Method and device for analyzing authorship characteristics of literary works based on multidimensional information

Abstract

The application discloses a literary work author creation characteristic analysis method and device based on multi-dimensional information, comprising: obtaining multi-dimensional work data of an author; performing space-time correlation modeling on the multi-dimensional work data to obtain a multi-dimensional space-time data model; based on the multi-dimensional space-time data model, layering work nodes according to time slices, and establishing directed track edges according to time sequence and geographical position changes to obtain an initial space-time track network; extracting node evaluation features and edge evaluation features from the multi-dimensional space-time data model and the initial space-time track network; determining node creation strength according to the node evaluation features and determining edge weights according to the edge evaluation features; configuring the node creation strength and the edge weights to the initial space-time track network to obtain a layered space-time track network; and determining a creation characteristic analysis result according to the layered space-time track network and theme attributes and social attributes and outputting the result to a geographic information terminal. The application is beneficial to improving the accuracy and traceability of creation track reconstruction and dynamic display.

Description

Technical Field

[0001] This application relates to the fields of literary data processing and digital humanities technology, specifically to a method and apparatus for analyzing the creative characteristics of authors of literary works based on multidimensional information. Background Technology

[0002] Analyzing the creative characteristics of literary authors typically requires integrating information such as the text of the work, the time of creation, the location of activities, the themes, and the social interactions of the authors. This information is scattered across collections of works, chronologies, biographies, letters, and geographical data, and includes unstructured text, structured metadata, spatial coordinates, and data on relationships between authors. It is characterized by heterogeneous sources, inconsistent record granularity, and long time spans.

[0003] Existing technologies typically perform separate tasks for geographic entity labeling, work quantity statistics, thematic analysis, or character relationship display. There is a lack of stable work-level correspondences between different analysis results. When switching time ranges, map scales, or expanding character relationships, issues easily arise where work, location, theme, and character information cannot be synchronized. Furthermore, it is difficult to continuously present changes in the author's creative activities over time and space.

[0004] Therefore, how to uniformly process heterogeneous literary materials, maintain traceable connections between the content of the works and their time, space, theme and character relationships, and form data results suitable for continuous analysis and interactive display is a technical problem that needs to be solved in this field. Summary of the Invention

[0005] This application proposes a method and apparatus for analyzing the creative characteristics of literary authors based on multidimensional information, aiming to improve the consistency of heterogeneous literary data and generate analytical results that can characterize the spatiotemporal evolution of the author's creative activities. This helps reduce ambiguity in the association of heterogeneous data and improves the accuracy and traceability of reconstructing and dynamically displaying creative trajectories.

[0006] Firstly, embodiments of this application provide a method for analyzing the creative characteristics of authors of literary works based on multi-dimensional information, executed by a creative characteristic analysis server. The method includes: Obtain multidimensional work data from the author; the multidimensional work data includes the content of the work as well as spatiotemporal, thematic, and social attributes. Spatiotemporal correlation modeling is performed on the multidimensional artwork data to obtain a multidimensional spatiotemporal data model; Based on the aforementioned multidimensional spatiotemporal data model, literary works are used as work nodes, and the work nodes are layered according to time slices. Directed trajectory edges are established according to temporal and geographical location changes to obtain an initial spatiotemporal trajectory network. Node evaluation features and edge evaluation features are extracted from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network. The node evaluation features include the amount of creations at a location, the local work density benchmark value, and the time slice weight. The edge evaluation features include the amount of creations at a location, the density of works outside the region, and the style changes of adjacent nodes. The node creation intensity is determined based on the node evaluation characteristics, and the edge weight is determined based on the edge evaluation characteristics. The node creation intensity and the edge weight are configured into the initial spatiotemporal trajectory network to obtain a hierarchical spatiotemporal trajectory network. Based on the hierarchical spatiotemporal trajectory network and the thematic and social attributes, the creative characteristic analysis results are determined and output to the geographic information terminal.

[0007] As can be seen, this embodiment associates the content, spatiotemporal attributes, thematic attributes, and social attributes of a work with work nodes, and maintains the temporal and geographical migration relationships between works through directed trajectory edges. The node creation intensity and edge weight can respectively characterize the level of creative activity at a location and the degree of change in adjacent creative activities, enabling multiple types of literary data to participate in the analysis collaboratively in the same trajectory network, thereby improving the accuracy and traceability of creative trajectory reconstruction and dynamic display.

[0008] In one possible example, extracting node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network includes: extracting node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network includes: reading the work content, geographical location, time slice, and theme attributes of the work nodes from the multidimensional spatiotemporal data model, and reading directed trajectory edges from the initial spatiotemporal trajectory network to obtain trajectory evaluation basic data containing the relationships between adjacent work nodes; grouping the work nodes in the trajectory evaluation basic data according to geographical location and time slice, calculating the location creation volume of each group, and reading the local work density benchmark value associated with the geographical location and the time slice weight associated with the time slice to obtain the node evaluation features; determining the density of works outside the region based on the number of work nodes with the "Travel Outside the Region" theme identifier in the trajectory evaluation basic data and the total number of work nodes in the corresponding group; determining the style changes of adjacent nodes based on the work content and theme attributes of adjacent work nodes in the trajectory evaluation basic data; combining the location creation volume, the density of works outside the region, and the style changes of adjacent nodes into the edge evaluation features to obtain the node evaluation features and the edge evaluation features.

[0009] As can be seen, this example uses the same trajectory evaluation base data as input to form node-side features and edge-side features respectively, which can avoid the inconsistency in data caliber caused by separate statistics on the number of works, theme identifiers and adjacent relationships.

[0010] In one possible example, determining the style change of adjacent nodes based on the content and theme attributes of adjacent work nodes in the trajectory evaluation basic data includes: reading the content and theme attributes of adjacent work nodes from the trajectory evaluation basic data to obtain adjacent node analysis data; performing semantic feature quantization on the adjacent node analysis data to obtain the node style features of each adjacent work node; comparing the node style features of each group of adjacent work nodes in the same dimension according to the relationship between the adjacent work nodes to obtain the dimensional difference value of each group of adjacent work nodes; and aggregating the dimensional difference values ​​to obtain the style change of adjacent nodes.

[0011] As can be seen, this example transforms the overall stylistic differences between adjacent works into multiple comparable feature dimensions, and aggregates the differences while maintaining the correspondence between the dimensions, which can reduce the randomness caused by relying solely on a single topic tag to judge style changes.

[0012] In one possible example, the semantic feature quantization of the adjacent node analysis data to obtain the node style features of each adjacent work node includes: performing text normalization and term segmentation on the work content in the adjacent node analysis data to obtain a standardized text sequence for each work node; matching the standardized text sequence with a preset thesaurus and a travel-related extraterritorial thesaurus to obtain term feature groups for each work node; determining word frequency features and term co-occurrence features based on the term feature groups, and encoding the theme attributes in the adjacent node analysis data to obtain node feature vectors for each work node; and normalizing each feature dimension of the node feature vectors to obtain the node style features of each adjacent work node.

[0013] As can be seen, this example transforms the term information, term combination relationships, and existing thematic attributes in the content of the work into a unified feature space, so that works of different lengths and creation periods can be compared on the same dimension.

[0014] In one possible example, matching the standardized text sequence with a preset thesaurus and a travel-related extraterritorial thesaurus to obtain the term feature groups of each work node includes: performing continuous term and phrase recognition on the standardized text sequence to obtain a candidate term set; matching the candidate term set with terms and synonyms in the preset thesaurus and the travel-related extraterritorial thesaurus respectively to obtain hit terms and hit positions; configuring classification labels for the hit terms according to the term affiliation and hierarchical relationship of the hit terms in the preset thesaurus or the travel-related extraterritorial thesaurus to obtain labeled terms; and aggregating the labeled terms and hit positions according to work nodes to obtain the term feature groups of each work node.

[0015] As can be seen, this example retains the hit terms, text locations, and category tags, enabling subsequent processing to distinguish the usage of the same terms in different works and contexts, and improving the consistency of subject terms and regional terms extraction.

[0016] In one possible example, configuring classification tags for the hit words based on their affiliation and hierarchical relationship in the preset thesaurus or the travel-related extraterritorial thesaurus to obtain labeled words includes: reading the affiliation and hierarchical relationship of the words corresponding to the hit words from the preset thesaurus and the travel-related extraterritorial thesaurus to obtain tag mapping data; determining the theme tag of the hit words based on the theme affiliation in the tag mapping data; determining the regional subcategory tag of the hit words based on the travel-related extraterritorial hierarchy in the tag mapping data; and associating at least one of the theme tag and the regional subcategory tag with the corresponding hit words to obtain the labeled words.

[0017] As can be seen, this example subdivides the broad theme of travel and expatriation through the thesaurus hierarchy, enabling work nodes to retain both the overarching theme and regional subcategories, facilitating subsequent cross-regional statistics, theme migration analysis, and layered map display.

[0018] In one possible example, determining the node creation intensity based on the node evaluation features and determining the edge weight based on the edge evaluation features includes: reading the location creation volume, local work density benchmark value, and time slice weight corresponding to each work node from the node evaluation features to obtain node intensity calculation data; normalizing the location creation volume in the node intensity calculation data according to the local work density benchmark value to obtain normalized creation volume; adjusting the normalized creation volume according to the time slice weight to obtain the node creation intensity; reading the location creation volume, external work density, and adjacent node style changes corresponding to each directed trajectory edge from the edge evaluation features, and normalizing the dimensions of each to obtain edge weight components; weighting and aggregating the edge weight components according to a preset contribution coefficient to obtain the edge weight, and outputting the node creation intensity and the edge weight.

[0019] As can be seen, this example first eliminates the influence of the number of works at different locations and different evaluation feature dimensions, and then determines the creative intensity of nodes and edge weights, which is conducive to maintaining the consistency of evaluation scales across different locations and time slices.

