Historical learning-oriented historical painter dynamic relation network calculation method

Abstract

The invention discloses a historical painter dynamic relation network calculation method for historical learning, which comprises the following steps: firstly, obtaining multi-source ancient literature basic data and event information, realizing data enhancement and event expansion through a large language model, filling a data gap and removing repeated data; constructing a basic relationship strength model (BaseWave) by adopting Beta distribution, constructing an event pulse model (EventPulse) by adopting a Gaussian pulse function, and determining the initial influence of the figure in combination with an extended dynamic PageRank algorithm; splitting the positive and negative relation intensities, carrying out vectorization calculation, and describing the dynamic propagation intensity of the social relation through an adjacent matrix; and finally, outputting dynamic network data including relationship evolution, influence change and culture carrier association. According to the method, the crossing of the relation strength from a static label to a spatio-temporal dynamic curve is realized, the social link effect of a short-term culture interaction effect and a culture carrier is quantified, and a scientific and efficient quantitative analysis tool is provided for history learning and literature creation.

Description

Technical Field

[0001] This invention belongs to the field of computer technology, specifically a method for calculating the dynamic relationship network of painters throughout history for historical learning. It can integrate multidimensional data of multiple literati, quantify the spatiotemporal evolution of the strength of relationships between literati, accurately measure the dynamic influence of key figures, and provide a quantitative analysis tool for historical learning and literary creation. Background Technology

[0002] The social circles of Chinese historical painters constitute a highly complex and diverse network of relationships, interwoven with multiple dimensions including family, culture, and politics. At the family level, blood ties formed through marriage and clan connections served as a crucial foundation for resource sharing and mutual support. At the cultural level, exchanges and interactions through calligraphy, painting, and letters fostered spiritual bonds, becoming core channels for cultural dissemination and intellectual exchange. At the political level, power networks based on relationships among relatives, teachers, and colleagues profoundly influenced the political development of ancient states. These interwoven networks collectively shaped the trajectory of ancient society, occupying a key position in cultural transmission and social evolution, and representing an important area for historical study.

[0003] Current research on historical painter networks faces numerous technical bottlenecks: traditional research methods rely on the manual collection and inductive reasoning of massive amounts of historical materials, which is not only time-consuming and labor-intensive but also prone to deficiencies in data, hindering the formation of rigorous arguments; existing data mining methods cannot efficiently handle the sparsity and non-reproducibility of historical data, making it difficult to meet the needs of dynamic network analysis; furthermore, most studies treat cultural carriers (such as paintings, calligraphy, and letters) as independent objects, neglecting their social attributes as social bonds and failing to quantify their driving role in the relationships between literati. Moreover, existing technologies largely focus on static relationship analysis, failing to capture the dynamic evolution of literati relationship networks over time and revealing the correlation between short-term interaction events and long-term relationship development.

[0004] Therefore, there is an urgent need for a network computing method that can integrate multidimensional data and quantify dynamic relationships and influences to fill the gaps in existing technologies. Summary of the Invention

[0005] To address the technical problems of existing methods in analyzing the relationship networks of historical painters, such as insufficient quantification of dynamic relationships, neglect of the social attributes of cultural carriers, and inefficient handling of data sparsity, this invention provides a method for calculating the dynamic relationship networks of painters throughout history for historical learning. This invention can quantify the spatiotemporal evolution of the strength of relationships among literati, accurately assess the dynamic influence of individuals within relationship networks, reveal the driving mechanism of cultural carriers as social bonds, and provide a scientific and efficient quantitative analysis tool for historical research and literary creation.

