A method for identifying cross-cultural contextual conflicts

By constructing a cultural context structure and calculating the transmission path, the problem of identifying the relationship between contexts in cross-border e-commerce commodity exports was solved, enabling accurate identification of potential cultural conflicts and improving the accuracy of cross-cultural context conflict identification.

CN123116141APending Publication Date: 2026-11-17GUANGZHOU INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611293800.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies lack the ability to express different contexts holistically when exporting goods through cross-border e-commerce, and cannot depict the relationships between contexts and the transmission process in the cultural environment, making it difficult to identify the implicit cultural conflicts formed by the combination of multiple contexts.

Method used

We construct cultural context structures, generate contextual expressions through text encoding and image feature extraction, expand paths and calculate dissemination paths, and identify potential cultural conflicts.

Benefits of technology

It enables the accurate identification of potential cultural conflict relationships formed by the combined effects of multiple contextual factors, thereby improving the accuracy of cross-cultural contextual conflict identification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116141A_ABST
    Figure CN123116141A_ABST
Patent Text Reader

Abstract

This invention relates to a method for identifying cross-cultural contextual conflicts, comprising: S0: extracting raw data of a target object; S1: preprocessing the raw data to obtain a target object data structure; S2: encoding cultural contextual features into the target object data structure to obtain a target object contextual expression; S3: mapping the target object contextual expression to the cultural context structure to determine an initial contextual position; S4: expanding paths within the cultural context structure based on the initial contextual position to form various contextual paths; S5: selecting contextual paths that meet legal conditions to form a contextual path set; S6: calculating the cultural differences between contextual paths in the contextual path set; S7: analyzing the contextual conflict relationship between each contextual path based on the cultural differences; S8: assessing the risk level based on the contextual conflict relationship. The cross-cultural contextual conflict identification method of this invention has the advantage of high accuracy in cultural conflict analysis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of cultural conflict identification, and in particular to a method for identifying cross-cultural contextual conflicts. Background Technology

[0002] When exporting cross-border e-commerce goods to different countries, it is necessary to make targeted adjustments to the visual elements and content of product information to suit the cultural characteristics of each country, avoiding offense to consumers in the target country and thus integrating the product into the local context. Therefore, it is essential to identify and review cross-cultural risks associated with cross-border e-commerce goods during the export process.

[0003] Currently, the main methods for reviewing product information include keyword matching based on a pre-set sensitive word database, logical judgment of product content based on manually constructed rules, or calculating similarity after converting product information into vector representations. However, these methods are all based on text or single semantics, lacking the ability to express contextual structure, failing to depict the relationships between different contexts, and struggling to reflect the transmission process of context within a cultural environment. They also cannot effectively identify implicit cultural conflicts formed by combinations of multiple contexts. Summary of the Invention

[0004] Based on this, the purpose of this invention is to improve the ability to identify cross-cultural context conflicts and to provide a method for identifying cross-cultural context conflicts.

[0005] A method for identifying cross-cultural contextual conflicts includes the following steps: S0: Extract the raw data of the target object; the raw data includes text information and image data of the target object; S1: Preprocess the raw data to obtain the target object data structure; the preprocessing includes cleaning and structuring. S2: Perform cultural context feature encoding on the target object data structure to obtain the target object context expression; the cultural context feature encoding performs text encoding on the preprocessed target object text information to obtain a text context vector, extracts image features from the target object image data to obtain an image context vector, and fuses the text context vector and the image context vector to finally obtain the target object context expression; S3: Map the contextual expression of the target object to the corresponding context node in the cultural context structure to determine the initial context position; S4: Based on the initial context position, expand the path in the cultural context structure to form various context paths; S5: Filter the context paths according to the preset context relationship constraints to form a set of context paths; the context relationship constraints filter the context paths in which there is a valid context relationship connection between two adjacent context nodes; S6: Calculate the cultural differences between the terminating context nodes of each context path in the context path set; S7: Analyze the contextual difference conflict relationship between the various contextual paths based on the aforementioned cultural differences; S8: Assess the risk level based on the contextual conflict relationship.