[0020] In one possible example, the spatiotemporal correlation modeling of the multidimensional work data to obtain a multidimensional spatiotemporal data model includes: standardizing the creation time, location name, and geographical coordinates in the multidimensional work data, and transforming the geographical coordinates to a unified geographical coordinate system to obtain standardized spatiotemporal metadata; associating the standardized spatiotemporal metadata with the corresponding work content, work version, collection affiliation, spatial imagery, theme attribute, and social attributes to obtain structured spatiotemporal work records; writing the structured spatiotemporal work records into a spatial database that supports geographic range queries to obtain spatiotemporal work data; and establishing work nodes, work sequence relationships, and travel route relationships in a graph database that supports extended relational queries based on the work identifier, collection affiliation, theme attribute, and social attributes in the structured spatiotemporal work records. The structured spatiotemporal work records are divided according to the time boundaries corresponding to the author's life stages to obtain time slices, and the corresponding time slice tags are written into the spatiotemporal work data and the relationship data. The same work identifier and time slice tag are written into the corresponding spatiotemporal work records and work nodes to form cross-database mapping items. Based on the cross-database mapping items, a graph query entry from the spatiotemporal work data to the relationship data and a spatial back lookup entry from the relationship data to the spatiotemporal work data are established, and the query order of first performing time range and spatial range filtering and then performing work relationship expansion is configured to obtain the dual-database interoperable query relationship. The spatiotemporal work data, the relationship data, the time slices and the dual-database interoperable query relationship are associated to obtain the multidimensional spatiotemporal data model.

[0021] As can be seen, this example uses the same work identifier and time slice label to maintain the correspondence between spatial data and relational data, enabling the creative characteristic analysis server to first perform spatiotemporal range filtering, then expand the work relationships around the filtering results, and be able to retrieve the spatial information of the work from the relation query results.

[0022] In one possible example, determining the creative characteristic analysis results and outputting them to a geographic information terminal based on the hierarchical spatiotemporal trajectory network and the theme attributes and social attributes includes: receiving time interval, map scale, and region selection information sent by the geographic information terminal; extracting a target trajectory subnet from the hierarchical spatiotemporal trajectory network based on the time interval, map scale, and region selection information; obtaining the theme attributes and social attributes associated with the target trajectory subnet to obtain target analysis data; generating a spatial distribution heatmap layer based on the geographical location and creative intensity of the work nodes in the target analysis data at a global scale; clustering the work nodes in the target analysis data according to the theme attributes at a regional scale to generate a regional theme clustering layer and a theme word cloud layer; and configuring the location marker size according to the creative intensity of the nodes at a location scale. The content of the works is associated with the corresponding work nodes to generate a location node layer; a curve trajectory layer is generated based on the work nodes with the "Travel Abroad" theme identifier and their corresponding directed trajectory edges, and the trajectory line width is configured according to the amount of creation at each location; the intensity of the social interaction is determined based on the geographical location of the social interaction objects, the frequency of responses, the number of collaborations, and the output of results in the social interaction attributes, and the radiation range and connection line weights are configured according to the social interaction intensity to generate a social interaction network radiation circle layer; the spatial distribution heat map layer, the regional theme clustering layer, the theme word cloud layer, the location node layer, the curve trajectory layer, and the social interaction network radiation circle layer are linked with the time axis and rendered in an electronic map that supports two-dimensional and three-dimensional switching to obtain the creative characteristic analysis results, including the spatiotemporal distribution of creation, theme migration path, and spatial diffusion trend of social interaction network, and output to the geographic information terminal.

[0023] As can be seen, this example configures different data representation methods according to map scale, and uses a time axis to synchronize changes in spatial distribution, theme migration, and cross-diffusion results, which can reduce display interference caused by loading irrelevant data at once.

[0024] Secondly, embodiments of this application provide a device for analyzing the creative characteristics of literary works authors based on multidimensional information, including: The data acquisition module is used to acquire multidimensional work data of the author; the multidimensional work data includes the content of the work as well as spatiotemporal, thematic, and social attributes. The spatiotemporal correlation modeling module is used to perform spatiotemporal correlation modeling on the multidimensional work data acquired by the data acquisition module to obtain a multidimensional spatiotemporal data model. The initial network construction module is used to obtain the initial spatiotemporal trajectory network based on the multidimensional spatiotemporal data model obtained by the spatiotemporal correlation modeling module, with literary works as work nodes, the work nodes are layered according to time slices, and directed trajectory edges are established according to temporal and geographical location changes. The evaluation feature extraction module is used to extract node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network. The node evaluation features include the amount of creations at a location, the local work density benchmark value, and the time slice weight. The edge evaluation features include the amount of creations at a location, the density of works outside the region, and the style changes of adjacent nodes. The evaluation parameter determination module is used to determine the node creation intensity based on the node evaluation features extracted by the evaluation feature extraction module, and to determine the edge weight based on the edge evaluation features extracted by the evaluation feature extraction module. The trajectory network configuration module is used to configure the node creation intensity and edge weight determined by the evaluation parameter determination module to the initial spatiotemporal trajectory network constructed by the initial network construction module, so as to obtain a hierarchical spatiotemporal trajectory network. The analysis output module is used to determine the creative characteristic analysis results and output them to the geographic information terminal based on the hierarchical spatiotemporal trajectory network obtained by the trajectory network configuration module and the theme attributes and social attributes obtained by the data acquisition module.

[0025] As can be seen, this embodiment implements the aforementioned method through interconnected functional modules, so that the input and output of each module are transmitted in the order of multidimensional work data, multidimensional spatiotemporal data model, initial spatiotemporal trajectory network, evaluation features, evaluation parameters, hierarchical spatiotemporal trajectory network and creative characteristic analysis results, which facilitates deployment and execution in servers or distributed devices.

[0026] Thirdly, embodiments of this application provide an electronic device including a processor, a memory, a communication interface, and one or more programs, the one or more programs being stored in the memory and configured to be executed by the processor, the programs including instructions for performing the steps in the first aspect of embodiments of this application.

[0027] Fourthly, embodiments of this application provide a computer storage medium storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps described in the first aspect of this embodiment.

[0028] As can be seen, in this embodiment of the application, compared with the existing analysis methods that only perform geographic entity annotation, work quantity statistics, or static map display, this application performs work-level spatiotemporal correlation and trajectory processing on heterogeneous work data. The creative characteristic analysis server of this application first acquires the author's multidimensional work data, then performs spatiotemporal correlation modeling on the multidimensional work data, then establishes an initial spatiotemporal trajectory network, and then extracts node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network, subsequently determining the node creation intensity and edge weights, then configuring the node creation intensity and edge weights to the initial spatiotemporal trajectory network, and finally combining the hierarchical spatiotemporal trajectory network, theme attributes, and interaction attributes to determine and output the creative characteristic analysis results. Because work nodes, directed trajectory edges, evaluation features, node creation intensity, and edge weights are linked according to a continuous data processing chain, work-level spatiotemporal migration, location creation activity, and changes in the style of adjacent works can be collaboratively represented in the same network, thereby reducing the ambiguity of heterogeneous data associations and improving the accuracy and traceability of creative trajectory reconstruction and dynamic display. Attached Figure Description

[0029] Figure 1 A schematic diagram of the hardware architecture of the literary work author creative characteristic analysis system provided in this application embodiment; Figure 2 A schematic diagram of the creative characteristic analysis software system provided in the embodiments of this application; Figure 3 A schematic diagram illustrating an author's cross-regional creative activity analysis scenario provided in this application embodiment; Figure 4 A flowchart illustrating the method for analyzing the creative characteristics of authors of literary works provided in this application embodiment; Figure 5 Example diagram of hierarchical spatiotemporal trajectory network construction and correlation analysis provided in the embodiments of this application; Figure 6 A flowchart illustrating the node evaluation feature and edge evaluation feature extraction process provided in the embodiments of this application; Figure 7 Example diagram of trajectory evaluation basic data grouping and evaluation feature formation provided in the embodiments of this application; Figure 8 A flowchart illustrating the process of determining the style change of adjacent nodes provided in an embodiment of this application; Figure 9 Example diagram of style feature comparison between adjacent work nodes provided in the embodiments of this application; Figure 10 A flowchart illustrating the node style feature determination process provided in this application embodiment; Figure 11 Example diagram of semantic feature quantification of a work provided in this application embodiment; Figure 12 Example diagram of multidimensional spatiotemporal data model and dual-database interoperability query provided in the embodiments of this application; Figure 13 A schematic diagram of the functional modules of the literary work author creative characteristic analysis device based on multidimensional information provided in the embodiments of this application; Figure 14 A schematic geographical distribution example of key spatiotemporal nodes in author B's poetry creation provided for embodiments of this application; Figure 15 This is a visualization example of the trajectory of author B's poems on the theme of traveling abroad, provided as an embodiment of this application. Detailed Implementation

[0030] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only used to explain this application and are not intended to limit the scope of protection of this application.

[0031] The term "work" in this application can refer to poetry, prose, novels, plays, letters, or other literary texts with author attribution and creative content. A work node is a network node established using a single work or a fragment of a work whose creation time can be independently determined as the data unit. A work identifier is a data identifier used to identify the same work across different databases, data tables, and query results.

[0032] The spatiotemporal attributes referred to in this application may include creation time, time precision, location name, administrative division, geographical coordinates, location source, and location confidence level. Thematic attributes may include preset theme tags, regional subcategory tags, spatial imagery, and thematic features extracted from the content of the work. Social attributes may include social contacts, relationship type, relationship duration, social interaction locations, collaborative works, cooperative matters, and data sources.

[0033] The term "time slice" in this application refers to the time grouping of works recorded according to the author's life stage, research time range, or a predetermined time granularity. In the hierarchical spatiotemporal trajectory network, "hierarchy" indicates that each work node belongs to a corresponding time slice, but does not require all time slices to have the same length.

[0034] In one possible example, see Figure 1The literary work author creative characteristic analysis system 100 provided in this application embodiment includes a creative characteristic analysis server 110, a literary data source device group 120, a spatial database server 130, a graph database server 140, a geographic information terminal 150, and a communication network 160. The creative characteristic analysis server 110 includes a processor 111, a memory 112, a communication interface 113, and an analysis data storage 114; the literary data source device group 120 includes a work text data source 121, a chronological data data source 122, a geographic data data source 123, and a social data data source 124.

[0035] Processor 111 executes the creative characteristic analysis program stored in memory 112. Memory 112 stores program instructions, a dictionary, parameter configurations, and intermediate processing results. Communication interface 113 receives literary data via communication network 160 and sends analysis results to geographic information terminal 150. Analysis data storage 114 stores work identifiers, data sources, processing status, and analysis task records.

[0036] Textual data source 121 can provide the original text, title, edition, collection affiliation, and length of the work. Chronological data source 122 can provide the author's activity period, creation time, location records, and life stages. Geographical data source 123 can provide standard location names, administrative divisions, geographical coordinates, and the historical evolution of place names. Social data source 124 can provide information on relationships, records of poems exchanged, collaborative works, correspondence, and corresponding times and locations.

