Cultural field-oriented knowledge graph space-time map display method

Abstract

The invention provides a cultural field-oriented knowledge graph space-time map display method, which comprises the following steps of: extracting culture entities and relationships thereof from multi-source heterogeneous data through a knowledge graph construction module, and constructing a structured knowledge graph; analyzing and normalizing the time and space information of the culture entity through a space-time processing module; through a map display module, displaying the processed culture entity and the time-space relationship thereof on a user terminal in a map form; through the interaction module, interaction operation between the user and the knowledge graph displayed in the map is achieved, and the user can more conveniently explore culture information; the method supports the display of multi-level cultural relationships, can effectively extract and display target information in a complex cultural network, has good applicability, can be widely applied to the fields of cultural heritage, historical research, archaeological research and the like, and enables cultural entities and relationships thereof to be more visually and dynamically presented in front of users.

Description

Technical Field

[0001] The present invention relates to the field of cultural information technology, and particularly to a method for displaying a spatio-temporal map of a knowledge graph for the cultural field. Background Art

[0002] In the field of cultural heritage protection and digital research, cultural entities and their spatio-temporal information have always been the core elements for constructing a research framework. With the in-depth application of technologies such as digital twin, 3D modeling, and geographic information system, the digitalization process of cultural resources is undergoing a transformation from static storage to dynamic analysis. The key challenges currently faced are how to construct an interdisciplinary technical system and how to effectively manage, display, and analyze this information, which has become a key issue.

[0003] The spatio-temporal display methods in the existing technologies usually can only provide information in a single time or space dimension, and it is difficult to intuitively display the correlation relationships of cultural entities in multiple dimensions. Moreover, the traditional knowledge graph and map display methods in the cultural field cannot realize the spatio-temporal relationships of cultural entities and their interactive display, resulting in the following problems: it is difficult to dynamically display the spatio-temporal changes of cultural entities; there is a lack of support for multi-level correlations, and it is difficult to display complex cultural relationship networks; the interactivity between users and the displayed content is poor, and the needs of in-depth exploration of cultural information cannot be met. Summary of the Invention

[0004] Aiming at the deficiencies of the existing technologies, the purpose of the present invention is to provide a method for displaying a spatio-temporal map of a knowledge graph for the cultural field to solve the problems raised in the above background art. The present invention enables cultural entities and their relationships to be presented more intuitively and dynamically in front of users, and cultural information can be explored more conveniently, with good applicability.

[0005] To achieve the above purpose, the present invention is implemented by the following technical solutions: A method for displaying a spatio-temporal map of a knowledge graph for the cultural field, the display method comprising the following steps:

[0006] Step 1: Extract cultural entities and their relationships from multi-source heterogeneous data through a knowledge graph construction module to construct a structured knowledge graph;

[0007] Step 2: Parse and standardize the time and space information of cultural entities through a spatio-temporal processing module;

[0008] Step 3: Through a map display module, display the processed cultural entities and their spatio-temporal relationships in the form of a map on a user terminal;

[0009] Step 4: Implement interactive operations between users and the knowledge graph displayed in the map through an interaction module.

[0010] Further, the knowledge graph construction module in the first step includes a data cleaning unit, an entity extraction unit, and a relationship extraction unit.

[0011] Further, the data cleaning unit is used to clean multi-source data; the entity extraction unit is used to extract cultural entities from the cleaned data; the relationship extraction unit is used to identify and mark the relationships between different cultural entities, and the relationship extraction unit introduces a spatio-temporal decay factor and develops a spatio-temporal aware relationship strength calculation model:

[0012]

[0013] Where S(t) is the signal strength at time t (such as cultural influence, information dissemination strength, etc.), S0 is the strength at the initial time t0, λ is the time decay coefficient, which controls the exponential decay rate over time, α is the space decay coefficient, which controls the linear decay of the strength with distance, d1 is a correction term that may be related to distance, or cultural relationship features extracted by the LLM.

[0014] Further, the spatio-temporal processing module in the second step includes a time parsing unit, a space parsing unit, a spatio-temporal backtracking reasoning unit, and a heterogeneous data fusion unit.