[0006] The technical solution adopted in this invention is: A method for calculating the dynamic relationship network of painters throughout history for historical learning, characterized by the following steps: S1. Data Acquisition and Augmentation. Basic data and event information of multiple literati are acquired as raw data. Missing attribute dimensions and event information in the raw data are categorized and organized. A Large Language Model (LLM) is used for attribute augmentation and data imputation to generate structured data, ensuring data integrity and structure. S2. Event Expansion and Validation: A retrieval template is constructed using a large language model to retrieve event information related to the target writer from public literature, forming an expanded event set; attribute enhancement processing is performed on the expanded event set to generate structured event data; Levenshtein distance is used to calculate the similarity between the expanded event data and the existing data, duplicate events are removed, and high-quality data is retained after sampling and verification to complete the event expansion; S3. Basic Relationship and Event Pulse Modeling: For each pair of literary figures, the basic relationship type (such as relatives and friends, teacher and student, political enemies, etc.) is extracted, and the relationship time interval is determined to be the intersection of the birth and death years of the two individuals. A Beta distribution is used as the time evolution shape of the basic relationship to construct a basic relationship strength model (BaseWave). The peak position and evolution trend of the relationship strength are controlled by the shape parameters, and the strength amplitude is optimized by combining the number of interaction events. A Gaussian impulse function is used to define the relationship event pulse (EventPulse). An event pulse strength model is constructed by using event timestamps and influence weights to quantify the instantaneous impact of a single event on the relationship. The classic PageRank algorithm is extended to a time series dynamic algorithm to determine the initial influence of a person at a certain moment based on the individual event pulse. S4. Calculation of Dynamic Relationship Strength and Propagation Strength: Combining the basic relationship strength and event impulse strength, construct positive relationship strength according to the two types of interactive events (positive and negative) and the basic relationship type. (t) and the strength of the negative relationship (t); The relationship strength is regarded as a two-dimensional vector, with positive strength as the horizontal axis and negative strength as the vertical axis. The total relationship strength is calculated by the vector magnitude, and the relationship tendency is determined by the vector angle; The adjacency matrix is ​​used to describe the dynamic propagation strength of social relationships. The matrix elements are the total relationship strength between two people at the corresponding time, which are used to define the network edge weights; S5. Dynamic Network Data Output: Combining the strength of dynamic relationships with the intensity of social relationship propagation, and integrating the calculation results of the dynamic influence of individuals, outputs dynamic network data that includes relationship evolution, changes in influence, and connections with cultural carriers, supporting subsequent analysis and applications.

[0007] Furthermore, in step S1, the basic data comes from multiple sources of historical data, including Artyx Cloud, the Chinese Historical Biographical Database, the "Chronological Table of Artists from the Song to the Qing Dynasties", and the National Treasure Museum of China. The large language model uses DeepSeek. When enhancing the attributes of a person, the set of identity attributes is defined as {literati, officials, merchants, others}. The person's occupation information is input through the structured prompt template to achieve automatic classification. When enhancing the attributes of an event, entity extraction, hierarchical classification and impact assessment are completed through the structured prompt template. The hierarchical classification includes macro types {art, politics, social, economic} and micro types {creation, traveling together, apprenticeship, etc., 8 categories}. The impact assessment weight w∈[-1,1].

[0008] Furthermore, in step S2, events with a confidence level below 0.7 are filtered out when the extended event set is generated; the structured extended event data is generated by reusing prompt templates and includes participants, initiators, cultural carriers, locations, event types, and influence weight attributes; high-quality data is retained and erroneous data is corrected during sampling verification.

[0009] Furthermore, in step S3, the expression for the basic relationship strength model is: (1)

[0010] BaseWave uses the probability density function of the beta distribution as the shape term, normalizing the time and shape parameters. Used to control peak position and skewness; the amplitude term employs a sublinear function to enhance the impact of the number of interactive events on the intensity amplitude; the time range is the intersection of the birth and death years of the individuals. Define the duration of the relationship. The list of interaction events between characters is denoted as , The number of events; Different shape parameters are preset for the beta distribution for different labeling relationships k. If no interaction events occur between graphs, the preset parameters of their relationship categories are used directly; if interactive events exist, the parameters are optimized using the weighted negative log-likelihood over the interactive event list E, where each event... All include timestamps and event weight : （2） .

[0011] Furthermore, in step S3, the expression for the relational event impulse model is:

[0012] in, To control the pulse's time decay rate, take , For event timestamps, Weighting of event impact, For the list of interactive events; the initial influence calculation expression is: (4) in, For character nodes, for A list of personal events at any given moment.