[0037] Spatial database server 130 is used to store spatiotemporal artwork data that supports filtering by time and spatial range. Graph database server 140 is used to store artwork nodes, character nodes, and relationships between artwork sequences, travel routes, and call-and-response relationships. Spatial database server 130 and graph database server 140 can be deployed on different devices or serve as two data service instances within creative characteristic analysis server 110.

[0038] The geographic information terminal 150 can be a computer, a mobile terminal, or a research and display terminal, used to submit author, time interval, map scale, and region selection information, and to display the spatiotemporal distribution of creation, theme migration path, and spatial diffusion trend of social networks.

[0039] The aforementioned hardware architecture enables literature data collection, spatial querying, relation expansion, trajectory calculation, and map display to be performed collaboratively by the corresponding devices, and allows database services and analysis services to be deployed separately according to data volume and access requirements.

[0040] See Figure 2The creative characteristic analysis software system 200 provided in this application embodiment includes a data access layer 210, a data specification layer 220, a model storage layer 230, a trajectory processing layer 240, an analysis service layer 250, and a visual interaction layer 260. The data access layer 210 includes a text interface 211, a metadata interface 212, a geographic interface 213, and a relational interface 214; the data specification layer 220 includes a text preprocessing unit 221, a time specification unit 222, a geographic specification unit 223, and an identifier association unit 224; the model storage layer 230 includes a spatial data management unit 231, a relational data management unit 232, a time slice management unit 233, and a cross-database query unit 234; the trajectory processing layer 240 includes an initial network construction unit 241, an evaluation feature extraction unit 242, an evaluation parameter determination unit 243, and a hierarchical network configuration unit 244; the analysis service layer 250 includes a spatiotemporal distribution analysis unit 251, a topic migration analysis unit 252, and a social diffusion analysis unit 253; and the visual interaction layer 260 includes a timeline unit 261, a scale switching unit 262, a node trajectory rendering unit 263, and a work content calling unit 264.

[0041] The data access layer 210 is used to receive work text, work metadata, geographical data, and social data. The data specification layer 220 is used to process the text format, creation time, location name, geographical coordinates, and work identifier, so that data from different sources have a consistent data format and associated primary key.

[0042] The model storage layer 230 is used to form spatiotemporal work data, relational data, time slices, and inter-database query relationships. The trajectory processing layer 240 is used to construct work nodes and directed trajectory edges, extract evaluation features, determine the creation intensity of nodes and edge weights, and form a hierarchical spatiotemporal trajectory network.

[0043] The analysis service layer 250 is used to determine the spatiotemporal distribution of creation, thematic migration paths, and spatial diffusion trends of social networks from the hierarchical spatiotemporal trajectory network. The visual interaction layer 260 is used to receive the user's time, space, and scale selections and to display the corresponding analysis layers in a linked manner.

[0044] Each software layer can be executed by a single process, or it can be deployed separately for data access, model storage, trajectory calculation, and display services. By passing work identifiers, time slice tags, and processing results through inter-layer interfaces, the data correspondence between different software units can be maintained.

[0045] See Figure 3In the scenario of analyzing the author's cross-regional creative activities provided in this application embodiment, taking author A as an example, the portfolio data 310 records the content of works W01 to W06 and the portfolio affiliation; the chronology data 320 records the creation year and the author's activity location corresponding to works W01 to W06; the geographical data 330 provides the standard name, region and geographical coordinates of each location; and the social record 340 records the correspondence, meetings or cooperation between author A and figures P01, P02 and P03.

[0046] The creative characteristic analysis server 110 uses the work identifiers of works W01 to W06 as the primary key to combine the content, time, location, theme, and social information of each work into multi-dimensional work data. The server can assign works W01 and W02 to the first time slice, works W03 and W04 to the second time slice, and works W05 and W06 to the third time slice, and establish directed trajectory edges according to the creation sequence and location changes.

[0047] When works W03, W04, W05, and W06 bear the "Traveling Abroad" theme identifier, the server can determine the density of works from outside the region based on their location, and determine the style changes of adjacent nodes by combining the themes, terms, and spatial imagery changes of adjacent works. After the node creation intensity and edge weights are configured to the corresponding nodes and trajectory edges, a creative activity trajectory across time slices can be formed.

[0048] The Geographic Information Terminal 150 displays author A's cross-regional migration path on a global scale, shows the clustering of different themes on a regional scale, and displays the original text of the work and the people he interacted with on a location scale. When the user drags the timeline, the work nodes, trajectory edges, theme layers, and interaction layers are updated synchronously according to the corresponding time slices.

[0049] This scenario is not limited to specific biographical or work data of a particular author. In practice, the number of works, time slices, and geographical categories can be set according to the research object, the completeness of the data, and the analysis task.

[0050] In one possible example, see Figure 4 The method for analyzing the creative characteristics of literary works provided in this application embodiment can be implemented through the following steps S101 to S107, which can be executed by the creative characteristic analysis server 110.

[0051] S101. Obtain multidimensional work data of the author; the multidimensional work data includes the content of the work as well as spatiotemporal, thematic, and social attributes.

[0052] The multidimensional work data uses the work identifier as the basic association primary key. Work content can include title, body text, paragraph structure, text version, and collection affiliation. Spatiotemporal attributes, thematic attributes, and social attributes can be directly derived from structured data or extracted by the server from text, chronology, and social data.

[0053] For works whose creation time can only be determined down to the year or time range, the time value and time precision can be saved. For locations where there are differences between ancient and modern place names, both the original location name and the standard location name can be saved. For cases where multiple versions of the same work exist, version identifiers and source identifiers can be retained to prevent different versions from being mistakenly identified as different works.

[0054] S102. Perform spatiotemporal correlation modeling on the multidimensional work data to obtain a multidimensional spatiotemporal data model.

[0055] The multidimensional spatiotemporal data model is used to maintain the correlation between work content, spatial records, relationship records, and time slices. The model can include spatiotemporal work data, relationship data, time slices, and inter-database query relationships, and is not limited to a single data table or a single database.

[0056] The server can combine the content, time, location, theme, and social interaction data of the same work into a structured record based on the work's identifier, and save the data source for each field. The data source field can be used to trace the original data in case of location conflicts, time conflicts, or version conflicts.

[0057] S103. Based on the multidimensional spatiotemporal data model, literary works are used as work nodes. The work nodes are layered according to time slices, and directed trajectory edges are established according to temporal and geographical location changes to obtain the initial spatiotemporal trajectory network.

[0058] Each work node is associated with at least the work identifier, creation time, geographical location, and time slice. The server can first sort the work nodes according to their creation time, and then establish directed trajectory edges between adjacent work nodes, pointing from earlier works to later works.

[0059] When two works have the same creation time granularity, their chronological relationship can be determined based on the order of recorded information, the arrangement order of the works in the collection, or more fine-grained time information. When the chronological relationship cannot be reliably determined, the two works can be configured as nodes at the same level, and directed trajectory edges can be established between each work and the node of the subsequent work.

[0060] Changes in geographical location can include changes in location coordinates, administrative regions, or research areas. For works created at adjacent times and in the same location, directed trajectory edges can still be established to represent consecutive creative activities occurring within the same location.

[0061] S104. Extract node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network; the node evaluation features include the amount of creation at a location, the local work density benchmark value and the time slice weight, and the edge evaluation features include the amount of creation at a location, the density of works outside the region and the style changes of adjacent nodes.

[0062] The location-based creation volume represents the number of work nodes corresponding to the target geographical location and the target time slice. The local work density benchmark value represents the reference data used to evaluate the degree of work clustering at that location, which can come from other time slices of the author, data of similar authors, or a pre-established research sample. The time slice weight is used to characterize the relative role of different life stages, research stages, or exhibition stages in the current analysis task.

[0063] The density of works from outside the target area indicates the proportion of work nodes in the target group that bear the theme of "travel outside the target area". The style variation of adjacent nodes indicates the degree of variation in features such as terms, term co-occurrence, theme and spatial imagery between two work nodes connected by a directed trajectory edge.

[0064] S105. Determine the node creation intensity based on the node evaluation characteristics, and determine the edge weight based on the edge evaluation characteristics.

[0065] The server can first normalize the creation volume of a location according to the local work density benchmark value, and then adjust the normalization result according to the time slice weight to obtain the node creation intensity. The node creation intensity can be configured to each work node under that location and time slice, or it can be configured to a location aggregation node formed by multiple work nodes.

[0066] Edge weights can be obtained by performing dimensional normalization on the amount of works created at a location, the density of works from outside the region, and the stylistic changes of adjacent nodes, and then aggregating them according to preset contribution coefficients. Each contribution coefficient can be configured according to the research task and saved in association with the parameter version.

[0067] S106. Configure the node creation intensity and the edge weight to the initial spatiotemporal trajectory network to obtain a hierarchical spatiotemporal trajectory network.

[0068] Configuring node creation intensity can include writing node creation intensity into the work node attributes and establishing a mapping between node creation intensity and display size, display level, or query sorting. Configuring edge weight can include writing edge weight into the directed trajectory edge attributes and establishing a mapping between edge weight and trajectory line width, trajectory saliency, or path filtering order.

[0069] As an enhanced example of steps S103 to S106, see Figure 5In the initial spatiotemporal trajectory network 510 provided in this application embodiment, the work nodes are arranged in layers according to the first time slice 520, the second time slice 530 and the third time slice 540; the node creation intensity label 550 is configured to the corresponding work node; the edge weight label 560 is configured to the corresponding directed trajectory edge; the theme association layer 570 and the social association layer 580 are connected to the corresponding work node through the work identifier; the configured data is provided to the creation characteristic result view 590.

[0070] For example, if the number of work nodes at a certain location within the second time slice 530 is higher than the corresponding local work density benchmark value, and this time slice has a high analysis weight, the work nodes at that location can have a relatively high node creation intensity. When there is both a high density of works from outside the region and obvious style changes between adjacent nodes between the second time slice 530 and the third time slice 540, the corresponding directed trajectory edge can have a relatively high edge weight.

[0071] S107. Based on the hierarchical spatiotemporal trajectory network and the theme attributes and social attributes, determine the creative characteristic analysis results and output them to the geographic information terminal.

[0072] The server can first obtain the target trajectory subnet in the hierarchical spatiotemporal trajectory network based on the filtering information submitted by the terminal, and then associate the theme attribute and social attribute based on the target trajectory subnet. The results of the creative characteristic analysis can include the distribution of creative locations, changes in creative intensity, theme clustering areas, theme migration direction, distribution of social objects, and the diffusion process of social relationships.

[0073] The server can output analysis results data, map layer data, or pre-generated visualization pages to the geographic information terminal. The output data can carry work identifiers, enabling the geographic information terminal to continue to retrieve the original text and data sources when the user selects a work node.

