A method for discovering an unregistered place name and inferring a spatial position and a related device

Abstract

The application discloses a method and related device for discovering unregistered place names and reasoning spatial positions, and relates to the technical field of geographic information. The method comprises the following steps: constructing a place name space-time derivation relationship knowledge graph based on an open source geographic database; inputting target place name text data into a sequence labeling model to obtain a labeling result, and extracting candidate derived place names based on the labeling result; performing common name segmentation on the candidate derived place names to obtain potential basic place names, and judging whether the potential basic place names meet preset association conditions based on the place name space-time derivation relationship knowledge graph; if yes, the potential basic place names are excluded; if no, the potential basic place names are determined as unregistered place names; and determining the spatial positions of the unregistered place names according to the common names of the unregistered place names and the place name space-time derivation relationship knowledge graph. The application can improve the automation degree of discovering unregistered place names and reasoning spatial positions.

Description

Technical Field

[0001] This application relates to the field of geographic information technology, and in particular to a method and related apparatus for discovering unregistered place names and inferring spatial location. Background Technology

[0002] With the acceleration of globalization, the demand for high-precision, comprehensive geographic information data is becoming increasingly urgent. In recent years, several large-scale geographic information construction projects have been launched internationally, aiming to build high-precision geographic information databases covering the globe to support economic and social development and international exchange and cooperation. These databases typically consist of various basic data, including digital orthophotos, digital land models, place names and administrative divisions, and land cover. Among these, place name data, as a key type of basic geographic information, reflects the distribution of natural resources, geographical environmental characteristics, local customs, and historical changes in a specific region, thus possessing particular significance. Place names are essentially the product of people conceptualizing and naming geographical elements or entities. They provide users with an intuitive and efficient way to identify and access geographical locations, greatly enhancing the retrieval, analysis, and visualization capabilities of geographic information data.

[0003] Despite the crucial role of place name data in global geographic information resources, its collection and database construction still face significant challenges. Particularly in some economically underdeveloped regions, the slow progress of geographic information infrastructure development, constrained by the long cycles and high costs of traditional surveying methods, has resulted in incomplete coverage and outdated data in existing place name databases. A large number of place names are not officially included in authoritative databases; these are often referred to as "unregistered place names." This phenomenon severely restricts the completeness and accuracy of geographic information systems, directly impacting applications such as urban planning, intelligent navigation, emergency disaster relief, and resource allocation. Filling in these missing place names is of profound significance for improving regional geographic information services, promoting economic and social development, and especially supporting the development of underdeveloped regions.

[0004] The academic community has conducted numerous exploratory studies on the rich linguistic and cultural phenomena inherent in place names. International scholars have analyzed the relationship between place name derivation and regional culture from the perspectives of linguistics and cultural geography, and have conducted comparative studies on the characteristics of derived affixes in different language families. In addition, domestic researchers have systematically sorted out the generation mechanism and classification system of derived place names, providing a theoretical basis for place name standardization. However, existing research mostly focuses on theoretical analysis and phenomenon summarization, and lacks technical solutions that can be practically applied to the identification and completion of unregistered place names, making it difficult to effectively support large-scale, automated place name data construction and maintenance.

[0005] Therefore, there is an urgent need to study an efficient and reliable technical method to solve the problem of missing data in existing place name databases, improve the coverage and data quality of global geographic information resources, and meet the growing application needs of multiple fields. Summary of the Invention

[0006] The purpose of this application is to provide a method and related apparatus for discovering unregistered place names and inferring spatial location, which can improve the automation level of discovering unregistered place names and inferring spatial location.

[0007] To achieve the above objectives, this application provides the following solution:

[0008] Firstly, this application provides a method for discovering unregistered place names and inferring their spatial location, including:

[0009] Based on the acquired open-source geographic database, a knowledge graph of place name spatiotemporal derivation relationships is constructed. The open-source geographic database includes multiple original place name data, which at least include place names, geographic entity types, and spatial coordinates. The knowledge graph of place name spatiotemporal derivation relationships includes entities, attributes, and relationships. The entities are place names, the attributes are proper names, common names, and spatial coordinates, and the relationships are spatial relationships and semantic similarity.