[0015] Further, the time parsing unit is used to convert non-standard time information into a standard format; the space parsing unit is used to geocode the place name and convert it into longitude and latitude coordinates; the spatio-temporal backtracking reasoning unit is used to generate the influence propagation path of the entity in the spatio-temporal dimension; a cultural data federated learning framework is established in the heterogeneous data fusion unit, and a multi-modal alignment mechanism is introduced in the data cleaning unit.

[0016] Further, the space parsing unit encodes the geographic coordinate information, and the data is parsed for geographic location encoding using the national open map or a third-party commercial map.

[0017] Further, it also includes building a self-built collection place name database through the space parsing unit, and combining fuzzy matching and the shortest edit distance algorithm to improve the parsing rate.

[0018] Further, the map display module in the third step is used to implement map display. During this process, the spatio-temporal data is stored in the database and geographic queries are supported.

[0019] Further, the map display module also provides dynamic display, uses D3.js to implement the time axis interaction animation, adjusts the map display according to the time period selected by the user, and dynamically loads the data for this time period through a front-end request.

[0020] Further, the interaction module in step 4 includes functions of magnification, reduction, and dragging; timeline sliding function; entity information query and multi-level display function.

[0021] Advantages of the present invention:

[0022] 1. Compared with traditional maps that only display positioning coordinate information, the method for displaying the spatio-temporal map of the knowledge graph for the cultural field enables cultural entities and their relationships to be presented more intuitively and dynamically to users through the combination of the spatio-temporal map and the knowledge graph.

[0023] 2. The method for displaying the spatio-temporal map of the knowledge graph for the cultural field improves the user's interaction experience, supports map zooming, dragging, and spatio-temporal sliding queries, enabling users to more conveniently explore cultural information; supports the display of multi-level cultural relationships and can effectively extract and display target information in a complex cultural network.

[0024] 3. The present invention achieves an accuracy rate of 92.3% in fuzzy time parsing, a 37% improvement compared to traditional NER methods, shortens the spatio-temporal query response time to 120 ms, and supports real-time interaction of tens of millions of entities.

[0025] 4. The method for spatio-temporal display of the knowledge graph specific to the cultural field provided by the present invention has good applicability and can be widely applied to fields such as cultural heritage, historical research, and archaeological research. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flowchart of a method for displaying a spatio-temporal map of a knowledge graph for the cultural field according to the present invention;

[0027] Figure 2 is a schematic diagram of the entity extraction result in an embodiment of the present invention;

[0028] Figure 3 is a schematic diagram of the relationship extraction result in an embodiment of the present invention;

[0029] Figure 4 is the process of cleaning, fusing, and constructing the knowledge graph in an embodiment of the present invention.

[0030] Figure 5 is the spatio-temporal map of the knowledge graph story in an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0031] In order to make the technical means, creative features, achieved objectives, and functions of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0032] Please refer to Figures 1 to 5, the present invention provides the following technical solutions: A method for displaying the spatio-temporal map of a knowledge graph in the cultural field, which method includes four units: a knowledge graph construction module, a spatio-temporal processing module, a map display module, and an interaction module, and specifically includes the following content:

[0033] 1. Through the knowledge graph construction module, cultural entities and their relationships are extracted from multi-source heterogeneous data to construct a structured knowledge graph. Among them, the knowledge graph construction module includes:

[0034] 1.1 A data cleaning unit, used for cleaning multi-source data;

[0035] 1.2 An entity extraction unit, used for extracting cultural entities from the cleaned data;

[0036] 1.3 A relationship extraction unit, used for identifying and marking the relationships between different cultural entities.

[0037] 2. Through the spatio-temporal processing module, the time and space information of cultural entities is analyzed and normalized. Among them, the spatio-temporal processing module includes:

[0038] 2.1 A time analysis unit, used for converting non-standard time information into a standard format;

[0039] 2.2 A space analysis unit, used for geocoding the place name and converting it into longitude and latitude coordinates.

[0040] 3. Through the map display module, the processed cultural entities and their spatio-temporal relationships are displayed in the form of a map on the user terminal. Among them, the functions included in the interaction module are: zoom in, zoom out, and drag functions; timeline sliding function; entity information query and multi-level display function.

[0041] 4. Through the interaction module, the user is enabled to perform interactive operations with the knowledge graph displayed in the map.