[0074] This embodiment forms a continuous processing chain by using multidimensional artwork data, a multidimensional spatiotemporal data model, an initial spatiotemporal trajectory network, evaluation features, evaluation parameters, and a hierarchical spatiotemporal trajectory network, ensuring that each intermediate result is used in subsequent steps. The creation time sequence, location changes, creative activity levels, and stylistic changes of adjacent artworks can remain correlated within the same network, thereby improving the consistency and traceability of the creative characteristic analysis results.

[0075] In one possible example, see Figure 6 The extraction of node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network in the aforementioned step S104 can be achieved through the following steps S201 to S205.

[0076] S201. Read the work content, geographical location, time slice and theme attribute of the work node from the multi-dimensional spatiotemporal data model, and read the directed trajectory edge from the initial spatiotemporal trajectory network to obtain the trajectory evaluation basic data containing the relationship between adjacent work nodes.

[0077] The trajectory evaluation data can be organized according to the work identifier. Each piece of basic data includes at least the current work node, the location of the current node, the time slice, the theme attribute, and the directed trajectory edge connected to the current node. The relationship between adjacent work nodes can include the predecessor node identifier, the successor node identifier, and the directed trajectory edge identifier.

[0078] The server can perform integrity checks on the read data. Nodes that lack content but retain thematic attributes can participate in the location creation statistics, but may not participate in style change calculations; nodes that lack reliable geographical locations can be marked as nodes to be verified and will not be added to the location group for the time being.

[0079] S202. Group the work nodes in the trajectory evaluation basic data according to geographical location and time slice, count the number of creations in each group, and read the local work density benchmark value associated with geographical location and the time slice weight associated with time slice to obtain the node evaluation features.

[0080] The geographical location can be represented by location coordinates, administrative division codes, or research area codes. The server can group work nodes that are at the same geographical location and belong to the same time slice into the same group, and determine the number of work nodes within the group as the location creation quantity.

[0081] Local work density benchmarks can be configured individually by location, or generated from statistical data of works from adjacent locations or similar regions. Time slice weights can be preset according to the current research task, or configured based on the data completeness of the time slice and the analysis stage selected by the user.

[0082] S203. Determine the density of works outside the region based on the number of work nodes with the "Travel Outside the Region" theme identifier and the total number of work nodes in the corresponding group in the trajectory evaluation basic data.

[0083] The "Travel Abroad" theme identifier can be generated from theme thesaurus matching results, manual annotation results, or a combination of both. The server counts the number of work nodes with the identifier within the same geographical location and time slice group, and determines the density of overseas works based on its proportion in the total number of work nodes in the group.

[0084] When a work node has multiple regional subcategory tags, the work node is still counted as a work node with the "Traveling Abroad" theme in the current group to avoid duplicate counting.

[0085] S204. Based on the content and theme attributes of adjacent work nodes in the trajectory evaluation basic data, determine the style changes of adjacent nodes.

[0086] The server can obtain the preceding and subsequent work nodes one by one based on the directed trajectory edges, and then extract the terms, term co-occurrence, theme category, and spatial imagery features of the two nodes respectively. The server forms a dimension difference value by comparing the differences in the values ​​of the same feature dimension, and aggregates the dimension difference values ​​to obtain the style changes of adjacent nodes.

[0087] In this extended architecture example, to address the issue of literal differences caused by different versions of the same work or quoted text being misjudged as style changes, the server can first identify duplicate segments, variant character replacement segments, and editorial punctuation differences according to the work identifier and version relationship. Then, the feature dimensions identified as version differences are configured with lower comparison weights, while newly emerging thematic terms, spatial imagery, and stable co-occurrence relationships are retained as the main comparison content. This can reduce the interference of version editing factors on style migration judgment.

[0088] S205. Combine the number of creations at the location, the density of works from outside the region, and the style changes of adjacent nodes into the edge evaluation features to obtain the node evaluation features and the edge evaluation features.

[0089] Among them, node evaluation features can be saved according to work node identifiers or location group identifiers, and edge evaluation features can be saved according to directed trajectory edge identifiers. When the location creation quantity is used by both the node side and the edge side, the reference identifier of the same statistical result can be saved to avoid numerical discrepancies caused by separate statistics.

[0090] As an enhanced example of steps S201 to S205, see Figure 7 In this embodiment of the application, the target time slice data 710 is grouped into locations according to geographical location 720; work nodes with the theme of "traveling outside the territory" are grouped into theme identifier group 730; nodes corresponding to directed trajectory edges are organized into adjacent node pairs 740; location grouping 720 is combined with local benchmark data to form node evaluation features 750; the number of creations at the location, the statistical results of theme identifier group 730 and the style changes of adjacent node pairs 740 form edge evaluation features 760.

[0091] For example, the target time slice data 710 contains five work nodes for location A, two of which have the theme "Travel Abroad". The server uses these five work nodes to determine the amount of work created at that location, determines the density of works created abroad by the proportion of the two labeled nodes relative to the total number of work nodes in that group, and obtains the style changes of adjacent nodes according to the directed trajectory edges. The above results together form the evaluation features corresponding to the nodes and the directed trajectory edges.

[0092] This embodiment generates node evaluation features and edge evaluation features from the same trajectory evaluation base data, and maintains the correspondence of evaluation results through work node identifiers, location group identifiers, and directed trajectory edge identifiers. This can reduce the statistical range misalignment caused by extracting features separately, and provide consistent input for calculating node creation intensity and edge weights.

[0093] In one possible example, see Figure 8 The determination of style changes of adjacent nodes based on the content and theme attributes of adjacent work nodes in the aforementioned step S204 can be achieved through the following steps S301 to S304.

[0094] S301. Read the work content and theme attributes of adjacent work nodes from the trajectory evaluation basic data to obtain adjacent node analysis data.

[0095] The server reads the node identifiers of the preceding and succeeding works based on the directed trajectory edge identifiers, and then retrieves the work content and theme attributes corresponding to the two nodes from the multidimensional spatiotemporal data model. Adjacent node analysis data can be saved as node pairs, carrying the directed trajectory edge identifiers.

[0096] Thematic attributes can include manually labeled topics, thesaurus-matched topics, regional subcategories, and spatial imagery. The content of the work can include the original text before normalization and the text version used for analysis, allowing for retrospective review of the analysis results.

[0097] S302. Semantic feature quantification is performed on the analysis data of the adjacent nodes to obtain the node style features of each adjacent work node.

[0098] Semantic feature quantization is used to convert the text and thematic attributes of a work into feature representations of the same dimension. Node style features may include thematic term distribution, regional term distribution, term co-occurrence distribution, spatial imagery distribution, sentence length features, and thematic category encoding.

[0099] Node style features within the same analysis task employ the same feature dimension order and normalization rules. The server can associate and save feature rule versions with node style features to identify different versions of analysis results after the thesaurus or parameters are updated.

[0100] S303. Based on the relationship between adjacent work nodes, compare the node style features of each group of adjacent work nodes in the same dimension to obtain the dimensional difference value of each group of adjacent work nodes.

[0101] In this context, same-dimensional comparison refers to comparing preceding and subsequent nodes on the same topic terms, the same geographical category, the same co-occurrence category, or the same topic encoding dimension. Dimensional difference values ​​can represent the degree to which corresponding features increase, decrease, or remain stable.

[0102] Extended Architecture Example: To address the issue of feature dimension mismatch caused by missing topic annotations or significant text length differences in some works, the server can generate valid dimension labels for each node's style features, performing comparisons only on dimensions that are valid in both nodes. New topics or new regional categories that are valid only in one node can be recorded separately as single-sided new dimensions. The server can also configure a trustworthy state for each dimension based on the text's completeness and the source of the topic attributes, ensuring that missing fields are not directly treated as style differences. This approach maintains the correspondence between comparison dimensions and reduces errors caused by missing data.

[0103] S304. Aggregate the dimensional difference values ​​to obtain the style changes of the adjacent nodes.

[0104] The server can aggregate dimensional difference values ​​according to feature groups such as theme, region, term co-occurrence, and spatial imagery. Then, the aggregation results of each feature group are used to form style variations of adjacent nodes according to a preset order of action or contribution coefficient. No specific mathematical formula is required during the aggregation process.

[0105] See Figure 9 In the example diagram comparing the style features of adjacent work nodes, the preceding work node 910 and the following work node 920 form the first node style feature 930 and the second node style feature 940, respectively; the same-dimensional comparison unit 950 compares the changes of the two nodes in the dimensions of theme terms, regional categories, spatial imagery and term co-occurrence; the difference aggregation unit 960 combines the effective dimensional differences into the style changes of adjacent nodes 970.

[0106] For example, the content of the preceding work node 910 mainly includes hometown scenery and local people, while the following work node 920 adds terms related to ports, voyages, and foreign locations. If the regional categories, spatial imagery, and co-occurrence of related terms in both nodes change in the same direction, the difference aggregation unit 960 can identify this change as a relatively obvious style change between adjacent nodes; if only a few low-frequency terms change, a lower degree of change can be retained.

[0107] This embodiment performs same-dimensional comparisons between adjacent work nodes defined by directed trajectory edges, and processes missing and added dimensions separately, so that style changes of adjacent nodes have clear data sources and comparison objects, which helps to improve the stability and interpretability of creative style migration judgment.

[0108] In one possible example, see Figure 10 The semantic feature quantization of the adjacent node analysis data in the aforementioned step S302 can be achieved through the following steps S401 to S404.

[0109] S401. Perform text normalization and term segmentation on the content of the works in the analysis data of the adjacent nodes to obtain the standardized text sequence of each work node.

[0110] Text standardization can include character encoding standardization, mapping between simplified and traditional Chinese characters, mapping between variant characters, handling of whitespace characters, and punctuation standardization. For poetry or works with a fixed paragraph structure, the boundaries of lines, sentences, and chapters can be preserved, and the entire text is not processed into an unstructured sequence of terms.

[0111] Term segmentation can be performed by combining general dictionaries, literary dictionaries, author proper names lists, and place name lists. Each term in a standardized text sequence can carry a work identifier, term position, sentence / segment position, and original text position.

[0112] S402. Match the standardized text sequence with the preset thesaurus and the travel lexicon to obtain the term feature groups of each work node.

[0113] The pre-set thesaurus can include thematic entries such as natural scenery, local memories, historical narratives, social life, and interpersonal relationships. The travel and expatriate thesaurus can include place names, travel activities, cross-regional spatial imagery, and their regional subcategories.

