Interactive comparative analysis platform and method for character relation graph of Chinese and Japanese literature

Abstract

The invention relates to the technical field of cross-culture literature atlas analysis, and particularly discloses an interactive comparative analysis platform and method for a Chinese and daily literature character relation atlas, and the method comprises the steps: collecting and integrating original texts, historical versions and translation data of Chinese and daily literature works, and constructing a structured text corpus; a convolutional neural network is adopted to calculate version variation feature values, and semantic offset degrees between different versions are quantified; constructing a vocabulary alignment matrix in combination with a pre-trained cross-language word vector model, extracting a semantic consistency score and a cultural image retention rate, and further calculating a cultural adaptation deviation feature value; fusing the two types of features into a composite credibility evaluation vector, inputting the composite credibility evaluation vector into a cross-culture text reliability evaluation model based on a random forest, and outputting a credibility score of character association; finally, differentiated visual display is achieved according to the credibility, and the functions of screening according to the credibility, version path tracking, cultural deviation thermodynamic diagram mapping and the like are supported.

Description

Technical Field

[0001] This invention relates to the field of cross-cultural literary mapping technology, specifically to an interactive comparative analysis platform and method for character relationship mapping in Chinese and Japanese literature. Background Technology

[0002] With the deepening of cultural exchanges between China and Japan, the mutual translation, dissemination, and reception of literary works between the two countries have become important topics in cross-cultural studies. Traditionally, the study of relationships between literary characters has relied mainly on close reading of texts and manual induction, which is insufficient to cope with the complexity and diversity brought about by massive amounts of textual data. In recent years, knowledge graph technology has been introduced into the field of literary studies, providing new methodological support for constructing structured networks of character relationships. However, in the process of constructing character relationship graphs in Chinese and Japanese literature, significant deviations in semantic transmission and cultural adaptation occur due to language differences, version evolution, and different cultural contexts, affecting the accuracy and reliability of graph nodes and edge relationships. Existing technologies mostly focus on rule-based or statistical methods for entity recognition and relationship extraction, lacking systematic modeling of semantic shifts and cultural variations in the translation process, making it difficult to meet the needs of constructing high-quality graphs across languages, versions, and cultural contexts.

[0003] The existing technology has the following shortcomings:

[0004] Current technologies generally overlook the deep semantic shifts caused by version evolution in literary texts throughout history. This means that textual variations in the same work across different historical periods, publishers, or commentators may not result in obvious lexical changes, but can fundamentally alter the implicit expressions of character relationships. For example, a character who is a core ally of the protagonist in the first edition may be relegated to a peripheral role in a later revised edition. Such changes often cannot be detected through traditional keyword matching or syntactic analysis; they require a combination of contextual semantic modeling and version evolution path analysis. This problem is not effectively modeled in most current literary graph construction systems, leading to the risk of "false connections," "missed connections," or "misjudgments" in contexts of version variation, severely limiting the depth and credibility of their application in academic research and cultural heritage protection. Summary of the Invention

[0005] The purpose of this invention is to provide an interactive comparative analysis platform and method for character relationship maps in Chinese and Japanese literature, in order to solve the problems mentioned above.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] An interactive comparative analysis method for character relationship maps in Chinese and Japanese literature includes the following steps:

[0008] S1: In the process of constructing a map of relationships between Chinese and Japanese literary figures, original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works are collected and integrated in real time.

[0009] S2: Calculate version variation feature values ​​based on the content differences and transmission paths between different versions to assess the degree of deviation of a certain version of text from the original version;

[0010] S3: Perform semantic and cultural element comparison analysis on translated text data during the language conversion process. Calculate the cultural adaptation deviation feature value based on the cultural image retention rate and semantic consistency in the translated text. This value is used to identify semantic shifts caused by cultural background differences during the translation process.

[0011] S4: Construct a composite credibility assessment vector from the version variation feature value and the cultural adaptation deviation feature value, and input it into the cross-cultural text reliability assessment model for fusion analysis to assess the credibility level of the association between people in the graph;

[0012] S5: Based on the credibility level, differentiated displays are obtained to achieve multi-dimensional comparative analysis of the relationship between Chinese and Japanese literary figures, thereby improving the depth and accuracy of cross-cultural research.