[0042] In this embodiment, the implementation details of the above knowledge graph construction module are also provided, specifically as follows:

[0043] 1. Data cleaning unit

[0044] Perform string matching on the entities and attributes of multi-source data to remove duplicate records, and this process is deduplication processing. Use regular expressions to filter out irrelevant information, such as HTML tags, noise characters, etc., and this process is denoising processing. Use the Pandas library to standardize data in diverse formats (such as XML, CSV, JSON) into a unified format, and this process is data format conversion. Fill in missing values through mean, median filling or prediction based on collaborative filtering algorithms (such as KNN algorithm), and this process is filling of missing values.

[0045] 2. Entity extraction unit

[0046] First, extract template-like regular data samples in the form of rules, and set matching rules (such as regular expressions) to extract specific cultural entities. For non-regular form data, use deep learning named entity recognition NER models (such as BERT, BiLSTM-CRF, etc.) to train a specific domain entity recognizer.

[0047] 3. Relationship extraction unit

[0048] Based on the entity co-occurrence relationship, count the frequency of their simultaneous occurrence and context information. Then define and construct a semantic model, defining relationships such as "located in" (Location), "belong to" (BelongTo), etc. Parse the syntactic structure through dependency syntactic analysis to identify the subject, predicate, and object relationships. A spatio-temporal decay factor can also be introduced in the relationship extraction unit to develop a spatio-temporal aware relationship strength calculation model as follows:

[0049]

[0050] Where S(t) is the signal strength at time t (such as cultural influence, information dissemination strength, etc.), S0 is the strength at the initial time t0, λ is the time decay coefficient, controlling the exponential decay rate over time, α is the spatial decay coefficient, controlling the linear decay of the strength with distance, d1 is a correction term that may be related to distance, or cultural relationship features extracted through the LLM.

[0051] In this embodiment, the implementation details of the above spatio-temporal processing module are also provided, as follows:

[0052] 4. Time parsing unit

[0053] For texts with clear time descriptions, based on pattern matching, design a time pattern library (such as regular expressions supporting "the 15th year of Jiaqing", etc.). Its conversion method is as follows: "the 15th year of Jiaqing" -> 1810 (obtain the Gregorian year by looking up the table).

[0054] For texts with vague time descriptions, such as "about xx years", normalize them to a specified range, such as converting "the early 20th century" to 1900 - 1910. In this unit, construct a knowledge-enhanced spatio-temporal disambiguation algorithm, and develop a probability event inference model that integrates historical event anchor points for vague expressions in Chinese ancient books such as "around the Kangxi period" and "the late Ming and early Qing dynasties". For example, by associating with the historical event time axis (such as "Li Zicheng Uprising (1627 - 1645)"), establish a time probability distribution network based on event association.

[0055] 5. Space parsing unit

[0056] Encode geographical coordinate information. The data can use the national open map or use third-party commercial maps such as Baidu Map API and Amap API for geographical location coding and parsing. At the same time, a collection place name database can also be built independently, and the parsing rate can be improved by combining fuzzy matching and the shortest edit distance algorithm.

[0057] 6. Spatiotemporal backtracking reasoning unit

[0058] Deeply associate and reason about spatiotemporal data combined with the content of the knowledge base to encapsulate the spatiotemporal backtracking reasoning engine. The "relationship tracing" function can be added to the interaction module. When the user clicks on a certain cultural entity, the system automatically generates the influence propagation path of the entity in the spatiotemporal dimension and visually displays the ripple effect of cultural influence.

[0059] 7. Heterogeneous data fusion unit

[0060] Establish a cultural data federated learning framework, introduce a multimodal alignment mechanism in the data cleaning unit. For example, perform dynamic conversion on "one li" in ancient books (in the Ming Dynasty, 1 li ≈ 576 meters; in the Qing Dynasty, 1 li ≈ 559 meters), combine historical GIS data for spatial calibration, and then adopt an improved HR+ tree index structure at the database layer, combined with the knowledge reasoning path query algorithm of the knowledge graph database, to improve the query efficiency from O(n) to O(log n).

[0061] In this embodiment, the implementation details of the above map display module are also provided as follows:

[0062] 1. Map display implementation

[0063] Store spatiotemporal data in a database (such as PostGIS) to support geographical queries. The WebGIS framework for front-end rendering can use Leaflet or OpenLayers to implement map rendering and data loading. Its mapping logic is that cultural entities are marked as points by longitude and latitude, and polygon shapes are used to display complex geographical information (such as historical regional boundaries).