[0006] Compared to existing technologies, this invention constructs a cultural context structure and calculates the propagation path, enabling contextual information to be propagated and analyzed within the cultural relationship structure. This allows for the identification of potential cultural conflict relationships formed by the combined effects of multiple contextual factors, resulting in more accurate cultural conflict analysis results than existing technologies that rely on keyword matching, rule base filtering, or similarity calculation.

[0007] Furthermore, feature encoding is performed on the product titles and product descriptions in the product structure data through text encoding. The specific steps are as follows: a. Perform word segmentation on the title and description respectively to obtain a term set; b. Map the term set to a cultural context feature set to obtain the cultural context matching degree corresponding to the term; the cultural context includes symbolic meaning features, emotional association features, and social context features. c. Calculate the context weight of each term based on its cultural context matching degree in the term set; d. The text context vector is obtained by aggregating the context weights of each term and their matching degree with each cultural context.

[0008] Furthermore, the expression for the context weight is: , The expression for aggregation is: , Among them, among them, For the i-th term, Let be the frequency of the i-th term in the target text. To determine the total number of texts in a predefined cultural context text set, For included terms The number of cultural context texts This is the cultural context distinguishing coefficient for this term.

[0009] Furthermore, the target object image data in the target object structure data is feature-encoded through image feature extraction; the image feature extraction analyzes the visual elements in the target object image data and maps them to the cultural context feature set to obtain an image context vector; the visual elements include dominant color information, image symbol elements, and image theme tags.

[0010] Furthermore, in step S3, the specific steps for performing the mapping include: S31: Read the feature values ​​of each context node in the cultural context structure under different cultural context dimensions, and construct the node feature vector; S32: Calculate the contextual expression of the target object and the contextual matching degree between each of the contextual nodes; S33: Select at least one of the context nodes as the starting context position based on the context matching degree of each context node.

[0011] Furthermore, the expression for the context matching degree is: , in, is the weight of the k-th context dimension.

[0012] Furthermore, the contextual path is modeled based on the contextual relationship weights and path structure, and quantified using a path propagation weight function, the expression of which is: , in, For the i-th context path, The weights represent the relationships between adjacent context nodes. For path length, This is the context propagation attenuation coefficient.

[0013] Furthermore, the formula for calculating the cultural differences is as follows: , in, Let k be the eigenvalues ​​of the k-th dimension of the target culture feature vector. The eigenvalue of the k-th dimension of the cultural feature vector to which the context node belongs. Let be the weight of the k-th cultural context feature dimension. It is a non-linear mapping function.

[0014] Furthermore, the expression for risk assessment is: , in, This represents the path weight.

[0015] Based on the same inventive concept, the present invention also includes an electronic device comprising a processor; a memory for storing a computer program executed by the processor; wherein, when the processor executes the computer program, it implements the cross-cultural context conflict identification method described above.

[0016] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0017] Figure 1 This is a flowchart of the identification method of the present invention. Detailed Implementation

[0018] This invention addresses the shortcomings of existing risk identification technologies, such as their lack of overall contextual expression and inability to depict inter-contextual relationships and transmission mechanisms within cultural environments. It proposes a cross-cultural contextual conflict identification method based on the contextual expression of a product and the termination categories of multiple contextual transmission paths. This method can fully quantify the intensity of contextual conflict, thereby accurately identifying potential cultural misunderstandings and conflict risks that a product may generate in the target cultural environment.

[0019] Specifically, please refer to Figure 1 , Figure 1 This is a flowchart of the identification method of the present invention. A cross-cultural context conflict identification method includes: S0: Extract the raw data of the target object. The raw data includes the target object's text information and image data. The target object's text includes a title, description, and target sales market. In this embodiment, taking a product as the target object as an example, the target object's text includes the product title, product description, and target sales market, and the target object's image data is a product image.