[0013] As a further aspect of the present invention: the evaluation of the degree of deviation of a certain version of text from the original version specifically includes:

[0014] In the process of constructing a map of relationships between Chinese and Japanese literary figures, the original texts and historical version evolution records of literary works from both countries are collected and integrated in real time. Based on the content differences and transmission paths between different versions of literary works from China and Japan, version variation characteristic values ​​are calculated. It is then determined whether the version variation characteristic value is greater than or equal to a preset threshold. If it is, then a certain version of the text deviates abnormally from the original version; otherwise, the deviation of a certain version of the text from the original version is normal.

[0015] As a further aspect of the present invention: the process of obtaining the version mutation feature value is as follows:

[0016] A convolutional neural network is used to model the text similarity between different versions of text, and based on this model, a similarity score matrix between each version is output, and then version variation feature values ​​are calculated; the specific steps include the following:

[0017] The different versions of the collected Chinese and Japanese literary works were preprocessed to construct a set of word vector sequences;

[0018] Convert the word vector sequence of each version of the text into a fixed-dimensional text matrix;

[0019] Local semantic features are extracted using a one-dimensional convolutional layer and the ReLU activation function, and then the feature vector representation of the version text is generated after max pooling.

[0020] Calculate the cosine similarity between each pair of all versions to obtain the text similarity matrix between versions;

[0021] Select a baseline version and calculate the average similarity deviation between all other versions and the baseline version to obtain the version variation feature value.

[0022] As a further aspect of the present invention: the identification of semantic shifts caused by cultural differences during the translation process specifically includes:

[0023] In constructing a character relationship map of Chinese and Japanese literature, translated text data from literary works in China and Japan are collected and integrated in real time. Semantic and cultural elements of the translated text data are compared and analyzed during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translation, the cultural adaptation deviation feature value is calculated. It is determined whether the cultural adaptation deviation feature value is greater than or equal to a preset threshold. If it is, it is a semantic shift caused by cultural background differences during the translation process. If not, it is a semantic shift caused by cultural background differences outside the translation process.

[0024] As a further aspect of the present invention: the process for obtaining the cultural adaptation deviation feature value is as follows:

[0025] During the translation text integration process, translated text data from Chinese and Japanese literary works are collected and integrated in real time. Semantic and cultural element comparisons are then performed on the translated text data during the language conversion process to calculate cultural adaptation bias characteristic values, specifically including:

[0026] A pre-trained cross-lingual word vector model is used to combine the source language text vocabulary set and the target language translation vocabulary set;

[0027] Constructing a word alignment matrix based on the cosine similarity function;

[0028] For each source language word, calculate its matching score with the most similar word in the target language, i.e. the maximum value in the word alignment matrix, denoted as the semantic consistency score vector. Identify the set of key terms with specific cultural images in the source language text and calculate their average semantic consistency score as the cultural image retention rate.

[0029] The cultural adaptation deviation characteristic value is obtained by calculating the ratio between the cultural image retention rate and the overall semantic consistency mean.

[0030] As a further aspect of the present invention: the construction of a composite credibility assessment vector from version variation feature values ​​and cultural adaptation deviation feature values, and its input into a cross-cultural text reliability assessment model for fusion analysis, specifically includes:

[0031] The version variation feature value and the cultural adaptation deviation feature value are obtained, and a composite credibility evaluation vector is constructed from the version variation feature value and the cultural adaptation deviation feature value. This vector is used as the input of the cross-cultural text reliability evaluation model. The training objective of the cross-cultural text reliability evaluation model is to minimize the error between the predicted cross-cultural text reliability score and the actual cross-cultural text reliability score. Based on the trained model, the cross-cultural text reliability score is output. The cross-cultural text reliability evaluation model is a random forest model.

[0032] As a further aspect of the present invention: the training process of the cross-cultural text reliability assessment model is as follows:

[0033] The composite credibility assessment vector is fed into the random forest model for training. The training dataset consists of multiple labeled samples, each containing a set of composite credibility assessment vectors and a corresponding human-assessed cross-cultural text reliability score. During training, the objective function is to minimize the mean squared error between the predicted score and the actual score. The model performance is optimized by continuously adjusting the model parameters, including the number of decision trees and the maximum depth. Finally, after the model training is completed, when a new translated text is input, the system automatically extracts its version variation feature value and cultural adaptation bias feature value, constructs a composite credibility assessment vector, and inputs it into the trained random forest model to output the cross-cultural text reliability score of the corresponding translated text.