[0013] Furthermore, in step S4, the calculation expressions for the strength of the positive relationship and the strength of the negative relationship are as follows:

[0014]

[0015] in, This is the basic relation strength model for the current k-relationship. =0.6, contributing weight to the pulse; It is a positive basic relation type. This is a list of positive events. >0; It is a negative basic relation type. This is a list of negative events. <0; The shape parameter k is the predefined labeling relation k for the beta distribution. The expressions for calculating the total relation strength and relation tendency are:

[0016]

[0017] in, The strength of a negative relationship is as follows: when θ∈(0°,45°), the relationship is positive; when θ∈(45°,90°), the relationship is negative.

[0018] Furthermore, in step S4, the elements of the adjacency matrix A(t) (t) = S(t), representing the node of the person at time t. to v The strength of relational propagation; the influence of dynamic figures is calculated iteratively, and the expression is: (9) Where V represents the set of people, For v The initial influence, k is the relationship identifier, v is the character node, and t is the current time.

[0019] Furthermore, in step S1, the basic data includes character attribute data, relationship data, and cultural carrier data, while the event information includes event time, participants, and event description.

[0020] The core technical concept of this invention lies in the following: First, by integrating multi-source data and enhancing a large language model, the problems of sparsity and insufficient structure of historical data are solved; second, by innovatively combining the BaseWave model with EventPulse, a breakthrough is achieved in transforming relationship strength from "static labels" to "spatiotemporal dynamic curves," considering both the evolutionary trend of long-term stable relationships and capturing the instantaneous impact of short-term interactive events; finally, by using an extended dynamic PageRank algorithm, the influence of personal events and the propagation effect of social relationships are integrated to achieve dynamic evaluation of a person's influence, ultimately outputting dynamic network data that reflects the spatiotemporal evolution of the literary relationship network.

[0021] Compared with the prior art, the beneficial effects of the present invention are reflected in: 1. This invention achieves spatiotemporal dynamic quantification of relationship strength by integrating the BaseWave model and the EventPulse model, breaking through the limitation of traditional methods that can only describe static relationships, and accurately capturing the dual effects of long-term relationship evolution and short-term event impact.

[0022] 2. This invention leverages a large language model to achieve data augmentation and event expansion, effectively solving the problem of historical data sparsity. At the same time, it ensures data reliability through a similarity verification mechanism, providing high-quality data support for dynamic network analysis.

[0023] 3. This invention achieves accurate determination of complex relationship tendencies by splitting and vectorizing the strength of positive and negative relationships, filling the gap in traditional methods that cannot quantify the mixed effects of positive and negative interactions.

[0024] 4. This invention quantifies the driving effect of cultural carrier-related events (such as the gifting of calligraphy and paintings, and academic debates) on social relationships. It achieves quantitative analysis of cultural activities as social bonds through impulse weighting, providing a new path for revealing the connection between cultural dissemination and the evolution of relationship networks.

[0025] 5. This invention, through the application of the dynamic PageRank algorithm, realizes the temporal evaluation of a person's influence, which can accurately reflect the changes in a person's social status in different historical stages and provide quantitative evidence for historical research. Attached Figure Description

[0026] Figure 1 is an overall flowchart of the present invention; Figure 2 is a technical framework diagram of the present invention, including data augmentation (S1) and dynamic network analysis (S2). Figure 3 is an example diagram of the relationship strength, showing that the relationship strength between Shen Zhou and Wen Zhengming (the top solid line) is formed by the superposition of the fundamental wave (the bottom solid line) and the event pulse (the middle dotted line). Detailed Implementation

[0027] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0028] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0029] The present invention will now be described in detail with reference to the accompanying drawings and exemplary embodiments.