[0114] The term feature set can include the hit term, the hit location, the term source, the topic tag, and the regional subcategory tag. The same hit term can correspond to one or more candidate tags, and disambiguation is performed through term hierarchy and context.

[0115] S403. Determine word frequency features and word co-occurrence features based on the term feature groups, and encode the topic attributes in the adjacent node analysis data to obtain the node feature vector of each work node.

[0116] Among them, word frequency features represent the degree to which various hit terms appear in the work. Term co-occurrence features represent the degree to which two or more terms appear together in the same sentence, adjacent lines of poetry, or a preset text window. Topic attribute encoding can map topic categories and regional subcategories to fixed feature dimensions.

[0117] An extended architecture example of this embodiment: To address the issue of fixed co-occurrence windows being unsuitable due to the different structures of poetry, prose, and long narrative texts, the server can select the co-occurrence range based on the genre of the work. For poetry, the same line and adjacent lines can be used as the co-occurrence range; for prose, the same paragraph or adjacent paragraphs can be used; for longer works, local co-occurrence results can be formed by chapter first, and then aggregated into work-level co-occurrence features. By adapting the co-occurrence range to the text structure, the erroneous co-occurrence relationships of terms that are far apart and lack semantic connection can be reduced.

[0118] S404. Normalize each feature dimension of the node feature vector to obtain the node style features of each adjacent work node.

[0119] Normalization is used to reduce the impact of work length and different feature value ranges on the comparison results. The server can process word frequency, term co-occurrence, and topic encoding dimensions separately, and then combine them into node style features according to a fixed dimensional order.

[0120] As an enhanced example of steps S401 to S404, see Figure 11 The original work content 1110 is normalized and segmented into a standardized text sequence 1120; the thesaurus matching unit 1130 obtains thematic terms and travel-related terms from the standardized text sequence 1120; the term co-occurrence unit 1140 establishes co-occurrence relationships within sentences or adjacent lines of poetry; the theme encoding unit 1150 encodes existing theme attributes; and the node feature vector 1160 is normalized to form node style features 1170.

[0121] For example, in the standardized text sequence 1120, "port" and "voyage" appear in the same line of poetry, while "foreign land" and "letter" appear in adjacent lines. The term co-occurrence unit 1140 can establish co-occurrence relationships between "port" and "voyage" and between "foreign land" and "letter," respectively. The topic encoding unit 1150 can write the theme of travel and the corresponding regional subcategory into a fixed dimension, so that these contents are jointly included in the node feature vector 1160.

[0122] This embodiment forms word frequency features, word co-occurrence features, and topic encoding while preserving text structure and word position. It also obtains node style features with a unified dimension through normalization, enabling works of different genres and lengths to have comparable semantic representations.

[0123] Based on the aforementioned node style feature embodiment, in one possible example, the matching of the standardized text sequence with the preset topic thesaurus and the travel outside the domain thesaurus in step S402 can be achieved through the following steps S501 to S504.

[0124] S501. Perform continuous term and phrase recognition on the standardized text sequence to obtain a candidate term set.

[0125] Herein, the server may sequentially identify individual terms, continuous terms and fixed phrases according to term positions, and retain the start position, end position and sentence segment position of the candidate term in the work. The continuous term recognition may adopt a matching order from long to short, so as to preferentially retain phrases with complete semantics.

[0126] For example, when "Nanyang" and "Guihang" appear consecutively in the standardized text sequence, they may respectively form individual candidate terms, and may also form the candidate phrase "Nanyang Guihang". Subsequent lexicon matching may select the corresponding granularity according to the entry inclusion situation.

[0127] S502. Match the candidate term set with the entries and synonym entries in the preset theme lexicon and the overseas travel lexicon respectively, to obtain hit terms and hit positions.

[0128] Herein, the entries may include standard entries, alias entries, historical place name entries and synonym entries. After the matching is successful, the server retains the lexicon identifier, entry identifier and hit position corresponding to the hit term.

[0129] For cross-sentence segment candidate phrases, it may be judged according to the work genre and phrase rules whether they can be used as complete hit terms. Cross-sentence segment combinations that cannot form a stable semantic relationship may not be included in the hit results.

[0130] S503. Configure a classification label for the hit term according to the entry affiliation and hierarchical relationship of the hit term in the preset theme lexicon or the overseas travel lexicon, so as to obtain an annotated term.

[0131] Herein, the entry affiliation represents the theme category or region category to which the hit term belongs, and the hierarchical relationship represents the relationship of the entry among the superordinate category, subordinate category and synonym entries. The annotated term may store the original hit term, standard entry, classification label and hit position at the same time.

[0132] Example of an extended architecture of this embodiment: To solve the problem of label conflict caused by a term having a place meaning, a person meaning or an imagery meaning at the same time, the server may read the context terms before and after the hit position, and acquire the place information and existing theme attributes of the current work to form a candidate label context. The server first excludes candidate labels that are inconsistent with the entry hierarchy, and then compares the compatibility of the candidate labels with the context terms, place information and theme attributes. When the ambiguity cannot be reliably eliminated, multiple candidate labels and a to-be-confirmed state may be retained, instead of directly determining a unique label. This manner can reduce wrong theme annotations caused by polysemous words.

[0133] S504. Aggregate the labeled terms and the hit positions according to the work nodes to obtain the term feature groups of each work node.

[0134] The server can aggregate tagged terms within the same work according to the work identifier and restore the order of the terms in the sentence segment according to the hit position. The term feature group can also retain the number of times the term appears, the number of different positions, and the adjacency relationship with other tagged terms.

[0135] This embodiment forms term feature groups through continuous term recognition, lexicon matching, hierarchical annotation, and work-level aggregation. It retains not only the term classification results but also their text position and contextual relationships, thereby providing traceable input for subsequent term frequency, term co-occurrence, and topic feature processing.

[0136] Based on the aforementioned example of term feature groups, in one possible example, configuring classification labels for the hit terms in step S503 can be achieved through the following steps S601 to S604.

[0137] S601. Read the term affiliation and hierarchical relationship corresponding to the hit term from the preset topic thesaurus and the travel outside the domain thesaurus to obtain tag mapping data.

[0138] The tag mapping data may include the term identifier, superordinate category, subordinate category, synonyms, applicable context, and thesaurus version. The server can read the tag mapping data based on the standard term corresponding to the matched term.

[0139] S602. Determine the topic tag of the hit term based on the topic term affiliation in the tag mapping data.

[0140] Thematic tags can include natural scenery, local memories, travel abroad, historical narratives, social life, and interpersonal relationships. Thematic tags can be expanded or adjusted according to the research task, and are not limited to the examples above.

[0141] When a hit term belongs to multiple topic categories, the server can combine other terms near the hit location with the existing topic attributes of the current work to determine the main topic tags and auxiliary topic tags.

[0142] S603. Determine the regional subcategory label of the hit term based on the travel domain level in the label mapping data.

[0143] The regional subcategory labels include at least one of the following: Travelogue of Location A, Landscape of Location B, Travel Traces of Location C, and Travel in Location D.

[0144] Among them, regional subcategories can be determined based on the region where the standard location belongs, the travel route recorded in the work, or the regional affiliation of the entry. Travelogues in Place A, landscapes in Place B, travel traces in Place C, and travels in Place D are just examples. The thesaurus can also add other regional subcategories according to the research object.

[0145] For hit terms that only represent travel activities but whose location cannot be determined, you can configure an "Outside the Travel Domain" theme tag instead of a location subcategory tag.

[0146] S604. Associate at least one of the topic tag and the regional subcategory tag with the corresponding hit term to obtain the labeled term.

[0147] The associated tagged terms can include hit terms, standard terms, topic tags, regional subcategory tags, hit location, and tag source. The server can write the tagged terms into the term feature group of the corresponding work node.

[0148] This embodiment utilizes term attribution and hierarchical relationships to form topic tags and regional subcategory tags, ensuring that terms at the same terminology level are stably mapped to the same tags, and allowing terms whose region cannot be determined to retain only the higher-level topic, thus balancing tag granularity and annotation reliability.

[0149] Based on the aforementioned evaluation feature extraction embodiment, in a possible example, the determination of node creation intensity based on node evaluation features and edge weight based on edge evaluation features in step S105 can be achieved through the following steps S701 to S705.

[0150] S701. Read the location creation volume, local work density benchmark value and time slice weight corresponding to each work node from the node evaluation features to obtain node intensity calculation data.

[0151] The node strength calculation data can be organized according to either work node identifiers or location group identifiers. When using location group identifiers, multiple work nodes under the same location and time slice can reference the same node strength calculation data.

[0152] S702. Normalize the location creation quantity in the node intensity calculation data according to the local work density benchmark value to obtain the normalized creation quantity.

[0153] The local artwork density benchmark is used to provide a reference scale for the location dimension. The server can compare the relative degree of artwork volume at a location with the corresponding benchmark value and convert the comparison results of different locations to a unified value range.

[0154] When a local work density benchmark value is missing, reference values ​​from the same administrative region or adjacent regions can be read; if a reliable reference value still cannot be obtained, the corresponding node can be marked as a benchmark to be supplemented.

[0155] S703. Adjust the normalized creation amount according to the time slice weight to obtain the node creation intensity.

[0156] The time slice weight can reflect the degree of attention the current research task pays to different creative stages, as well as the coverage of data within the time slice. The server will adjust the results as the node creation intensity and associate them with the corresponding work nodes or location aggregation nodes.

[0157] S704. Read the location creation quantity, density of works outside the domain, and style changes of adjacent nodes corresponding to each directed trajectory edge from the edge evaluation features, and perform dimension normalization respectively to obtain the edge weight components.

[0158] The creation quantity corresponding to the directed trajectory edge can be the creation quantity of the location of the subsequent work node, or it can be a combination of the statistics of the preceding and subsequent locations. The same value selection rules are used for the same analysis task.

[0159] After each edge weight is normalized, the server can retain the location creation quantity component, the density component of works outside the region, and the style variation component to support geographic information terminals in viewing the basis for edge weight formation.

[0160] S705. The edge weight components are weighted and aggregated according to the preset contribution coefficient to obtain the edge weight, and the node creation intensity and the edge weight are output.

[0161] The preset contribution coefficient can be configured by the system default or by researchers according to the analysis objectives. Parameter configurations are saved in association with the analysis task identifier, thesaurus version, and time slicing scheme, enabling the reproducibility of analysis results under the same parameter conditions.