[0034] As a further aspect of the present invention: the credibility level of the association between individuals in the evaluation map specifically includes:

[0035] Determine whether the cross-cultural textual reliability score of the relationships between people in the graph is greater than or equal to a preset threshold. If yes, the relationships between people in the graph are reliable; otherwise, the relationships between people in the graph are unreliable.

[0036] As a further aspect of the present invention: the method of obtaining differentiated displays based on credibility levels to achieve multi-dimensional comparative analysis of the relationships between Chinese and Japanese literary figures specifically includes:

[0037] Based on the credibility score output by the cross-cultural text reliability assessment model, the character nodes and related edges in the graph are visualized and rendered in a hierarchical manner according to their credibility level. Credible character relationships are displayed in a highlighted or bolded form, while untrustworthy relationships are presented in a dashed or grayed-out form. The user interface provides functions such as filtering by credibility threshold, tracking version variation paths, and mapping cultural adaptation deviation heatmaps, enabling researchers to dynamically observe the changing trends of character relationships and their cultural semantic shifts under different versions and translation backgrounds.

[0038] An interactive comparative analysis platform for character relationship maps in Chinese and Japanese literature, including:

[0039] The multi-source text acquisition and integration module collects and integrates original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works in real time during the construction of a relationship map of Chinese and Japanese literary figures.

[0040] The version variation feature calculation module calculates version variation feature values ​​based on the content differences and transmission paths between different versions, which are used to assess the degree of deviation of a certain version of text from the original version.

[0041] The cultural adaptation deviation analysis module compares and analyzes the semantics and cultural elements of the translated text data during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translated text, it calculates the cultural adaptation deviation feature value, which is used to identify the semantic shift caused by cultural background differences during the translation process.

[0042] The composite credibility assessment modeling module constructs a composite credibility assessment vector from version variation feature values ​​and cultural adaptation deviation feature values, and inputs it into the cross-cultural text reliability assessment model for fusion analysis to assess the credibility level of the association between people in the graph.

[0043] The multidimensional visualization and comparative analysis module obtains differentiated displays based on credibility levels, enabling multidimensional comparative analysis of the relationship between Chinese and Japanese literary figures, thereby improving the depth and accuracy of cross-cultural research.

[0044] The beneficial effects of this invention are:

[0045] (1) This invention introduces two core evaluation indicators, version variation feature value and cultural adaptation deviation feature value, to construct a multi-dimensional, cross-language, and cross-version semantic shift quantitative analysis mechanism. This enables systematic modeling and in-depth analysis of semantic variation phenomena in Chinese and Japanese literary works on two key dimensions: text evolution path and language and cultural differences. Among them, the version variation feature value extracts local semantic features of the text based on a one-dimensional convolutional neural network and accurately captures deep semantic shifts caused by version circulation, revision, and reprinting by constructing a cross-version similarity matrix. The cultural adaptation deviation feature value integrates a pre-trained cross-language word vector model and cultural image recognition technology to characterize language conversion distortion caused by cultural background differences during translation from the perspectives of semantic consistency and cultural symbol retention rate. On this basis, this invention further integrates the two types of feature values ​​to construct a composite credibility evaluation vector, which is then fed into a random forest model trained with the goal of minimizing prediction error, to realize the intelligent generation and dynamic evaluation of the reliability scores of character nodes and their relationships in the graph. This mechanism not only effectively improves the structural stability and semantic consistency of the graph in cross-version and cross-language scenarios, but also enhances its explanatory power and adaptability in cross-cultural research at a deeper level, providing solid technical support and methodological innovation for constructing a high-precision and highly interpretable graph of relationships between Chinese and Japanese literary figures.

[0046] (2) This invention not only achieves accurate modeling and credibility assessment of semantic shifts in the character relationship graph of Chinese and Japanese literature at the technical level, but also innovatively integrates visualization analysis and interactive exploration mechanisms at the application level, constructing an intelligent interactive system that integrates multi-dimensional display, dynamic filtering, and path tracking. Specifically, based on the credibility score output by the cross-cultural text reliability assessment model, the system realizes hierarchical rendering of graph nodes and edge relationships, supporting differentiated visualization strategies such as highlighting high-credibility associations and graying out low-credibility connections; at the same time, the system also provides a variety of interactive functions such as filtering by credibility threshold, tracking the evolution path of version variation, and mapping the heat map of cultural adaptation deviation, enabling researchers to dynamically explore the semantic evolution trajectory and cultural adaptation degree of the same character relationship in different historical versions and translation contexts in a unified interface. This interactive design that deeply integrates semantic analysis, knowledge modeling, and user behavior not only significantly enhances the interpretability and operability of the graph system, but also pioneeringly promotes the paradigm shift of comparative studies of Chinese and Japanese literature from static text analysis to dynamic knowledge discovery, providing a new methodological support and technical implementation path for cross-cultural literary research. Attached Figure Description

[0047] The invention will now be further described with reference to the accompanying drawings.