[0010] The acquired target place name text data is input into the sequence labeling model to obtain the labeling results. Based on the labeling results, candidate derived place names are extracted. The sequence labeling model is a combination of a trained bidirectional pre-trained language model and a conditional random field. The labeling results are the labeling of proper nouns, generic nouns and directional words in the place names. The candidate derived place names are composite place names that include multiple generic noun combinations or combinations of generic nouns and directional words.

[0011] The candidate derived place names are segmented into common names to obtain potential basic place names. Based on the place name spatiotemporal derivation relationship knowledge graph, it is determined whether the potential basic place names meet the preset association conditions. If yes, the potential basic place names are excluded. If no, the potential basic place names are determined to be unregistered place names. Unregistered place names are potential basic place names that do not exist in the open source geographic database.

[0012] Based on the generic name of the unregistered place name, determine the geographic entity type of the unregistered place name, and based on the geographic entity type of the unregistered place name and the place name spatiotemporal derivation relationship knowledge graph, use predefined mapping rules to determine the spatial representation and spatial coordinates of the corresponding original place name;

[0013] Based on the spatial representation and spatial coordinates of the original place names, a geometric algorithm is used to determine the spatial location of unregistered place names.

[0014] Optionally, based on the acquired open-source geographic database, a knowledge graph of the spatiotemporal derivation relationship of place names is constructed, specifically including:

[0015] Based on the acquired open-source geographic database, a knowledge graph of place name spatiotemporal derivation relationships is constructed using top-down and bottom-up methods. The top-down method defines the entities, attributes, and relationships in the knowledge graph. The bottom-up method extracts entities, attributes, and relationships from the open-source geographic database according to its data type, and then cleans and merges these entities, attributes, and relationships before filling them into the knowledge graph.

[0016] Optionally, based on the data type of the open-source geodatabase, entity, attribute, and relationship extraction is performed on the open-source geodatabase, specifically including:

[0017] When the open-source geodatabase is structured data, spatial database or geographic information system tools are used to automatically extract entities and spatial relationships between entities, and text similarity algorithms are used to calculate the semantic similarity between entities.

[0018] When the open-source geodatabase contains unstructured data, natural language processing techniques are used to extract entities and the spatial relationships between them.

[0019] Optionally, the candidate derived place names are subjected to generic segmentation to obtain potential base place names, specifically including:

[0020] Candidate derived place names are segmented into generic terms according to predefined derivation patterns to obtain potential base place names; the derivation patterns include “locative term + proper name + derived generic term”, “derived generic term + proper name + locative term”, “numeral + proper name + derived generic term”, “derived generic term + proper name + numeral”, “proper name + original generic term + derived generic term”, and “derived generic term + original generic term + proper name”.

[0021] Optionally, the preset association conditions specifically include:

[0022] The correlation between the geographic entity type of the potential base place name and the geographic entity type of the corresponding original place name in the place name spatiotemporal derivation relationship knowledge graph is greater than the first threshold.

[0023] The common name similarity between the potential base place name and the original place name is greater than the second threshold;

[0024] The proper name similarity between the potential base place name and the original place name is greater than the third threshold;

[0025] The similarity of geographic entity types between the potential base place name and the original place name is greater than the fourth threshold.

[0026] Optionally, based on the spatial representation and spatial coordinates of the original place names, a geometric algorithm is used to determine the spatial location of unregistered place names, specifically including:

[0027] Based on the spatial representation of the original place name, the minimum bounding rectangle of the original place name is calculated on the boundary corresponding to the directional word of the original place name, and the minimum bounding rectangle is determined as the first predicted position.

[0028] Based on the spatial coordinates of the original place name, a buffer zone with a fixed radius is determined, and the buffer zone with the fixed radius is determined as the second predicted location;

[0029] Based on the first predicted location and the second predicted location, the spatial location of the unregistered place name is determined.

[0030] Optionally, the method for discovering unregistered place names and inferring spatial location further includes:

[0031] The discovered unregistered place names and their spatial locations are compared with the acquired multi-source verification data. The accuracy rate of unregistered place name discovery and the reasonableness rate of spatial location inference are calculated through manual evaluation, simulation verification, or multi-source data comparison.

[0032] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the unregistered place name discovery and spatial location reasoning method described in any one of the above.