[0030] refer to Figures 1 to 3 The present invention provides a method for calculating the dynamic relationship network of painters throughout history for historical learning, the specific steps of which are as follows: S1. Data Acquisition and Augmentation: Acquire basic data and event information from multiple historical painters; classify and organize missing attribute dimensions and event information in the original data; use a large language model for attribute augmentation and data imputation to generate structured data; details are as follows: As attached Figure 2 As shown in the knowledge graph section, data is collected from multiple sources, including personal attribute data (name, birth and death years, place of origin, occupation, etc.), relationship data (basic relationship types, association proofs, etc.), event data (time, description, participants, etc.), and cultural carrier data (basic information and associated records of calligraphy, paintings, letters, etc.). Data sources include Artyx Cloud, the Chinese Historical Biographical Database, the "Chronological Table of Artists from the Song to the Qing Dynasties", and the National Treasure Museum of China. For the data on individuals, we define a set of identity attributes = {literati, officials, merchants, others}. We input the individual's occupation information through a structured prompt template and use the DeepSeek large language model for automatic classification to complete the identity attribute enhancement. For event data, the system uses an event structured prompt template to extract entities (participants, initiators, cultural carriers, locations), classify them hierarchically (macro types: art, politics, social, economic; micro types: creation, travel together, apprenticeship, etc., 8 categories), and assess their impact (weight w∈[-1,1]), generating structured event data and filling the missing attributes in the original data.

[0031] S2, Event Expansion and Verification: A retrieval template is constructed using a large language model to retrieve event information related to the target writer from public documents, forming an expanded event set. The expanded event set is then augmented with attributes to generate structured expanded event data. The similarity between the structured expanded event data and the existing event data is calculated using Levenshtein distance, duplicate events are removed, and high-quality data is retained after sampling verification. Specifically, events with a confidence level below 0.7 are filtered out when the extended event set is generated; structured extended event data is generated by reusing prompt templates and includes participants, initiators, cultural carriers, locations, event types, and impact weight attributes; high-quality data is retained and erroneous data is corrected during sampling and verification to form the extended event set E'; The structured hint template is reused to enhance the attributes of the extended event set E', generating the structured extended event Sevent'; Calculate the Levenshtein distance between the structured extended event Sepent' and the existing event data, remove duplicate events with similarity higher than a set threshold, sample and verify the remaining events, correct erroneous data, and complete the event extension.

[0032] S3. Basic Relationship and Event Impulse Modeling: For each pair of literary figures, the basic relationship type is extracted, and the relationship time interval is determined to be the intersection of the birth and death years of the two individuals. A basic relationship strength model is constructed using a Beta distribution, and the model strength amplitude is optimized by combining the number of interaction events. A relationship event impulse model is constructed using a Gaussian impulse function to quantify the instantaneous impact of a single event on the relationship. The classic PageRank algorithm is extended to a time series dynamic algorithm to determine the initial influence of a person at a certain moment based on personal event impulses. Specifically, the BaseWave model simulates a probability distribution that roughly conforms to the changes in relationships. When there is a basic relationship between individuals labeled by experts, this waveform will serve as a benchmark for relationship strength, providing prior enhancement for the strength evolution of labeled relationships. Its shape is constructed from the beta distribution as shown in formula (1): （1） BaseWave uses the probability density function of the beta distribution as the shape term, normalizing the time and shape parameters. Used to control peak position and skewness; the amplitude term employs a sublinear function to enhance the impact of the number of interactive events on the intensity amplitude; the time range is the intersection of the birth and death years of the individuals. Define the duration of the relationship. The list of interaction events between characters is denoted as , The number of events; Different shape parameters are preset for the beta distribution for different labeling relationships k. If no interaction events occur between graphs, the preset parameters of their relationship categories are used directly; if interactive events exist, the parameters are optimized using the weighted negative log-likelihood over the interactive event list E, where each event... All include timestamps and event weight : （2） .

[0033] The EventPluse relational event pulse is defined using a Gaussian impulse function, and the total event pulse intensity is a linear superposition of the individual events:

[0034] in, To control the pulse's time decay rate, take , For event timestamps, Weighting of event impact, This is a list of interactive events.

[0035] As attached Figure 3 As shown, the events related to the relationship between Shen Zhou and Wen Zhengming are mainly distributed between 1490 and 1510, with a relatively even distribution and a relatively flat basic waveform, with the peak around 1495. The final relationship strength (red line) is formed by superimposing the basic waveform (blue line) and the event pulses (red line).