[0048] Figure 1This is a flowchart of the interactive comparative analysis method for the Chinese-Japanese literary character relationship map of the present invention;

[0049] Figure 2 This is a flowchart of the interactive comparative analysis platform for the relationship map of Chinese and Japanese literary figures in this invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figure 1 As shown, this invention is an interactive comparative analysis method for character relationship maps in Chinese and Japanese literature, comprising the following steps:

[0052] S1: In the process of constructing a map of relationships between Chinese and Japanese literary figures, original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works are collected and integrated in real time.

[0053] S2: Calculate version variation feature values ​​based on the content differences and transmission paths between different versions to assess the degree of deviation of a certain version of text from the original version;

[0054] S3: Perform semantic and cultural element comparison analysis on translated text data during the language conversion process. Calculate the cultural adaptation deviation feature value based on the cultural image retention rate and semantic consistency in the translated text. This value is used to identify semantic shifts caused by cultural background differences during the translation process.

[0055] S4: Construct a composite credibility assessment vector from the version variation feature value and the cultural adaptation deviation feature value, and input it into the cross-cultural text reliability assessment model for fusion analysis to assess the credibility level of the association between people in the graph;

[0056] S5: Based on the credibility level, differentiated displays are obtained to achieve multi-dimensional comparative analysis of the relationship between Chinese and Japanese literary figures, thereby improving the depth and accuracy of cross-cultural research.

[0057] In S1, during the construction of a relationship map of literary figures in China and Japan, original texts, historical version evolution records, and translated text data from literary works in both countries are collected and integrated in real time. Specifically, this includes:

[0058] In constructing a graph of relationships between literary figures in Chinese and Japanese literature, this invention utilizes a multi-source heterogeneous data acquisition mechanism to collect and integrate original texts, historical version evolution records, and translated text data from literary works in both countries in real time. Specifically, the system first connects to multiple national-level digital libraries, open access platforms, and literary database resources, such as the National Library of China, the National Diet Library of Japan, CNKI, JSTOR, and Project Gutenberg, using web crawler technology to extract the full-text texts of classic literary works from both China and Japan. Simultaneously, for classical literary works of significant cultural value, the system also accesses bibliographical research databases and manuscript archives to obtain records of version evolution from different historical periods, including first editions, revised editions, translations, and annotated editions.

[0059] Furthermore, in terms of acquiring translated text data, the system calls authoritative translation corpora through API interfaces, such as the Chinese-Japanese Bilingual Translation Corpus, the LDC Corpus, and the OPUS open-source translation corpus project. It also combines the work of a human annotation team to perform secondary proofreading and semantic annotation on the translated versions of representative literary works. All collected text data undergoes a unified cleaning process, including noise reduction, word segmentation, standardization, language recognition, and version annotation, ultimately forming a structured "source text - translated text - version path" triple dataset. This provides a high-quality data foundation for subsequent version variation analysis, cultural adaptation modeling, and character relationship extraction.

[0060] In S2, based on the content differences and transmission paths between different versions, version variation feature values ​​are calculated to assess the degree of deviation of a particular version of text from the original version. Specifically, this includes:

[0061] In the process of constructing a map of relationships between Chinese and Japanese literary figures, the original texts and historical version evolution records of literary works from both countries are collected and integrated in real time. Based on the content differences and transmission paths between different versions of literary works from China and Japan, version variation characteristic values ​​are calculated. It is then determined whether the version variation characteristic value is greater than or equal to a preset threshold. If it is, then a certain version of the text deviates abnormally from the original version; otherwise, the deviation of a certain version of the text from the original version is normal.

[0062] The process for obtaining the version mutation feature value is as follows:

[0063] A convolutional neural network is used to model the text similarity between different versions of text, and based on this model, a similarity score matrix between each version is output, and then version variation feature values ​​are calculated; the specific steps include the following:

[0064] The different versions of the collected Chinese and Japanese literary works were preprocessed to construct a set of word vector sequences;

[0065] Convert the word vector sequence of each version of the text into a fixed-dimensional text matrix;

[0066] Local semantic features are extracted using a one-dimensional convolutional layer and the ReLU activation function, and then the feature vector representation of the version text is generated after max pooling.