[0033] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the unregistered place name discovery and spatial location reasoning method described above.

[0034] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the unregistered place name discovery and spatial location reasoning method described above.

[0035] According to the specific embodiments provided in this application, this application has the following technical effects:

[0036] This application provides a method and related apparatus for discovering and spatially inferring unregistered place names. It uses a sequence labeling model to annotate target place name text data, automatically identifying proper names, generic names, and directional words, and extracting candidate derived place names, reducing manual intervention and improving the accuracy and efficiency of place name recognition. Furthermore, by segmenting candidate derived place names into generic terms and combining this with a knowledge graph of place name spatiotemporal derivation relationships to determine the association conditions of potential base place names, it accurately identifies and excludes registered place names, ultimately determining unregistered place names. Simultaneously, based on the geographic entity type of the unregistered place names, it utilizes spatial relationships and coordinate information from the knowledge graph, combined with geometric algorithms, to accurately infer the spatial location of the unregistered place names, ensuring high accuracy in geographic location inference. Moreover, the automated sequence labeling and inference process significantly improves the automation level of place name discovery and spatial location inference, reducing the need for human intervention and adapting to large-scale data processing applications. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is an application environment diagram of an unregistered place name discovery and spatial location reasoning method according to an embodiment of this application;

[0039] Figure 2 A flowchart illustrating a method for discovering unregistered place names and inferring spatial location, provided as an embodiment of this application;

[0040] Figure 3 A schematic diagram of a knowledge graph of spatiotemporal derivation relationships of place names provided in an embodiment of this application;

[0041] Figure 4 A schematic diagram illustrating an example of an unregistered place name list with spatial information provided in an embodiment of this application;

[0042] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] The method for discovering unregistered place names and inferring spatial location provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on other servers. Terminal 102 can send an open-source geographic database and target place name text data to server 104. Server 104 constructs a place name spatiotemporal derivation relationship knowledge graph based on the open-source geographic database; inputs the target place name text data into a sequence labeling model to obtain labeling results, and extracts candidate derived place names based on the labeling results; performs generic segmentation on the candidate derived place names to obtain potential basic place names, and determines whether the potential basic place names meet preset association conditions based on the place name spatiotemporal derivation relationship knowledge graph; if yes, the potential basic place names are excluded; if not, the potential basic place names are identified as unregistered place names; and determines the spatial location of the unregistered place names based on their generic names and the place name spatiotemporal derivation relationship knowledge graph.

[0046] The terminal 102 can be, but is not limited to, various desktop computers, laptops, and IoT devices. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers, or it can be a cloud server.

[0047] In one exemplary embodiment, such as Figure 2 As shown, a method for discovering unregistered place names and inferring spatial location is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 205. Wherein:

[0048] Step 201: Based on the acquired open-source geographic database, construct a knowledge graph of place name spatiotemporal derivation relationships; the open-source geographic database includes multiple original place name data, each of which includes at least a place name, geographic entity type, and spatial coordinates; the knowledge graph of place name spatiotemporal derivation relationships is as follows: Figure 3 As shown, it includes entities, attributes, and relationships. Entities are place names, attributes are proper names, common names, and spatial coordinates, and relationships are spatial relationships and semantic similarity.

[0049] Step 202: Input the acquired target place name text data into the sequence labeling model to obtain the labeling results, and extract candidate derived place names based on the labeling results; the sequence labeling model is a combination of a trained bidirectional pre-trained language model and a conditional random field, and the labeling results are the labeling of proper nouns, generic nouns and directional words in the place name, and the candidate derived place names are composite place names including combinations of multiple generic nouns or combinations of generic nouns and directional words.

[0050] Step 203: Perform generic segmentation on the candidate derived place names to obtain potential basic place names, and determine whether the potential basic place names meet the preset association conditions based on the place name spatiotemporal derivation relationship knowledge graph; if yes, exclude the potential basic place names; if no, determine the potential basic place names as unregistered place names; unregistered place names are potential basic place names that do not exist in the open source geographic database.

[0051] Step 204: Based on the generic name of the unregistered place name, determine the geographic entity type of the unregistered place name, and based on the geographic entity type of the unregistered place name and the knowledge graph of the spatiotemporal derivation relationship of place names, use predefined mapping rules to determine the spatial representation and spatial coordinates of the corresponding original place name.