[0020] S1: Preprocess the raw data to obtain the target object data structure. The preprocessing includes cleaning and structuring. Specifically, cleaning involves removing invalid characters from the target object text, standardizing it by ensuring consistent encoding format and text segmentation and keyword extraction; and performing format validation on the target object image data. The structuring process encapsulates the cleaned raw data into a uniformly formatted data structure.

[0021] S2: Cultural context feature encoding is performed on the target object data structure to obtain the target object context expression. The cultural context feature encoding involves text encoding of the preprocessed target object text information to obtain a text context vector, extracting image features from the target object image data to obtain an image context vector, and fusing the text context vector and the image context vector to finally obtain the target object context expression; the target object context expression represents the state distribution of the product in cultural context features. The cultural context feature encoding includes text encoding and image feature extraction, and this encoding process does not involve the construction of a semantic relationship network between words or the generation of text structure, but rather prepares for subsequent processing; wherein, the text encoding performs feature encoding on the title and description, specifically including: a. Perform word segmentation on the title and description respectively to obtain a term set; b. Map the term set to a cultural context feature set to obtain the cultural context matching degree corresponding to each term. The cultural context includes symbolic features such as power, auspiciousness, and danger; emotional association features such as joy, fear, and solemnity; and social context features such as religion, taboo, and status symbol. The cultural context feature set includes a cultural context text set and a cultural context image set, and each cultural context has a corresponding preset cultural context feature value in the cultural context feature set. The cultural context text set consists of a preset series of cultural context texts. Mapping the terms to the corresponding cultural context maps the product information to a predefined cultural context feature space to represent the contextual state of the product in different cultural environments. Each term has a different cultural context matching degree corresponding to different cultural contexts, thereby measuring the cultural context of the term.

[0022] c. Calculate the context weight of each term based on its matching with the cultural context in the term set. Specifically, the cultural discrimination ability of a term is measured by statistically analyzing the ratio between the total number of cultural context texts and the number of cultural context texts containing each term. The context weight is then assigned to each term based on this cultural discrimination ability, reducing the influence of terms that commonly appear in multiple cultural contexts and enhancing the weight of terms with cultural discrimination characteristics. The expression for the context weight is: , in, For the i-th term, Let be the frequency of the i-th term in the target text. To determine the total number of texts in a predefined cultural context text set, For included terms The number of cultural context texts represents the prevalence of a term in the corresponding cultural context; This is the cultural context differentiation coefficient for the term, used to reduce the influence of terms that are common in multiple cultural contexts, thereby enhancing the weight of terms with cultural differentiation characteristics.

[0023] d. The text context vector is obtained by aggregating the context weights of each term and their matching degrees with each cultural context. The aggregation expression is: , in, For terms Contextual weight, For terms The degree of cultural context matching with various cultural contexts.

[0024] The image feature extraction analyzes the visual elements in the target object image data and maps them to the cultural context feature set to obtain an image context vector; the visual elements include dominant color information, image symbol elements, and image theme tags.

[0025] The fusion calculation involves weighted fusion of the text context vector and the image context vector to obtain the contextual representation of the target object. The expression for this fusion calculation is as follows: , in, and These are weighted coefficients, and their sum is 1. For text context vectors, This is the image context vector.

[0026] S3: Map the contextual expression of the target object to the corresponding context nodes in the cultural context structure to determine the initial context position. The cultural context structure includes a set of context nodes and a set of context relationships; each context node represents a cultural context element, and its symbolic meaning, social context labels, and emotional associations within the cultural environment are described by node feature values; each context relationship has a corresponding relationship weight to represent the strength of the contextual association between two context nodes. The larger the relationship weight, the stronger the association between the context nodes, and the greater its contribution in the process of propagating and forming contextual paths within the cultural context structure. The context relationship weight is obtained based on historical context data statistics and is calculated by combining the semantic contribution relationship of context elements in the target cultural environment, the frequency of cultural associations, and user feedback information; the historical context data in the cross-border e-commerce field includes cross-border e-commerce product review records, user behavior data, cultural content corpus data, or public semantic data resources.