[0036] S4. Calculation of dynamic relationship strength and propagation strength: Strength of positive relationship Considering only positive relation types And a list of positive relationship events ( ); Strength of negative relationship Considering only negative relation types and a list of negative relationship events ( ):

[0037]

[0038] in, This is the basic relation strength model for the current k-relationship. =0.6, contributing weight to the pulse; It is a positive basic relation type. This is a list of positive events. >0; It is a negative basic relation type. This is a list of negative events. <0; This is the shape parameter of the pre-defined labeling relationship k for the beta distribution; We treat relationship strength as a two-dimensional planar vector, with positive relationship strength on the horizontal axis and negative relationship strength on the vertical axis, ultimately resulting in a total relationship strength. Defined as the magnitude of the composite vector, the angle between the vectors Able to determine relationship tendency, if Indicates a positive tendency in the relationship, if This indicates a negative tendency in the relationship:

[0039]

[0040] in, Strength of negative relationship.

[0041] S5, Dynamic Network Data Output: We define an individual's influence as stemming from two sources: their social network and the independent impact of personal events (such as creative work or holding office). Social network influence is measured through a vector angle. The impact of individual events is quantified, while that of personal events is calculated using EventPluse, a relationship event impulse mechanism. Unlike relationship strength, the influence of personal events is subject to positive and negative cancellation, which better reflects the actual fluctuations in a character's influence.

[0042] Based on this, we extend the classic PageRank algorithm to a dynamic algorithm for time series. Let the set of people be... At any moment The personal event list is Each node The initial influence weight is determined by the individual event pulse: (4) in, For character nodes, for A list of personal events at any given moment.

[0043] The dynamic propagation strength of social relations is determined by the adjacency matrix. Description, its elements express From time to time arrive The strength of the relationship is determined by the angle between the vectors. Calculation, i.e. This is used to define edge weights. The dynamic influence of a character is solved iteratively. (9) Where V represents the set of people, For v The initial influence, k is the relationship identifier, v is the character node, and t is the current time.

[0044] Quantifying the impact of personal events on individuals In time Independent influence, other nodes in the social relationship dissemination item Influence is determined by the strength of the relationship. Passed proportionally , Finally, by integrating information such as overall relationship strength, relationship tendency, dynamic influence, and cultural carrier-related events, complete dynamic network data is output.

[0045] Currently, quantifying the dynamic evolution of relationships among historical figures is a crucial breakthrough in relationship network analysis and a necessary process for historical learning. Research on the relationship networks of historical painters mainly follows two paths: one is static topological analysis based on fixed social relationships (such as blood ties and teacher-student relationships), and the other is inductive analysis based on discrete events recorded in historical documents. The former reflects long-term stable relationships by constructing clan genealogies or school transmission networks, but struggles to capture the impact of short-term cultural interactions; the latter, while recording specific interaction events, lacks mathematical models to quantify changes in relationship strength. Existing methods fail to effectively integrate the social ties of cultural carriers (such as calligraphy, paintings, and letters) and also fail to address the network sparsity problem caused by incomplete historical data. This invention proposes a dynamic calculation method that integrates basic relationship waveforms and event impulse intensity, achieving spatiotemporal continuous modeling of the strength of literati relationships for the first time. Through the reconstruction of historical event data enhanced by a large model, the evolutionary analysis of cultural relationship networks achieves both historical credibility and mathematical rigor.