[0064] 2. Dynamic display

[0065] Use D3.js to implement the time-axis interaction animation, adjust the map display according to the time period selected by the user, and dynamically load the data of this time period through front-end requests (AJAX / WebSocket). The map rendering data supports various knowledge instances and relationship contents of the knowledge graph, and can organize the corresponding knowledge expressions according to the relevant information of the instances, such as the occurrence process of historical events, the historical evolution of buildings, and the life stories of figures.

[0066] In this embodiment, the basic principles, main features, and advantages of the present invention have been shown and described. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0067] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative manner of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for displaying a knowledge graph spatiotemporal map in the cultural field, characterized in that: The display method includes the following steps: Step 1: Extract cultural entities and their relationships from multi-source heterogeneous data through the knowledge graph construction module to build a structured knowledge graph; Step 2: Analyze and normalize the time and space information of cultural entities through the time and space processing module; Step 3: Through the map display module, the processed cultural entities and their spatiotemporal relationships are displayed on the user terminal in the form of a map; Step 4: Use the interactive module to enable users to interact with the knowledge graph displayed in the map.

2. According to claim 1, a method for displaying a knowledge graph spatiotemporal map in the cultural field is characterized by: The knowledge graph construction module in step one includes a data cleaning unit, an entity extraction unit, and a relationship extraction unit.

3. According to claim 2, a method for displaying a knowledge graph spatiotemporal map in the cultural field is characterized by: The data cleaning unit is used to clean multi-source data; the entity extraction unit is used to extract cultural entities from the cleaned data; the relationship extraction unit is used to identify and mark the relationship between different cultural entities, and the relationship extraction unit introduces a spatiotemporal attenuation factor to develop a spatiotemporal-aware relationship strength calculation model: Where S(t) is the signal strength at time t, S0 is the strength at the initial time t0, λ is the time attenuation coefficient, which controls the exponential decay rate over time, α is the spatial attenuation coefficient, which controls the linear attenuation of intensity with distance, and d1 is a correction term that may be related to distance, or a cultural relationship feature extracted by LLM.

4. According to claim 1, a method for displaying a knowledge graph spatiotemporal map in the cultural field is characterized by: The spatiotemporal processing module in step 2 includes a time parsing unit, a space parsing unit, a spatiotemporal backtracking reasoning unit, and a heterogeneous data fusion unit.

5. According to claim 4, a method for displaying a knowledge graph spatiotemporal map in the cultural field is characterized by: The time parsing unit is used to convert non-standardized time information into a standardized format; the space parsing unit is used to geocode place names and convert them into longitude and latitude coordinates; the space-time backtracking reasoning unit is used to generate the influence propagation path of entities in the space-time dimension; a cultural data federated learning framework is established in the heterogeneous data fusion unit, and a multimodal alignment mechanism is introduced in the data cleaning unit.

6. The method for displaying a knowledge graph spatiotemporal map in the cultural field according to claim 5, characterized in that: The spatial resolution unit encodes the geographic coordinate information, and the data is coded and resolved for geographic location using a national open map or a third-party commercial map.

7. The method for displaying a knowledge graph spatiotemporal map in the cultural field according to claim 6, characterized in that: It also includes building a database of place names through the spatial resolution unit, combining fuzzy matching and the shortest edit distance algorithm to improve the resolution rate.

8. According to claim 1, a method for displaying a knowledge graph spatiotemporal map in the cultural field is characterized by: The map display module in step three is used to realize map display. In this process, the spatiotemporal data is stored in the database and geographic query is supported.

9. The method for displaying a knowledge graph spatiotemporal map in the cultural field according to claim 8, characterized in that: The map display module also provides dynamic display, uses D3.js to implement timeline interactive animation, adjusts the map display according to the time period selected by the user, and dynamically loads the time period data through the front-end request.

10. The method for displaying a knowledge graph spatiotemporal map in the cultural field according to claim 1, characterized in that: The interactive module in step 4 includes zooming in, zooming out, and dragging functions; a timeline sliding function; and entity information query and multi-level display functions.

Images