[0027] The specific steps involved in the mapping process include: S31: Read the feature values ​​of each context node in the cultural context structure under different cultural context dimensions, and construct the node feature vector. The cultural context dimensions include symbolic meaning dimension, social context dimension, and emotional association dimension.

[0028] S32: Calculate the contextual matching degree between the target object's contextual expression and each of the contextual nodes. The contextual matching degree is the contextual distance, which in this embodiment is calculated using weighted Euclidean distance, and its expression is: , in, Let be the weight of the k-th context dimension, and n be the number of cultural context feature dimensions. Let k be the weight of the k-th cultural context dimension. Let be the feature value of the product context expression vector in the k-th cultural context dimension. Let be the feature value of the j-th context node in the k-th cultural context dimension.

[0029] S33: Select at least one of the context nodes as the starting context position based on the context matching degree of each context node. For example, selection can be made through Top-K selection, distance threshold filtering, or a selection mechanism based on probability distribution. In this embodiment, the top K context nodes with the highest to lowest context matching degree are selected as the starting context position, and K is greater than or equal to 1.

[0030] S4: Based on the initial context position, path expansion is performed within the cultural context structure to form various context paths. Each context relationship in the cultural context structure is assigned a corresponding context relationship weight to represent the strength of the contextual association between two context nodes. The context path characterizes the propagation process of the target object's context within the cultural context structure; this propagation process is modeled based on the context relationship weights and path structure, and quantified using a path propagation weight function to reflect the propagation intensity and evolutionary characteristics of context in different cultural environments.

[0031] The expression for the propagation weight function is: , in, For the i-th context path, The weights represent the relationships between adjacent context nodes. For path length, This is the contextual propagation attenuation coefficient. The contextual propagation attenuation coefficient adjusts the degree of attenuation of context during cross-cultural communication. The greater the cultural difference, the higher the value and the faster the propagation attenuation; the smaller the cultural difference, the lower the value, thus having a wider scope of contextual propagation.

[0032] S5: Filter the context paths according to preset contextual relationship constraints to form a set of context paths. For any sequence of context nodes in a context path, if it meets the contextual relationship constraints, the context path is considered valid; the contextual relationship constraints filter context paths in which there is a valid contextual relationship connection between two adjacent context nodes, and its expression is: , in, Let i be the k-th context node in the context node sequence. Let i be the (k+1)th context node in the context node sequence i, and E be the set of context relations in the cultural context structure.

[0033] Furthermore, a propagation depth constraint is simultaneously applied to filter the contextual path. This propagation depth constraint uses the maximum propagation depth and its expression is: , Where I represents the propagation depth, This represents the maximum propagation depth.

[0034] S6: Calculate the cultural differences between the terminating context nodes of each context path in the context path set. The terminating context node is the context node at the end of the context path, and the cultural difference is the distance between the terminating context nodes. The formula for calculating the cultural difference is: , in, Let k be the eigenvalues ​​of the k-th dimension of the target culture feature vector. The eigenvalue of the k-th dimension of the cultural feature vector to which the context node belongs. Let be the weight of the k-th cultural context feature dimension. It is a nonlinear mapping function. The target cultural feature vector describes the distribution of contextual features in the target cultural environment constructed from the cultural context data of the target market; the cultural context data includes product content data, user behavior data, local cultural corpus data, and social media text data of the target market; the target cultural feature vector is obtained by statistically analyzing the contextual elements in the cultural context data, extracting the distribution features on the preset contextual feature dimensions, and mapping them into vector components.