[0052] Step 205: Based on the spatial representation and spatial coordinates of the original place names, use a geometric algorithm to determine the spatial location of the unregistered place names.

[0053] By implementing steps 201 to 205 above, this application uses a sequence labeling model to annotate the target place name text data, automatically identifying proper names, generic names, and locative words, extracting candidate derived place names, reducing manual intervention, and improving the accuracy and efficiency of place name recognition. Furthermore, by segmenting candidate derived place names into generic names and combining them with a knowledge graph of place name spatiotemporal derivation relationships to determine the association conditions of potential base place names, it accurately identifies and excludes registered place names, ultimately determining unregistered place names. Simultaneously, based on the geographic entity type of unregistered place names, it utilizes spatial relationships and coordinate information in the knowledge graph, combined with geometric algorithms, to accurately infer the spatial location of unregistered place names, ensuring high accuracy in geographic location reasoning. In addition, the automated sequence labeling and reasoning process greatly improves the automation level of place name discovery and spatial location reasoning, reducing the need for human intervention and adapting to large-scale data processing applications.

[0054] Furthermore, step 201 involves constructing a knowledge graph of the spatiotemporal derivation relationship of place names based on the acquired open-source geographic database, specifically including:

[0055] Based on the acquired open-source geodatabase, a knowledge graph of place name spatiotemporal derivation relationships is constructed using both top-down and bottom-up methods. Specifically: the open-source geodatabase consists of vector data (such as points, lines, and polygons) from open-source geospatial databases such as Open Street Map (OSM) and GeoNames; the top-down method defines the entities, attributes, and relationships in the knowledge graph, as shown in Table 1; the bottom-up method extracts entities, attributes, and relationships from the open-source geodatabase according to its data type, cleans and merges these entities, attributes, and relationships to eliminate ambiguity, and finally populates the knowledge graph (i.e., the graph database); the extraction of entities, attributes, and relationships from the open-source geodatabase according to its data type specifically includes:

[0056] (1) When the open-source geographic database is structured data, use Geographic Information System (GIS) tools to automatically extract entities and spatial relationships between entities (such as inclusion, adjacency, and intersection). For example, calculate whether two polygons are adjacent or whether a point is located in a face. Use text similarity algorithms, such as those based on edit distance, Jaccard similarity, and word vector cosine similarity, to calculate the semantic similarity between entities and discover homologous or abbreviation derivation relationships.

[0057] A nine-intersection model based on dimension expansion is adopted to construct a relation matrix by analyzing the intersection dimensions (DIM) of the interior (Interior, I), boundary (B), and exterior (Exterior, E) of geographic entities a and b, so as to extract the spatial topological relationships of geographic entities.

[0058] ;

[0059] In the formula: a and b represent two geographic entities; I, B, and E represent the interior, boundary, and exterior of the geographic entities, respectively; DIM represents the dimension. Considering the characteristics of spatiotemporal derivation relationships, this application mainly extracts seven types of topological relationships: containment, spanning, equal to, overlapping, touching, contained, and disjoint.

[0060] For geographical entities that are topologically disjoint, the maximum observation distance between derived features and native features is statistically determined by manual annotation and used as the spatial proximity threshold.

[0061] The topological relationship and the proximity distance threshold together constitute the spatial relationship attribute set, in which the topological relationship takes precedence over the distance judgment.

[0062] Similarity extraction, specifically:

[0063] This application utilizes Word2Vec to calculate place name similarity and designs differentiated processing strategies for different components of place names (proper names, common names, and categories). The specific methods and implementation steps are as follows:

[0064] (1) Calculation of proper name similarity; a string similarity algorithm based on edit distance (such as Levenshtein distance) is adopted:

[0065] ;

[0066] in, Indicates the length of the two place name strings; Indicates will Convert to Minimum number of editing operations (insert, delete, replace) required.

[0067] (2) Common name similarity calculation; based on the conceptual hierarchy of common names in the knowledge graph, combined with semantic distance, concept depth, density, and relation type, the similarity is calculated as follows:

[0068] ;

[0069] in, This represents the shortest path length between common names A and B in the knowledge graph; and Indicates the depth of common names A and B (the root node has a depth of 1); P represents the number of direct children of a common ancestor; P represents the strength of the relationship (e.g., the "equal" relationship is stronger than the "part-whole" relationship). Indicates the weighting adjustment factor ( ).

