywords in the set to obtain the total traffic contribution of the common keywords to the target ASIN. The percentage of total traffic contribution from common keywords to the target ASIN relative to the total effective traffic of the target ASIN is used as the traffic overlap. Based on the set traffic overlap threshold and the calculated traffic overlap, competitors with traffic overlap exceeding the threshold are filtered out, resulting in a list of highly overlapping competitors.

8. The product traffic analysis and competitor identification method based on ASIN and keywords according to claim 1, characterized in that, The identification of stable competitors includes the following steps: A competitor that simultaneously meets both the similarity threshold condition and the volatility stability condition is considered a stable competitor. The similarity threshold condition is that the traffic competition similarity of three consecutive windows in the highly overlapping competitor list is greater than or equal to the preset similarity threshold. The fluctuation stability condition is that the fluctuation of the traffic competition similarity of the three windows is less than or equal to a predetermined threshold.

9. A product traffic analysis and competitor identification method based on ASIN and keywords according to claim 8, characterized in that, Obtaining the negative keywords for the target ASIN includes the following steps: Extract the shared keyword set between the target ASIN and a single stable competitor. For each shared keyword, calculate the difference in traffic share between the competitor and the target ASIN. Then, based on the set threshold for the difference in traffic share, filter out keywords whose traffic share difference is greater than or equal to the threshold as disadvantageous keywords.

10. The product traffic analysis and competitor identification method based on ASIN and keywords according to claim 1, characterized in that, The temporary competitor identification includes the following steps: If a competitor meets both the surge condition and the seasonal keyword verification, it is considered a temporary competitor. The surge condition is when the traffic competition similarity between the competitor and the target ASIN within a predetermined window is greater than or equal to a predetermined similarity threshold, and the growth rate is greater than or equal to a predetermined growth rate threshold. The seasonal keyword verification involves extracting keywords that cause a sudden increase in the similarity of the current window, and verifying the presence of nodular or event-related keywords among the seasonal keywords.

Images