[0162] This embodiment performs benchmark normalization and dimension normalization before determining the node creation intensity and edge weight, so that the number of works at a location, the degree of the theme outside the region, and the degree of style variation can participate in the processing under a compatible scale, and retains the versions of each component and parameter to facilitate the interpretation and review of the evaluation results.

[0163] In one possible example, the spatiotemporal correlation modeling of the multidimensional artwork data in step S102 can be achieved through steps S801 to S808. See Figure 12In the example diagram of the multidimensional spatiotemporal data model and dual-database interoperability query, the spatiotemporal specification input 1210 is processed to form a structured spatiotemporal work record 1220; the structured spatiotemporal work record 1220 is entered into the spatial database 1230 and the graph database 1240 respectively; the time slice data 1250 is associated with the two types of data respectively; the cross-database mapping item 1260 connects the spatial record and graph node of the same work; the query arrangement unit 1270 calls the two types of databases according to the preset query order; the association result forms a multidimensional spatiotemporal data model 1280.

[0164] S801. Standardize the creation time, location name, and geographical coordinates in the multidimensional work data, and convert the geographical coordinates to a unified geographical coordinate system to obtain standardized spatiotemporal metadata.

[0165] The creation time standardization converts year, month, day, and time range into a unified time field, and saves the time value and time precision respectively. The location name standardization maps aliases, historical place names, and abbreviations to standard location identifiers, while retaining the original location names.

[0166] A unified geographic coordinate system can be used, such as the WGS84 coordinate system or other coordinate systems used in the study area. For locations whose coordinates cannot be directly determined, the administrative region center, candidate coordinates, and pending confirmation status can be linked first.

[0167] S802. Associate the standardized spatiotemporal metadata with the corresponding work content, work version, collection affiliation, spatial imagery, theme attribute and social attributes to obtain a structured spatiotemporal work record.

[0168] Each structured spatiotemporal work record must have at least one work identifier. When a work has multiple creation locations or creation times recorded, multiple candidate spatiotemporal records can be formed, and the relationship between the candidate records is maintained through source identifiers, trusted status, and master record identifiers.

[0169] Spatial imagery refers to descriptions of scenery, buildings, routes, or environments related to geographical space within the content of a work, which differs from the actual location where the work was created. The server stores the creation location and spatial imagery separately to avoid directly identifying locations appearing in the text as the creation location.

[0170] S803. Write the structured spatiotemporal works record into a spatial database that supports geographic range querying to obtain spatiotemporal works data.

[0171] The spatial database can be configured with fields for artwork identification, time, location, coordinates, spatial index, and time index. The server can filter spatiotemporal artwork records based on time intervals, geographical ranges, or a combination of both.

[0172] S804. Based on the work identifier, collection affiliation, theme attribute, and social attributes in the structured spatiotemporal work record, establish work nodes, work sequence relationships, travel route relationships, and call-and-response relationships in a graph database that supports relational extended queries, and obtain relational data.

[0173] Among these, the sequence relationship of works can represent the arrangement order or creation order within the same collection. The travel route relationship can represent the migration relationship between adjacent creation locations. The echoing relationship can represent the relationship of mutual reference, response, or shared event between two works.

[0174] Graph databases can also create person nodes, location nodes, and portfolio nodes, and use work identifiers to map work nodes to structured spatiotemporal work records.

[0175] S805. Divide the structured spatiotemporal works record according to the time boundaries corresponding to the author's life stages to obtain time slices, and write the corresponding time slice labels into the spatiotemporal works data and the relationship data.

[0176] The biographical phase can be determined based on chronological data, migration events, creative stages, or time boundaries set by researchers. Each time slice has a time slice identifier, start time, end time, and slice name.

[0177] When the creation time of a work spans two time slices, the time slice to which it belongs can be determined based on the main creation time, or reference records can be created in the two time slices respectively, while retaining the same work identifier.

[0178] S806. Write the same work identifier and time slice label into the corresponding spatiotemporal work record and work node to form a cross-database mapping item.

[0179] The cross-database mapping item includes at least the work identifier, spatial record identifier, graph node identifier, and time slice label. When a work corresponds to multiple spatial records, the cross-database mapping item can record the main spatial record and candidate spatial records.

[0180] S807. Based on the cross-database mapping item, establish a graph query entry from the spatiotemporal work data to the relational data and a spatial back-query entry from the relational data to the spatiotemporal work data, and configure the query order to first perform time range and spatial range filtering, and then perform work relationship expansion, to obtain the inter-database query relationship.

[0181] The graph query entry is used to convert the work identifiers returned by the spatial database into work node identifiers in the graph database. The spatial recursion entry is used to convert the work node identifiers obtained from the extended query of the graph database into corresponding spatial record identifiers.

[0182] For example, the server can first filter work records within the target time interval and target area in the spatial database 1230 to obtain a set of work identifiers; the query arrangement unit 1270 enters the graph database 1240 based on the cross-database mapping item 1260 to expand the query to include work sequence relationships, travel route relationships, and echo relationships related to the aforementioned works; for the newly obtained work nodes, their geographical locations are then read through the spatial back lookup entry. This query order can limit the initial scope of relationship expansion and maintain the correspondence between the relationship results and spatial locations.

[0183] S808. Associate the spatiotemporal work data, the relational data, the time slices, and the dual-database interoperability query relationship to obtain the multidimensional spatiotemporal data model.

[0184] The multidimensional spatiotemporal data model can be managed through model identifiers and version identifiers. When the thesaurus, time slices, or raw data are updated, the server can generate a new model version while retaining the cross-database mapping items and analysis records corresponding to the old version.

[0185] This embodiment sets up a bidirectional query entry based on work identifiers and time slice tags between the spatial database and the graph database, enabling spatial filtering results to continue to perform relation expansion, and relation expansion results to also be able to retrieve spatial locations, thereby maintaining the consistency of work-level spatiotemporal data and relation data.

[0186] Based on the aforementioned process of forming a hierarchical spatiotemporal trajectory network, in a possible example, the determination of the creative characteristic analysis results in step S107 based on the hierarchical spatiotemporal trajectory network, theme attributes, and social attributes can be achieved through the following steps S901 to S907.

[0187] S901. Receive the time interval, map scale, and region selection information sent by the geographic information terminal; extract the target trajectory subnet from the hierarchical spatiotemporal trajectory network according to the time interval, map scale, and region selection information; and obtain the theme attribute and interaction attribute associated with the target trajectory subnet to obtain target analysis data.

[0188] The target trajectory subnet includes work nodes that meet the screening criteria, directed trajectory edges, node creation intensity, and edge weights. The server reads relevant theme attributes and interaction attributes based on the work identifiers to ensure that the analysis data corresponds to the nodes in the target trajectory subnet.

[0189] S902. On a global scale, a spatial distribution heat map is generated based on the geographical location and creation intensity of the work nodes in the target analysis data.

[0190] The global scale indicates that the current map covers multiple countries, regions, or a large geographical area. The server can determine the degree of influence of a location on the heat map area based on the node creation intensity, and ensure a continuous transition in the display area of ​​adjacent locations.

[0191] S903. At the regional scale, cluster the work nodes in the target analysis data according to the theme attributes to generate a regional theme clustering layer and a theme word cloud layer.

[0192] The regional scale can correspond to a country, administrative region, or research area. The regional theme clustering layer is used to display the spatial clustering of nodes of works on different themes, and the theme word cloud layer is used to display representative theme terms within the target region.

[0193] The terms in the topic word cloud layer can be associated with a set of work identifiers. When a user selects a term, the terminal can display work nodes that contain that term.

[0194] S904. At the location scale, configure the location marker size according to the node creation intensity, and associate the work content with the corresponding work node to generate a location node layer.

[0195] The location scale can correspond to a city, district / county, specific location, or a location defined by researchers. The location marker size is used to characterize the node creation intensity of the corresponding location, and the location node layer can also display the number of works, time slices, and main themes.

[0196] After a user selects a location marker, the geographic information terminal can retrieve the work's title, original text, creation time, and source of information based on the work's identifier.

[0197] S905. Generate a curved trajectory layer based on the work nodes with the theme of "Traveling Abroad" and their corresponding directed trajectory edges, and configure the trajectory line width according to the amount of creation at each location.

[0198] The curve trajectory can be generated based on the geographic coordinates of adjacent locations, while maintaining the direction of the directed trajectory edges. The trajectory line width is used to represent the amount of creation at the corresponding location, and the trajectory salience can also be configured in conjunction with edge weights.

[0199] S906. Determine the interaction intensity based on the geographical location of the interaction objects, the frequency of responses, the number of collaborations, and the output of results in the interaction attributes. Configure the radiation range and connection line weights according to the interaction intensity to generate an interaction network radiation circle layer.

[0200] The individuals or groups to whom one interacts can be authors, editors, organizations, or other related figures. The frequency of exchanges, the number of collaborations, and the amount of output can be converted to a unified range first, and then the intensity of the interaction can be determined according to the pre-set relationships.

[0201] The radiation range can represent the spatial distribution range between the target author and the interaction object, and the weight of the connecting line can represent the relative activity level of the corresponding interaction relationship.

[0202] S907. Link the spatial distribution heat map layer, the regional theme clustering layer, the theme word cloud layer, the location node layer, the curve trajectory layer, and the social network radiation circle layer with the time axis, and render them in an electronic map that supports two-dimensional and three-dimensional switching to obtain the creative characteristic analysis results including the creative spatiotemporal distribution, theme migration path, and social network spatial diffusion trend, and output them to the geographic information terminal.

[0203] When the timeline changes, the server redetermines the target time slice and provides the terminal with the corresponding data for adding, deleting, or updating layers. The 2D and 3D maps use the same artwork identifiers, node creation intensity, and edge weights; only the spatial projection and rendering methods differ.

[0204] This embodiment forms adaptive analysis layers according to global, regional and location scales, and maintains the linkage between different layers through work identifiers and timelines, so that users can view the relationship between themes, works and people step by step from the overall spatial distribution, while keeping the data source of each display result consistent.

[0205] In one possible example, see Figure 13 The literary work author creation characteristic analysis device 1300 based on multidimensional information provided in this application embodiment includes a data acquisition module 1310, a spatiotemporal correlation modeling module 1320, an initial network construction module 1330, an evaluation feature extraction module 1340, an evaluation parameter determination module 1350, a trajectory network configuration module 1360, and an analysis output module 1370.

[0206] The data acquisition module 1310 is used to acquire multidimensional work data of the author; the multidimensional work data includes the content of the work as well as spatiotemporal, thematic attributes and social attributes.