[0070] (3) Calculation of category similarity:

[0071] ① Define the category hierarchy, and let the category set be... Each category It contains a group of words of the same category;

[0072] ② Calculate the category center vector for each category. Calculate the mean (center vector) of all its member word vectors:

[0073] ;

[0074] in, For words The pre-trained vectors.

[0075] ③ Category similarity calculation: Cosine similarity is used to measure the similarity between the center vectors of two categories.

[0076] ;

[0077] (2) When the open-source geographic database is unstructured data, natural language processing technology is used to extract entities and spatial relationships between entities; for example, statements with implicit derivation relationships such as “River A flows into River B” and “Village C is located on the southern slope of Mountain D” are transformed into spatial relationships in the knowledge graph.

[0078] Table 1. Classes (i.e., entities), their attributes, and the relationships between classes.

[0079]

[0080] A knowledge graph is a structured representation of entities and their relationships in the objective world, aiming to construct a knowledge network that can be understood by computers in a semantic way. Its core is to organize complex knowledge in the form of a graph, where nodes represent entities or concepts, and edges represent semantic relationships between entities. Knowledge graphs describe the relationships between nodes and edges in the graph using triples, as shown below:

[0081] ;

[0082] Where E represents the set of entities in the knowledge graph; R represents the set of relations in the knowledge graph; and S represents the set of triples.

[0083] Knowledge graph construction methods include bottom-up, top-down, and a combination of both, as detailed below:

[0084] Bottom-up approach: This method uses data mining and natural language processing techniques to collect and acquire instance data, adding it to the knowledge graph data layer. This connects loosely connected knowledge points from different domains and data sources, forming a structured, hierarchical knowledge network. These knowledge elements are then summarized and organized, gradually abstracting upwards into concepts, ultimately forming the schema layer. This approach focuses on directly mining and constructing knowledge from data, without pre-designing conceptual structures, making it more suitable for handling large-scale data. While the bottom-up approach offers rapid updates and advantages in processing massive amounts of data, the schema layer built solely using this method suffers from poor comprehensiveness, accuracy, and standardization.

[0085] Top-down approach: Building knowledge graphs from the top-down perspective is a method that starts at an abstract level and gradually refines and deepens the information. First, it's necessary to clearly define the scope and objectives of the knowledge graph, identifying the relevant subject area. Then, the high-level structure and model of the knowledge graph are designed, including elements such as entities, attributes, and relationships. Ontology construction is conducted by domain experts to determine core concepts and their relationships. Subsequently, rules and constraints are formulated to guide the organization and interrelationships of entities within the knowledge graph. Once the high-level structure is established, instances and detailed information are gradually added to form a complete knowledge graph. This method emphasizes defining the overall framework and rules first, then supplementing and expanding knowledge through gradual refinement. It possesses strong logic and systematicity, effectively ensuring the consistency and accuracy of the knowledge graph. While the top-down approach can better represent the hierarchy between concepts, its construction is not only more labor-intensive and time-consuming, but also lacks automatic updates. Therefore, this method is more suitable for knowledge domains with smaller datasets.

[0086] A hybrid approach combining top-down and bottom-up methods creates a comprehensive and collaborative modeling strategy. First, a top-down approach clearly defines the overall goals, scope, and high-level structure of the knowledge graph, designing clear concepts, attributes, and relationships. Then, a bottom-up approach collects and extracts low-level data, including instances and attributes. This low-level data is gradually mapped to the previously designed high-level structure using entity extraction and relation modeling techniques, enriching the knowledge graph's details and practical content. Through continuous iteration, the diversity of the low-level data is analyzed to optimize the high-level structure. The validation and maintenance phases ensure the accuracy and consistency of the knowledge graph, maintaining its adaptability to new data. This comprehensive approach fully leverages the advantages of both bottom-up and top-down methods, aiming to create a comprehensive knowledge graph that is both guided by a high-level structure and incorporates practical details to better meet the needs of complex domains. It is typically used for knowledge graph construction in most cases.