[0067] For each pair of versions, cosine similarity is calculated to obtain the text similarity matrix between versions. The calculation expression is as follows:

[0068] ;

[0069] In the formula, This represents the feature vector of version j. Let S represent the feature vector of version k, S represent the text similarity matrix between versions, and j and k represent two different feature vectors;

[0070] Selecting a baseline version Calculate the average similarity deviation between all other versions and the baseline version to obtain the version variation feature value. The calculation expression is:

[0071] ;

[0072] Where N represents the total number of versions, V represents the similarity score between the j-th version and the baseline version, and VD represents the version variation feature value.

[0073] It should be noted that this invention innovatively introduces a one-dimensional convolutional neural network for cross-version text similarity modeling during the calculation of version variation feature values. By automatically extracting local semantic features from the text and constructing a high-precision similarity matrix, it achieves a quantitative assessment of the content evolution paths between different versions of Chinese and Japanese literary works. This method breaks through the traditional version comparison methods based on string matching or word frequency statistics, and can more sensitively capture deep semantic shifts caused by language style, paragraph reconstruction, plot adjustments, etc., thereby improving the accuracy and robustness of version variation analysis. On this basis, by selecting a benchmark version and calculating the average similarity deviation between it and other versions, version variation feature values ​​are further generated. This enables the system to identify text versions that deviate abnormally from the original version under dynamic threshold judgment, providing highly interpretable and culturally sensitive data support for the subsequent credibility assessment of nodes and edges in the character relationship graph, significantly enhancing the structural stability and semantic consistency of the graph in cross-version and cross-cultural contexts.

[0074] In S3, semantic and cultural element comparison analysis is performed on the translated text data during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translated text, cultural adaptation deviation feature values ​​are calculated to identify semantic shifts caused by cultural background differences during the translation process. Specifically, this includes:

[0075] In constructing a character relationship map of Chinese and Japanese literature, translated text data from literary works in China and Japan are collected and integrated in real time. Semantic and cultural elements of the translated text data are compared and analyzed during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translation, the cultural adaptation deviation feature value is calculated. It is determined whether the cultural adaptation deviation feature value is greater than or equal to a preset threshold. If it is, it is a semantic shift caused by cultural background differences during the translation process. If not, it is a semantic shift caused by cultural background differences outside the translation process.

[0076] The process for obtaining the cultural adaptation deviation feature value is as follows:

[0077] During the translation text integration process, translated text data from Chinese and Japanese literary works are collected and integrated in real time. Semantic and cultural element comparisons are then performed on the translated text data during the language conversion process to calculate cultural adaptation bias characteristic values, specifically including:

[0078] A pre-trained cross-lingual word vector model is used to combine the source language text vocabulary set and the target language translation vocabulary set;

[0079] The word alignment matrix is ​​constructed based on the cosine similarity function, and the calculation expression is as follows:

[0080] ;

[0081] In the formula, A represents the word alignment matrix. Represents the vocabulary of the source language text. Indicates the target language vocabulary;

[0082] For each source language word, calculate its matching score with the most similar word in the target language, i.e. the maximum value in the word alignment matrix, denoted as the semantic consistency score vector. Identify the set of key terms with specific cultural images in the source language text and calculate their average semantic consistency score as the cultural image retention rate.

[0083] The cultural adaptation deviation characteristic value is obtained by calculating the ratio between the cultural image retention rate and the overall semantic consistency mean.

[0084] It should be noted that this invention innovatively integrates a pre-trained cross-lingual word vector model and a cultural imagery recognition mechanism in the calculation of cultural adaptation deviation feature values. By quantitatively modeling the semantic alignment and cultural element retention of Chinese-Japanese translated texts at the lexical level, it achieves accurate identification of semantic shifts caused by cultural background differences during the translation process. This method breaks through the traditional translation quality assessment methods based on syntactic structure or surface-level word matching, introducing two key indicators: semantic consistency score vector and cultural imagery retention rate. These indicators characterize the cultural adaptation degree of the translated text from two dimensions: overall semantic fidelity and the effect of conveying specific cultural symbols, respectively. Furthermore, it constructs cultural adaptation deviation feature values ​​through weighted fusion and combines a dynamic threshold judgment mechanism to distinguish whether semantic shifts originate from cultural differences. This provides a highly culturally sensitive and reliable assessment basis for nodes involving cross-lingual associations in the Sino-Japanese literary character relationship graph, significantly improving the semantic expression accuracy and knowledge reasoning ability of the graph in a cross-cultural context.