[0207] The spatiotemporal correlation modeling module 1320 is used to perform spatiotemporal correlation modeling on the multidimensional work data acquired by the data acquisition module 1310 to obtain a multidimensional spatiotemporal data model.

[0208] The initial network construction module 1330 is used to obtain an initial spatiotemporal trajectory network based on the multidimensional spatiotemporal data model obtained by the spatiotemporal correlation modeling module 1320, with literary works as work nodes, the work nodes are layered according to time slices, and directed trajectory edges are established according to temporal and geographical location changes.

[0209] The evaluation feature extraction module 1340 is used to extract node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network. The node evaluation features include the amount of creation at a location, the local work density benchmark value, and the time slice weight. The edge evaluation features include the amount of creation at a location, the density of works outside the region, and the style changes of adjacent nodes.

[0210] The evaluation parameter determination module 1350 is used to determine the node creation intensity based on the node evaluation features extracted by the evaluation feature extraction module 1340, and to determine the edge weight based on the edge evaluation features extracted by the evaluation feature extraction module 1340.

[0211] The trajectory network configuration module 1360 is used to configure the node creation intensity and edge weight determined by the evaluation parameter determination module 1350 to the initial spatiotemporal trajectory network constructed by the initial network construction module 1330, so as to obtain a hierarchical spatiotemporal trajectory network.

[0212] The analysis output module 1370 is used to determine the creative characteristic analysis results and output them to the geographic information terminal based on the hierarchical spatiotemporal trajectory network obtained by the trajectory network configuration module 1360 and the theme attributes and social attributes obtained by the data acquisition module 1310.

[0213] The modules described above can be implemented by the processor executing program instructions from memory, or by communicating software services. Data is transferred between modules in the following order: multidimensional artwork data, multidimensional spatiotemporal data model, initial spatiotemporal trajectory network, evaluation features, evaluation parameters, hierarchical spatiotemporal trajectory network, and creative characteristic analysis results.

[0214] This embodiment implements the aforementioned method through a clear input-output relationship between functional modules, enabling data acquisition, spatiotemporal correlation, trajectory construction, evaluation processing, and result output to be executed on a single server, or distributed across multiple devices according to the processing load, and achieving the technical effects corresponding to the aforementioned method embodiment.

[0215] In the following example, author B is used as the target author, and the poems created by author B are considered literary works. The server can obtain multi-dimensional work data from collections such as "Collection 1," "Collection 2," "Collection 3," "Collection 4," "Collection 5," "Collection 6," "Collection 7," and "Collection 8," as well as author chronologies, prefaces and postscripts, geographical data, and social data. Each poem can form a corresponding work node, carrying a work identifier, creation time, creation location, geographical coordinates, collection affiliation, theme attributes, and social attributes. This example is only used to illustrate one specific application of the aforementioned method and does not constitute a limitation on the target author or the type of literary work.

[0216] See Figure 14 In the example geographical distribution of key spatiotemporal nodes for author B's poetry creation provided in this application embodiment, the schematic geographical base map 1410 is used to carry the work nodes determined according to standardized geographical coordinates. The server maps the geographical coordinates of the work nodes to the corresponding positions in the schematic geographical base map 1410, and organizes work nodes located in the same location or adjacent geographical areas into corresponding work node groups. At the example geographical display scale, multiple work nodes in the same location can be aggregated and displayed as location-level key nodes; this aggregation display does not change the independent work identifier of each work node in the multidimensional spatiotemporal data model and hierarchical spatiotemporal trajectory network.

[0217] The first regional work node group 1430 can include work nodes corresponding to locations such as City a in Country A, City b in Country A, and Country B. For example, work nodes from the period in City a in Country A in 1933 can be categorized into the early creation time slice; work nodes after moving to City b in Country A in 1949 can be categorized into the early time slice of City b in Country A; and work nodes created after the first trip to Country B in 1954 can be categorized into the travel time slice. The work nodes corresponding to City b in Country A can also serve as important spatiotemporal nodes connecting work nodes from different periods and regions.

[0218] The second regional work node group 1440 can include work nodes corresponding to locations such as Country D, Country E, and Country G. Among them, the work node of the "Third Collection" formed in Country D in 1966 can be organized according to the order of the works and the changes in the creation location; the work node of the "Eighth Collection" formed during the travels in Country E and Country G in 1976 can include work nodes corresponding to locations such as City p, City q, and City r in Country E.

[0219] The third regional work node group 1450 can include work nodes corresponding to locations such as City c in Country H, City d in Country H, and City f in Country H. For example, the work nodes of the "Fourth Collection" related to City c in Country H around 1970 and the work nodes of the "Fifth Collection" related to City d and City f in Country H in the 1960s and 1970s can be arranged according to their verifiable creation time and geographical coordinates.

[0220] The fourth region's work node group 1460 can include work nodes corresponding to locations such as country J, country K, and country L. Among them, work nodes related to travel in country L can be linked to "Collection 7". When the creation location can only be determined down to the country, region, or travel range, the server can display work nodes according to the corresponding geographic precision level and retain that geographic precision level to distinguish them from work nodes with specific city or scenic spot coordinates.

[0221] This example selects three themes from the thematic attributes carried by the work nodes: Traveling Abroad, Reflecting on History, and Landscapes, and distinguishes them using the theme category tag 1470. The three themes can be represented by different node outlines, internal fills, or combined line types, rather than using color as the sole distinguishing method. When works with different themes exist at the same location, different theme category tags can be displayed side-by-side at that location, or different theme category tags can be combined into the same location-level key node.

[0222] The node creation intensity marker 1480 is used to characterize the node creation intensity within a corresponding location and time slice. The server can determine the node creation intensity based on the location's creation volume, the local work density benchmark, and the time slice weight, and express the relative intensity through node size or the number of outer rings. For example, when the work nodes corresponding to "Third Collection" are relatively concentrated within the time slice of country D, and this time slice has a high analysis weight, the corresponding key nodes in country D can have a relatively high node creation intensity marker. Therefore, Figure 14 It can simultaneously present the geographical coverage, regional distribution differences, thematic category distribution, and location intensity of author B's poetry creation.

[0223] See Figure 15 In the visualization example of the travel trajectory of author B's poems on the theme of "traveling abroad" provided in this application embodiment, the server can filter work nodes representing the theme attribute of "traveling abroad" from the hierarchical spatiotemporal trajectory network, and form a sequence according to the time slice label, creation time and geographical location of the work node. Figure 15 The thematic trajectory is shown. The central node 1520 in City b of Country A represents the spatiotemporal reference position of author B, who conducted academic activities and poetry creation in different regions with City b of Country A as the main activity center. Directed trajectory edges 1580 can be established between the central node 1520 in City b of Country A and the work nodes in adjacent travel phases, based on the confirmed chronological order and geographical location changes in the structured spatiotemporal work records.

[0224] The first regional work node group 1430 can present the work nodes formed after the first trip to country B in 1954.

[0225] The second regional work node group 1440 can include work nodes from Country D's "Third Collection" in 1966 and work nodes from Country E and Country G's "Eighth Collection" in 1976. Work nodes in "Third Collection" can form a local trajectory according to the travel order in Country D, while retaining the social attributes; work nodes in "Eighth Collection" can form another local trajectory according to the travel order from south-central Country E to Country G.

[0226] The third region's work node group 1450 can include work nodes from the "Fourth Collection" related to City c in Country H, and work nodes from the "Fifth Collection" related to locations such as City d and City f in Country H. The server forms the node order based on the verifiable creation time of the corresponding work, and does not forcibly connect work nodes from different times or different collections into the same continuous trajectory just because the works are all located in the third region.

[0227] The fourth region's work node group 1460 can include travel-related work nodes from locations such as J country, K country, and L country. Among them, work nodes related to travel in L country can form a local trajectory according to the order of works recorded in the "Seventh Works Collection". When there are no work records between two regions that can confirm a continuous travel relationship, the server can preserve them as different trajectory segments to avoid inferring travel processes that are not supported by multidimensional work data based on geographical proximity.

[0228] The node marker 1570 for works on the theme of "Traveling Abroad" is assigned to each of the aforementioned work nodes to distinguish works on this theme from other works on other themes. In the black-and-white accompanying map, the node marker 1570 for works on the theme of "Traveling Abroad" can be represented by a uniform solid node. The position of the node in the schematic geographical base map 1410 corresponds to its creation location, and the arrangement order of the nodes along the directed trajectory edge 1580 corresponds to the creation time of the work or the verifiable chronological order within the collection of works.

[0229] Directed trajectory edge 1580 is used to connect temporally adjacent work nodes that have changed geographical locations, with the arrow direction indicating the temporal direction of the work node. Continuous directed trajectory segments can be formed within different work sets, while cross-time slice trajectory edges can be established between different time slices through work nodes with clear temporal succession relationships.

[0230] Edge weight marker 1590 is configured to the corresponding directed trajectory edge 1580. The server can determine the edge weight based on the amount of works created at a location, the density of works created outside the travel area, and the stylistic changes of adjacent nodes, and express the relative level of the edge weight through the line width or parallel outline of the directed trajectory edge. For example, when the trajectory segment corresponding to "Third Collection" has a high density of works created outside the travel area, and the works before and after this trajectory segment show a consistent direction of change in terms of regional category, spatial imagery, and term co-occurrence, the corresponding directed trajectory edge can have a relatively high edge weight marker; trajectory segments with only a few low-frequency term changes can retain a relatively low edge weight marker.

[0231] The server can also, based on the work's identifier, Figure 15The work nodes are associated with the work content, collection affiliation, theme attributes, and social attributes. When a user selects the "Third Collection" work node in the second region work node group 1440 through the geographic information terminal, they can obtain the corresponding work content, the time information of August 1966, the creation location in country D, and social attributes; when the user selects the third region work node group 1450, they can obtain the work records corresponding to the "Fourth Collection" or the "Fifth Collection" respectively.

[0232] pass Figure 15 The visualization example shown can present the distribution of author B's poems on the theme of traveling abroad, the direction of temporal migration, local travel trajectory and trajectory edge weights in the same schematic geographic view, so that the geographical location of the works, the degree of theme concentration and the stylistic changes of adjacent works can be combined to form traceable creative characteristic analysis results.

[0233] This application also provides a computer storage medium storing a computer program for electronic data interchange, which causes a computer to perform some or all of the steps of any of the methods described in the above method embodiments, wherein the computer includes an electronic device.