[0035] The cultural feature vector to which the context node belongs describes the distribution of contextual features of the context node in a specific cultural environment, and is constructed based on the cultural semantic information corresponding to the context node. The cultural semantic information includes the cultural label information corresponding to the context node, the frequency of occurrence of the context node in different cultural corpora, the co-occurrence relationship with other context nodes, and the socio-semantic attributes of emotional tendency; the cultural feature vector to which the context node belongs is obtained by vectorizing the cultural semantic information.

[0036] Furthermore, this application automatically adjusts the context propagation attenuation coefficient, which represents the degree of context propagation attenuation, based on the aforementioned cultural distance and different cultural environments, thereby adjusting the propagation weight and propagation path length, thus making the language propagation model more consistent with the characteristics of cross-cultural context propagation. Specifically, the formula for calculating the context propagation attenuation coefficient is as follows: , in, This is a cultural distance mapping function.

[0037] During path expansion, the process is triggered from each starting node in the context structure graph, and the nodes are expanded according to the context relationship edges to form a context propagation path.

[0038] S7: Analyze the contextual conflict relationships between the contextual paths based on the aforementioned cultural differences. For any two contextual paths, if the cultural context categories corresponding to the terminating nodes are different, then a contextual conflict is determined to exist between these two contextual paths, expressed as: ; S8: Assess the risk level based on the aforementioned contextual conflict relationships. The expression for risk assessment is: , in, This represents the path weight.

[0039] In another embodiment, the relation weight is obtained by normalizing the frequency of occurrence of the context elements corresponding to the context nodes in the target cultural environment and combining it with the context association strength.

[0040] In another embodiment, the target market cultural feature vector is calculated by statistically analyzing the frequency of occurrence, co-occurrence relationships, and emotional tendencies of different contextual elements in the target cultural environment, and then mapped to the same contextual feature space through normalization. The cultural feature vector to which the context node belongs is normalized by normalizing the distribution characteristics of contextual elements in different cultural environments, thereby enabling different contextual nodes to be compared and calculated in the same contextual feature space.

[0041] Taking the product "Red Dragon Pattern T-shirt" as an example, after extracting terms from the product title, we obtained the terms "red," "dragon pattern," and "T-shirt," which were then mapped to a set of cultural contexts {auspicious, power, danger, religion, daily consumption}. The specific mapping results are shown in the table below:

[0042] After calculation, the contextual weight of the term "red" was found to be 0.6, the weight of the term "dragon pattern" was 0.9, and the weight of the term "T-shirt" was 0.5. The resulting text context vector is... This indicates that, within the context of Eastern culture, the product title has strong auspicious attributes, a clear symbol of power, and implies a certain degree of danger or conflict.

[0043] Simultaneously, after extracting the visual elements from the product's image data, the image was found to contain a red main color and a dragon pattern, which were then mapped to obtain an image context vector. .

[0044] Subsequently, the image context vector and the text context vector are fused together. and The product context expression is obtained by taking values ​​of 0.6 and 0.4 respectively and then calculating.

[0045] Compared to existing technologies, this invention constructs a cultural context structure and calculates the propagation path, enabling contextual information to be propagated and analyzed within the cultural relationship structure. This allows for the identification of potential cultural conflict relationships formed by the combined effects of multiple contextual factors, resulting in more accurate cultural conflict analysis results than existing technologies that rely on keyword matching, rule base filtering, or similarity calculation.

[0046] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of this application. The singular forms “a,” “the,” and “the” used in the embodiments and claims of this application are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that, unless otherwise stated, “a plurality” means two or more; the terms “first,” “second,” “third,” etc., are used only to distinguish and not to describe a particular order or sequence, nor should they be construed as indicating or implying relative importance. The term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items. When the above description relates to drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. In the description of this application, those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0047] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and the present invention also intends to include these modifications and variations.