[0085] In S4, version variation feature values ​​and cultural adaptation deviation feature values ​​are used to construct a composite credibility assessment vector, which is then input into a cross-cultural text reliability assessment model for fusion analysis. This vector is used to assess the credibility level of the associations between figures in the graph, specifically including:

[0086] The version variation feature value and the cultural adaptation deviation feature value are obtained, and a composite credibility evaluation vector is constructed from the version variation feature value and the cultural adaptation deviation feature value. This vector is used as the input of the cross-cultural text reliability evaluation model. The training objective of the cross-cultural text reliability evaluation model is to minimize the error between the predicted cross-cultural text reliability score and the actual cross-cultural text reliability score. Based on the trained model, the cross-cultural text reliability score is output. The cross-cultural text reliability evaluation model is a random forest model.

[0087] The training process of the cross-cultural text reliability assessment model is as follows:

[0088] The composite credibility assessment vector is fed into the random forest model for training. The training dataset consists of multiple labeled samples, each containing a set of composite credibility assessment vectors and a corresponding human-assessed cross-cultural text reliability score. During training, the objective function is to minimize the mean squared error between the predicted score and the actual score. The model performance is optimized by continuously adjusting the model parameters, including the number of decision trees and the maximum depth. Finally, after the model training is completed, when a new translated text is input, the system automatically extracts its version variation feature value and cultural adaptation deviation feature value, constructs a composite credibility assessment vector, and inputs it into the trained random forest model to output the cross-cultural text reliability score of the translated text.

[0089] Determine whether the cross-cultural textual reliability score of the relationships between people in the graph is greater than or equal to a preset threshold. If yes, the relationships between people in the graph are reliable; otherwise, the relationships between people in the graph are unreliable.

[0090] In S5, differentiated presentations are obtained based on credibility levels, enabling multi-dimensional comparative analysis of the relationships between Chinese and Japanese literary figures. This improves the depth and accuracy of cross-cultural research, specifically including:

[0091] Based on the credibility scores output by the cross-cultural text reliability assessment model, the character nodes and related edges in the graph are visualized and rendered hierarchically according to their credibility levels. Credible character relationships are highlighted or bolded, while unreliable relationships are presented as dashed lines or grayscale. The user interface provides functions such as filtering by credibility threshold, tracking version variation paths, and mapping cultural adaptation deviation heatmaps. This allows researchers to dynamically observe the changing trends of character relationships and their cultural semantic shifts under different versions and translation backgrounds. This enables multi-dimensional and multi-level comparative analysis from semantic consistency, cultural adaptability to historical evolution paths, thereby enhancing the explanatory power and application value of the Sino-Japanese literary character relationship graph in cross-cultural research.

[0092] Please see Figure 2 As shown, the interactive comparative analysis platform for character relationship maps in Chinese and Japanese literature is characterized by including:

[0093] The multi-source text acquisition and integration module collects and integrates original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works in real time during the construction of a relationship map of Chinese and Japanese literary figures.

[0094] The version variation feature calculation module calculates version variation feature values ​​based on the content differences and transmission paths between different versions, which are used to assess the degree of deviation of a certain version of text from the original version.

[0095] The cultural adaptation deviation analysis module compares and analyzes the semantics and cultural elements of the translated text data during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translated text, it calculates the cultural adaptation deviation feature value, which is used to identify the semantic shift caused by cultural background differences during the translation process.

[0096] The composite credibility assessment modeling module constructs a composite credibility assessment vector from version variation feature values ​​and cultural adaptation deviation feature values, and inputs it into the cross-cultural text reliability assessment model for fusion analysis to assess the credibility level of the association between people in the graph.

[0097] The multidimensional visualization and comparative analysis module obtains differentiated displays based on credibility levels, enabling multidimensional comparative analysis of the relationship between Chinese and Japanese literary figures, thereby improving the depth and accuracy of cross-cultural research.