[0234] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0235] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0236] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0237] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0238] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0239] If the aforementioned integrated units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0240] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0241] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for analyzing the creative characteristics of literary works authors based on multidimensional information, characterized in that, Performed by the creation feature analysis server, including: Obtain multidimensional work data of the author; the multidimensional work data includes work content as well as spatiotemporal attributes, thematic attributes, and social attributes; Spatiotemporal correlation modeling is performed on the multidimensional artwork data to obtain a multidimensional spatiotemporal data model. This includes: standardizing the creation time, location name, and geographical coordinates in the multidimensional artwork data, and transforming the geographical coordinates to a unified geographical coordinate system to obtain standardized spatiotemporal metadata; associating the standardized spatiotemporal metadata with the corresponding artwork content, artwork version, artwork collection affiliation, spatial imagery, theme attribute, and social attributes to obtain structured spatiotemporal artwork records; writing the structured spatiotemporal artwork records into a spatial database that supports geographic range queries to obtain spatiotemporal artwork data; and establishing artwork nodes, artwork sequence relationships, travel route relationships, and call-and-response relationships in a graph database that supports extended relational queries based on the artwork identifiers, artwork collection affiliations, theme attributes, and social attributes in the structured spatiotemporal artwork records. The system obtains relational data; it divides the structured spatiotemporal work records according to the time boundaries corresponding to the author's life stages, obtains time slices, and writes the corresponding time slice tags into the spatiotemporal work data and the relational data; it writes the same work identifier and time slice tag into the corresponding spatiotemporal work records and work nodes to form a cross-database mapping item; based on the cross-database mapping item, it establishes a graph query entry from the spatiotemporal work data to the relational data and a spatial back lookup entry from the relational data to the spatiotemporal work data, and configures the query order to first perform time range and spatial range filtering, and then perform work relationship expansion, to obtain the dual-database interoperable query relationship; it associates the spatiotemporal work data, the relational data, the time slices, and the dual-database interoperable query relationship to obtain the multidimensional spatiotemporal data model; Based on the aforementioned multidimensional spatiotemporal data model, literary works are used as work nodes, and the work nodes are layered according to time slices. Directed trajectory edges are established according to temporal and geographical location changes to obtain an initial spatiotemporal trajectory network. Extracting node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network includes: reading the work content, geographical location, time slice, and theme attributes of the work nodes from the multidimensional spatiotemporal data model, and reading directed trajectory edges from the initial spatiotemporal trajectory network to obtain trajectory evaluation basic data containing the relationships between adjacent work nodes; grouping the work nodes in the trajectory evaluation basic data according to geographical location and time slice, counting the number of creations in each group, and reading the local work density benchmark value associated with geographical location and the time slice weight associated with time slice to obtain the node evaluation features; and based on the travel domain in the trajectory evaluation basic data... The density of works outside the domain is determined by the number of work nodes with external theme identifiers and the total number of work nodes in the corresponding group; the style changes of adjacent nodes are determined based on the work content and theme attributes of adjacent work nodes in the trajectory evaluation basic data; the location creation volume, the density of works outside the domain, and the style changes of adjacent nodes are combined into the edge evaluation features to obtain the node evaluation features and the edge evaluation features; the node evaluation features include the location creation volume, the local work density benchmark value, and the time slice weight; the edge evaluation features include the location creation volume, the density of works outside the domain, and the style changes of adjacent nodes; the local work density benchmark value represents the reference data used to evaluate the degree of aggregation of works at a location; The node creation intensity is determined based on the node evaluation characteristics, and the edge weight is determined based on the edge evaluation characteristics. The node creation intensity and the edge weight are configured into the initial spatiotemporal trajectory network to obtain a hierarchical spatiotemporal trajectory network. Based on the hierarchical spatiotemporal trajectory network and the thematic and social attributes, the creative characteristic analysis results are determined and output to the geographic information terminal.

2. The method according to claim 1, characterized in that, The step of determining the style changes of adjacent work nodes based on the work content and theme attributes of adjacent work nodes in the trajectory evaluation basic data includes: The work content and theme attributes of adjacent work nodes are read from the trajectory evaluation basic data to obtain adjacent node analysis data; Semantic feature quantization is performed on the analysis data of the adjacent nodes to obtain the node style features of each adjacent work node; Based on the relationship between adjacent work nodes, the node style characteristics of each group of adjacent work nodes are compared in the same dimension to obtain the dimensional difference value of each group of adjacent work nodes. The style changes of adjacent nodes are obtained by aggregating the dimensional difference values.

3. The method according to claim 2, characterized in that, The semantic feature quantization of the adjacent node analysis data to obtain the node style features of each adjacent work node includes: The text content of the works in the analysis data of the adjacent nodes is normalized and segmented into terms to obtain the standardized text sequence of each work node; The standardized text sequence is matched with a preset thesaurus and a travel lexicon to obtain the term feature groups of each work node; Based on the term feature groups, the term frequency features and term co-occurrence features are determined, and the topic attributes in the adjacent node analysis data are encoded to obtain the node feature vector of each work node; Normalize each feature dimension of the node feature vector to obtain the node style features of adjacent work nodes.

4. The method according to claim 3, characterized in that, The process of matching the standardized text sequence with a preset thesaurus and a travel-related thesaurus to obtain term feature groups for each work node includes: The standardized text sequence is subjected to continuous word and phrase recognition to obtain a candidate word set; The candidate word set is matched with the words and synonyms in the preset topic thesaurus and the travel outside the territory thesaurus to obtain the matched words and their positions. Based on the term affiliation and hierarchical relationship of the hit terms in the preset topic thesaurus or the travel outside the domain thesaurus, classification tags are configured for the hit terms to obtain labeled terms; By aggregating the labeled terms and the hit positions according to the work nodes, the term feature groups of each work node are obtained.

5. The method according to claim 4, characterized in that, The step of configuring category tags for the hit terms based on the term affiliation and hierarchical relationship in the preset topic thesaurus or the travel-related extraterritorial thesaurus, to obtain labeled terms, includes: The tag mapping data is obtained by reading the term affiliation and hierarchical relationship corresponding to the hit terms from the preset topic thesaurus and the travel outside the domain thesaurus; The topic tag of the hit term is determined based on the attribution of the topic entries in the tag mapping data; Based on the travel domain level in the tag mapping data, determine the regional subcategory tag of the hit term; The labeled terms are obtained by associating at least one of the topic tags and the regional subcategory tags with the corresponding hit terms.

6. The method according to any one of claims 1 to 5, characterized in that, The step of determining node creation intensity based on the node evaluation features and determining edge weights based on the edge evaluation features includes: The location creation volume, local work density baseline value and time slice weight corresponding to each work node are read from the node evaluation features to obtain node strength calculation data; The location creation quantity in the node intensity calculation data is normalized based on the local work density benchmark value to obtain the normalized creation quantity; The normalized creation amount is adjusted according to the time slice weight to obtain the node creation intensity; The location creation volume, density of works from outside the domain, and style changes of adjacent nodes corresponding to each directed trajectory edge are read from the edge evaluation features, and the dimensions are normalized to obtain the edge weight components. The edge weight components are weighted and aggregated according to a preset contribution coefficient to obtain the edge weight, and the node creation intensity and the edge weight are output.

7. The method according to any one of claims 1 to 5, characterized in that, The step of determining the creative characteristic analysis results and outputting them to the geographic information terminal based on the hierarchical spatiotemporal trajectory network, the theme attributes, and social interaction attributes includes: The system receives time interval, map scale, and region selection information sent by the geographic information terminal. Based on the time interval, map scale, and region selection information, it extracts the target trajectory subnet from the hierarchical spatiotemporal trajectory network and obtains the theme attributes and interaction attributes associated with the target trajectory subnet to obtain target analysis data. On a global scale, a spatial distribution heat map is generated based on the geographical location and creation intensity of the work nodes in the target analysis data. At the regional scale, the work nodes in the target analysis data are clustered according to the theme attributes to generate a regional theme clustering layer and a theme word cloud layer. At the location scale, the location marker size is configured according to the node creation intensity, and the work content is associated with the corresponding work node to generate a location node layer; Based on the work nodes with the theme of "Traveling Abroad" and their corresponding directed trajectory edges, a curved trajectory layer is generated, and the trajectory line width is configured according to the amount of creation at each location. The intensity of social interaction is determined based on the geographical location of the social interaction partners, the frequency of responses, the number of collaborations, and the output of results in the social interaction attributes. The radiation range and connection line weights are configured according to the social interaction intensity to generate a social interaction network radiation circle layer. The spatial distribution heat map, the regional theme clustering map, the theme word cloud map, the location node map, the curve trajectory map, and the social network radiation circle map are linked with the time axis and rendered in an electronic map that supports two-dimensional and three-dimensional switching to obtain the creative characteristic analysis results, including the spatiotemporal distribution of creation, theme migration path, and social network spatial diffusion trend, and output to the geographic information terminal.

8. A device for analyzing the creative characteristics of literary works authors based on multidimensional information, characterized in that, The method for analyzing the author's creative characteristics of literary works based on multidimensional information as described in any one of claims 1-7 includes: The data acquisition module is used to acquire multidimensional work data of the author; the multidimensional work data includes the content of the work as well as spatiotemporal, thematic, and social attributes. The spatiotemporal correlation modeling module is used to perform spatiotemporal correlation modeling on the multidimensional work data acquired by the data acquisition module to obtain a multidimensional spatiotemporal data model. The initial network construction module is used to obtain the initial spatiotemporal trajectory network based on the multidimensional spatiotemporal data model obtained by the spatiotemporal correlation modeling module, with literary works as work nodes, the work nodes are layered according to time slices, and directed trajectory edges are established according to temporal and geographical location changes. The evaluation feature extraction module is used to extract node evaluation features and edge evaluation features from the multidimensional spatiotemporal data model and the initial spatiotemporal trajectory network. The node evaluation features include the amount of creations at a location, the local work density benchmark value, and the time slice weight. The edge evaluation features include the amount of creations at a location, the density of works outside the region, and the style changes of adjacent nodes. The evaluation parameter determination module is used to determine the node creation intensity based on the node evaluation features extracted by the evaluation feature extraction module, and to determine the edge weight based on the edge evaluation features extracted by the evaluation feature extraction module. The trajectory network configuration module is used to configure the node creation intensity and edge weight determined by the evaluation parameter determination module to the initial spatiotemporal trajectory network constructed by the initial network construction module, so as to obtain a hierarchical spatiotemporal trajectory network. The analysis output module is used to determine the creative characteristic analysis results and output them to the geographic information terminal based on the hierarchical spatiotemporal trajectory network obtained by the trajectory network configuration module and the theme attributes and social attributes obtained by the data acquisition module.

Images