[0046] The embodiments described in this specification are merely examples of implementations of the inventive concept. The scope of protection of this invention should not be considered as limited to the specific forms stated in the embodiments. The scope of protection of this invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. A method for calculating the dynamic relationship network of painters throughout history for historical learning, characterized in that, Includes the following steps: S1. Data Acquisition and Enhancement: Acquire basic data and event information of multiple historical painters as raw data, classify and organize the missing attribute dimensions and event information in the raw data, use a large language model to enhance attributes and fill data, and generate structured data. S2. Event Expansion and Validation: A retrieval template is constructed using a large language model to retrieve event information related to the target writer from public documents, forming an expanded event set; attribute enhancement processing is performed on the expanded event set to generate structured expanded event data; Levenshtein distance is used to calculate the similarity between the structured expanded event data and the existing event data, duplicate events are removed, and high-quality data is retained after sampling and verification; S3. Basic Relationship and Event Impulse Modeling: Extract the basic relationship type for each pair of literary figures and determine the relationship time interval as the intersection of the birth and death years of the two individuals; A basic relationship strength model is constructed using a Beta distribution, and the model strength amplitude is optimized by combining the number of interaction events. A relationship event impulse model is constructed using a Gaussian impulse function to quantify the instantaneous impact of a single event on the relationship. The classic PageRank algorithm is extended to a time series dynamic algorithm to determine the initial influence of a person at a certain moment based on the individual event impulse. S4. Calculation of Dynamic Relationship Strength and Propagation Strength: Construct positive relationship strength according to two categories of interactive events (positive and negative) and basic relationship types. (t) and the strength of the negative relationship (t); Treat the relationship strength as a two-dimensional vector, calculate the total relationship strength through the vector magnitude, and determine the relationship tendency through the vector angle; Use an adjacency matrix to describe the dynamic propagation strength of social relationships, where the matrix elements are the total relationship strength between the two people at the corresponding time; S5. Dynamic Network Data Output: Combining the strength of dynamic relationships with the intensity of social relationship propagation, and integrating the calculation results of the dynamic influence of individuals, outputs dynamic network data that includes relationship evolution, changes in influence, and connections with cultural carriers.

2. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S1, the basic data comes from multiple sources of historical data, including Artyx Cloud, the Chinese Historical Biographical Database, the "Chronological Table of Artists from the Song to the Qing Dynasties", and the National Treasure Museum of China. The large language model uses DeepSeek. When enhancing the attributes of a person, the set of identity attributes is defined as {literati, officials, merchants, others}. The person's occupation information is input through the structured prompt template to achieve automatic classification. When enhancing the attributes of an event, entity extraction, hierarchical classification and impact assessment are completed through the structured prompt template. The hierarchical classification includes macro types and micro types, and the impact assessment weight w∈[-1,1].

3. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S2, events with a confidence level below 0.7 are filtered out when the extended event set is generated; the structured extended event data is generated by reusing prompt templates and includes participants, initiators, cultural carriers, locations, event types, and influence weight attributes. High-quality data is retained and erroneous data is corrected during sampling verification.

4. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S3, the expression for the basic relationship strength model is: (1) BaseWave uses the probability density function of the beta distribution as the shape term, normalizing the time and shape parameters. Used to control peak position and skewness; the amplitude term employs a sublinear function to enhance the impact of the number of interactive events on the intensity amplitude; the time range is the intersection of the birth and death years of the individuals. Define the duration of the relationship. The list of interaction events between characters is denoted as , The number of events; Different shape parameters are preset for the beta distribution for different labeling relationships k. If no interaction events occur between graphs, the preset parameters of their relationship categories are used directly; if interactive events exist, the parameters are optimized using the weighted negative log-likelihood over the interactive event list E, where each event... All include timestamps and event weight : （2） 。 5. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S3, the expression for the relational event impulse model is: in, To control the pulse's time decay rate, it is generally taken as... , For event timestamps, Weighting the impact of events. For the list of interactive events; the initial influence calculation expression is: （4） in, For character nodes, for A list of personal events at any given moment.

6. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S4, the calculation expressions for the strength of the positive relationship and the strength of the negative relationship are as follows: in, This is the basic relation strength model for the current k relation. =0.6, contributing weight to the pulse; It is a positive basic relation type. This is a list of positive events. >0; It is a negative basic relation type. This is a list of negative events. <0; The shape parameter k is the predefined labeling relation k for the beta distribution. The expressions for calculating the total relation strength and relation tendency are: in, The strength of a negative relationship is as follows: when θ∈(0°,45°), the relationship is positive; when θ∈(45°,90°), the relationship is negative.

7. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S4, the elements of the adjacency matrix A(t) S(t) = S(t), representing the node of the person at time t. to v The strength of relational propagation; the influence of dynamic figures is calculated iteratively, and the expression is: （9） Where V represents the set of people, For v The initial influence, k is the relationship identifier, v is the character node, and t is the current time.

8. The method for calculating the dynamic relationship network of painters throughout history for historical learning as described in claim 1, characterized in that, In step S1, the basic data includes character attribute data, relationship data, and cultural carrier data, and the event information includes event time, participants, and event description.

Images