[0098] The working principle of this invention: This invention aims to improve the accuracy and explanatory power of modeling literary character relationships in cross-cultural contexts. The method constructs a multi-source heterogeneous text data acquisition mechanism, integrating original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works to form a structured "original text-translated text-version path" triple dataset. Based on this, version variation feature values ​​and cultural adaptation deviation feature values ​​are calculated separately: the former uses a one-dimensional convolutional neural network to extract local semantic features of the text and construct a similarity matrix, thereby quantifying the degree of semantic shift between different versions; the latter constructs a vocabulary alignment matrix based on a pre-trained cross-linguistic word vector model, integrating two dimensions—semantic consistency and cultural imagery retention rate—to identify translation semantic shifts caused by differences in cultural background. Subsequently, the above two types of feature values ​​are used to construct a composite credibility evaluation vector, which is input into a random forest model trained with minimizing prediction error as the objective function, outputting a cross-cultural text reliability score to evaluate the credibility level of character associations in the graph. Ultimately, the system achieves a visually differentiated display based on credibility scores, supporting interactive functions such as highlighting, filtering, path tracing, and heatmap mapping. This enables researchers to conduct in-depth analysis of the similarities and differences in the relationships between Chinese and Japanese literary figures from multiple dimensions, including semantic consistency, cultural adaptability, and historical evolution paths, significantly improving the intelligence level and academic application value of the mapping in cross-cultural research.

[0099] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0100] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0101] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0102] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0103] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. An interactive comparative analysis method for character relationship maps in Chinese and Japanese literature, characterized in that, Includes the following steps: S1: In the process of constructing a map of relationships between Chinese and Japanese literary figures, original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works are collected and integrated in real time. S2: Calculate version variation feature values ​​based on the content differences and transmission paths between different versions to assess the degree of deviation of a certain version of text from the original version; S3: Perform semantic and cultural element comparison analysis on translated text data during the language conversion process. Calculate the cultural adaptation deviation feature value based on the cultural image retention rate and semantic consistency in the translated text. This value is used to identify semantic shifts caused by cultural background differences during the translation process. S4: Construct a composite credibility assessment vector from the version variation feature value and the cultural adaptation deviation feature value, and input it into the cross-cultural text reliability assessment model for fusion analysis to assess the credibility level of the association between people in the graph; S5: Based on the credibility level, differentiated displays are obtained to achieve multi-dimensional comparative analysis of the relationships between Chinese and Japanese literary figures.

2. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 1, characterized in that, The assessment of the degree of deviation of a certain version of text from the original version specifically includes: In the process of constructing a map of relationships between Chinese and Japanese literary figures, the original texts and historical version evolution records of literary works from both countries are collected and integrated in real time. Based on the content differences and transmission paths between different versions of literary works from China and Japan, version variation characteristic values ​​are calculated. It is then determined whether the version variation characteristic value is greater than or equal to a preset threshold. If it is, then a certain version of the text deviates abnormally from the original version; otherwise, the deviation of a certain version of the text from the original version is normal.

3. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 2, characterized in that, The process for obtaining the version mutation feature value is as follows: A convolutional neural network is used to model the text similarity between different versions of text, and based on this model, a similarity score matrix between each version is output, and then version variation feature values ​​are calculated; the specific steps include the following: The different versions of the collected Chinese and Japanese literary works were preprocessed to construct a set of word vector sequences; Convert the word vector sequence of each version of the text into a fixed-dimensional text matrix; Local semantic features are extracted using a one-dimensional convolutional layer and the ReLU activation function, and then the feature vector representation of the version text is generated after max pooling. Calculate the cosine similarity between each pair of all versions to obtain the text similarity matrix between versions; Select a baseline version and calculate the average similarity deviation between all other versions and the baseline version to obtain the version variation feature value.

4. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 1, characterized in that, The identification of semantic shifts caused by cultural differences during the translation process specifically includes: In constructing a character relationship map of Chinese and Japanese literature, translated text data from literary works in China and Japan are collected and integrated in real time. Semantic and cultural elements of the translated text data are compared and analyzed during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translation, the cultural adaptation deviation feature value is calculated. It is determined whether the cultural adaptation deviation feature value is greater than or equal to a preset threshold. If it is, it is a semantic shift caused by cultural background differences during the translation process. If not, it is a semantic shift caused by cultural background differences outside the translation process.

5. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 4, characterized in that, The process for obtaining the cultural adaptation deviation feature value is as follows: During the translation text integration process, translated text data from Chinese and Japanese literary works are collected and integrated in real time. Semantic and cultural element comparisons are then performed on the translated text data during language conversion to calculate cultural adaptation bias characteristic values. Specifically, this includes: A pre-trained cross-lingual word vector model is used to combine the source language text vocabulary set and the target language translation vocabulary set; Constructing a word alignment matrix based on the cosine similarity function; For each source language word, calculate its matching score with the most similar word in the target language, i.e. the maximum value in the word alignment matrix, denoted as the semantic consistency score vector. Identify the set of key terms with specific cultural imagery in the source language text and calculate their average semantic consistency score as the cultural imagery retention rate. The cultural adaptation deviation characteristic value is obtained by calculating the ratio between the cultural image retention rate and the overall semantic consistency mean.

6. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 1, characterized in that, The process of constructing a composite credibility assessment vector from version variation feature values ​​and cultural adaptation deviation feature values, and inputting it into a cross-cultural text reliability assessment model for fusion analysis, specifically includes: The version variation feature value and the cultural adaptation deviation feature value are obtained, and a composite credibility evaluation vector is constructed from the version variation feature value and the cultural adaptation deviation feature value. This vector is used as the input of the cross-cultural text reliability evaluation model. The training objective of the cross-cultural text reliability evaluation model is to minimize the error between the predicted cross-cultural text reliability score and the actual cross-cultural text reliability score. Based on the trained model, the cross-cultural text reliability score is output. The cross-cultural text reliability evaluation model is a random forest model.

7. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 6, characterized in that, The training process of the cross-cultural text reliability assessment model is as follows: The composite credibility assessment vector is fed into the random forest model for training. The training dataset consists of multiple labeled samples, each containing a set of composite credibility assessment vectors and a corresponding human-assessed cross-cultural text reliability score. During training, the objective function is to minimize the mean squared error between the predicted score and the actual score. The model performance is optimized by continuously adjusting the model parameters, including the number of decision trees and the maximum depth. Finally, after the model training is completed, when a new translated text is input, the system automatically extracts its version variation feature value and cultural adaptation bias feature value, constructs a composite credibility assessment vector, and inputs it into the trained random forest model to output the cross-cultural text reliability score of the corresponding translated text.

8. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 1, characterized in that, The credibility level of the associations between individuals in the evaluation map specifically includes: Determine whether the cross-cultural textual reliability score of the relationships between people in the graph is greater than or equal to a preset threshold. If yes, the relationships between people in the graph are reliable; otherwise, the relationships between people in the graph are unreliable.

9. The interactive comparative analysis method for Sino-Japanese literary character relationship maps according to claim 1, characterized in that, The method of obtaining differentiated displays based on credibility levels enables multi-dimensional comparative analysis of the relationships between Chinese and Japanese literary figures, specifically including: Based on the credibility score output by the cross-cultural text reliability assessment model, the character nodes and related edges in the graph are visualized and rendered in a hierarchical manner according to their credibility level. Credible character relationships are displayed in a highlighted or bolded form, while untrustworthy relationships are presented in a dashed or grayed-out form. The user interface provides functions such as filtering by credibility threshold, tracking version variation paths, and mapping cultural adaptation deviation heatmaps, enabling researchers to dynamically observe the changing trends of character relationships and their cultural semantic shifts under different versions and translation backgrounds.

10. An interactive comparative analysis platform for character relationship maps in Chinese and Japanese literature, characterized in that: An interactive comparative analysis method for the Sino-Japanese literary character relationship map as described in any one of claims 1-9 includes: The multi-source text acquisition and integration module collects and integrates original texts, historical version evolution records, and translated text data from Chinese and Japanese literary works in real time during the construction of a relationship map of Chinese and Japanese literary figures. The version variation feature calculation module calculates version variation feature values ​​based on the content differences and transmission paths between different versions, which are used to assess the degree of deviation of a certain version of text from the original version. The cultural adaptation deviation analysis module compares and analyzes the semantics and cultural elements of the translated text data during the language conversion process. Based on the retention rate of cultural imagery and semantic consistency in the translated text, it calculates the cultural adaptation deviation feature value, which is used to identify the semantic shift caused by cultural background differences during the translation process. The composite credibility assessment modeling module constructs a composite credibility assessment vector from version variation feature values ​​and cultural adaptation deviation feature values, and inputs it into the cross-cultural text reliability assessment model for fusion analysis to assess the credibility level of the association between people in the graph. The multidimensional visualization comparison and analysis module obtains differentiated displays based on the credibility level, realizing a multidimensional comparative analysis of the relationship between Chinese and Japanese literary figures.

Images