Claims

1. A method for identifying cross-cultural contextual conflicts, characterized in that, Includes the following steps: S0: Extract the raw data of the target object; the raw data includes text information and image data of the target object; S1: Preprocess the raw data to obtain the target object data structure; the preprocessing includes cleaning and structuring. S2: Perform cultural context feature encoding on the target object data structure to obtain the target object context expression; the cultural context feature encoding performs text encoding on the preprocessed target object text information to obtain a text context vector, extracts image features from the target object image data to obtain an image context vector, and fuses the text context vector and the image context vector to finally obtain the target object context expression; S3: Map the contextual expression of the target object to the corresponding context node in the cultural context structure to determine the initial context position; S4: Based on the initial context position, expand the path in the cultural context structure to form various context paths; S5: Filter the context paths according to the preset context relationship constraints to form a set of context paths; the context relationship constraints filter the context paths in which there is a valid context relationship connection between two adjacent context nodes; S6: Calculate the cultural differences between the terminating context nodes of each context path in the context path set; S7: Analyze the contextual difference conflict relationship between the various contextual paths based on the aforementioned cultural differences; S8: Assess the risk level based on the contextual conflict relationship.

2. The cross-cultural context conflict identification method according to claim 1, characterized in that, The product title and product description in the product structure data are feature-encoded using text encoding. The specific steps are as follows: a. Perform word segmentation on the title and description respectively to obtain a term set; b. Map the term set to a cultural context feature set to obtain the cultural context matching degree corresponding to the term; the cultural context includes symbolic meaning features, emotional association features, and social context features. c. Calculate the context weight of each term based on its cultural context matching degree in the term set; d. The text context vector is obtained by aggregating the context weights of each term and their matching degree with each cultural context.

3. The cross-cultural context conflict identification method according to claim 2, characterized in that, The expression for the context weight is: , The expression for aggregation is: , Among them, among them, For the i-th term, Let be the frequency of the i-th term in the target text. To determine the total number of texts in a predefined cultural context text set, For included terms The number of cultural context texts This is the cultural context distinguishing coefficient for this term.

4. The cross-cultural context conflict identification method according to claim 1, characterized in that, Image feature extraction is used to encode the target object image data in the target object structure data; the image feature extraction analyzes the visual elements in the target object image data and maps them to the cultural context feature set to obtain the image context vector; the visual elements include dominant color information, image symbol elements and image theme tags.

5. The cross-cultural context conflict identification method according to claim 1, characterized in that, In step S3, the specific steps for mapping include: S31: Read the feature values ​​of each context node in the cultural context structure under different cultural context dimensions, and construct the node feature vector; S32: Calculate the contextual expression of the target object and the contextual matching degree between each of the contextual nodes; S33: Select at least one of the context nodes as the starting context position based on the context matching degree of each context node.

6. The cross-cultural context conflict identification method according to claim 5, characterized in that, The expression for the context matching degree is: , in, Let be the weight of the k-th context dimension, and n be the number of cultural context feature dimensions. Let k be the weight of the k-th cultural context dimension. Let be the feature value of the product context expression vector in the k-th cultural context dimension. Let be the feature value of the j-th context node in the k-th cultural context dimension.

7. The cross-cultural context conflict identification method according to claim 1, characterized in that, The contextual path is modeled based on the contextual relation weights and path structure, and is quantified using a path propagation weight function, the expression of which is: , in, For the i-th context path, The weights represent the relationships between adjacent context nodes. For path length, This is the context propagation attenuation coefficient.

8. The cross-cultural context conflict identification method according to claim 1, characterized in that, The formula for calculating the cultural differences is: , in, Let k be the eigenvalues ​​of the k-th dimension of the target culture feature vector. The eigenvalue of the k-th dimension of the cultural feature vector to which the context node belongs. Let be the weight of the k-th cultural context feature dimension. It is a non-linear mapping function.

9. The cross-cultural context conflict identification method according to claim 1, characterized in that, The expression for risk assessment is: , in, This represents the path weight.

10. An electronic device, characterized in that, processor; Memory for storing computer programs executed by the processor; Wherein, when the processor executes the computer program, it implements the cross-cultural context conflict identification method according to any one of claims 1-9.