[0087] Further, in step 202, the obtained target place name text data is input into the sequence labeling model (i.e., the pre-trained BERT-CRF model) to obtain the labeling results. The labeling results are proper nouns, common nouns, locative words and quantifiers of place names. Each character in the text can be labeled with a BIO tag (such as B-PER (start of proper noun), I-PER (middle of proper noun), B-GEN (start of common noun), B-DIR (start of locative word), O (irrelevant character)).

[0088] The BERT-CRF model is a named entity recognition model that combines a pre-trained language model (Bidirectional Encoder Representations from Transformers, BERT) and a Conditional Random Field (CRF), and is widely used in natural language processing tasks. It extracts contextual features through BERT layers and optimizes sequence labeling using CRF layers, thereby improving the accuracy of entity recognition.

[0089] In this application, place name data was extracted proportionally from 23 categories, manually annotated, and a dataset of 3000 data points was constructed. The manual annotation was performed by experts, master's students, and technicians in the field of place name research to ensure the accuracy of the dataset. Precision, recall, and F1 score were used as data metrics to evaluate the model's prediction accuracy; the formulas for calculating precision, recall, and F1 score are shown below:

[0090] ;

[0091] ;

[0092] ;

[0093] In the formula: TP (True Positive) means that all components (proper nouns, common nouns, locative words, quantifiers, etc.) of a place name are correctly labeled by the model; FP (False Positive) means that in the sequence predicted by the model as a place name, at least one component is incorrectly labeled; FN (False Negative) means that a place name component that actually exists but was not predicted by the model. For accuracy; Precision is the proportion of samples that the model predicts as positive, but which are actually positive. Recall measures how many samples that are actually positive were correctly predicted by the model. The F1 score takes both precision and recall into account and provides a more comprehensive evaluation of the model's performance in the form of a harmonic mean.

[0094] Furthermore, in step 203, generic segmentation is performed on the candidate derived place names to obtain potential base place names, specifically including:

[0095] Candidate derived place names are segmented into generic terms based on predefined derivation patterns to obtain potential base place names. Derivation patterns include "locative word + proper name + derived generic term", "derived generic term + proper name + locative word", "numeral + proper name + derived generic term", "derived generic term + proper name + numeral", "proper name + original generic term + derived generic term", and "derived generic term + original generic term + proper name". The original generic term refers to the generic term in the word-forming components of the original place name (also called the original place name), while the derived generic term refers to the generic term of the derived place name derived from the original place name.

[0096] Further, in step 203, based on the knowledge graph of place name spatiotemporal derivation relationships, it is determined whether the potential basic place name meets the preset association conditions; if yes, the potential basic place name is excluded; if no, the potential basic place name is identified as an unregistered place name; an unregistered place name means that the potential basic place name does not exist in the open-source geographic database; wherein, the preset association conditions specifically include: the correlation degree between the geographic entity type of the potential basic place name and the geographic entity type of the corresponding original place name in the knowledge graph of place name spatiotemporal derivation relationships is greater than the first threshold; the generic similarity between the potential basic place name and the original place name is greater than the second threshold; the proper name similarity between the potential basic place name and the original place name is greater than the third threshold; the geographic entity type similarity between the potential basic place name and the original place name is greater than the fourth threshold.

[0097] The knowledge graph is used to query the geographic entity type association, semantic relationship, common name similarity, proper name similarity, and category similarity (i.e., geographic entity type similarity) to check whether the categories of potential base place names and original place names meet the strong association relationship defined in the knowledge graph. If the category association is weak and does not reach the set thresholds for common name similarity, proper name similarity, and category similarity, the candidate pair is excluded.

[0098] Furthermore, in step 204, the geographic entity type of the unregistered place name is determined based on the generic name of the unregistered place name, and the spatial representation and spatial coordinates of the corresponding original place name are determined by using predefined mapping rules based on the geographic entity type of the unregistered place name and the knowledge graph of the spatiotemporal derivation relationship of place names (e.g., "door" may be mapped to "point", "square" may be mapped to "area"), thereby determining its approximate spatial representation (point, line, area).

[0099] Furthermore, in step 205, based on the spatial representation and spatial coordinates of the original place names, a geometric algorithm is used to determine the spatial location of unregistered place names, specifically including:

[0100] Orientation reasoning: Based on the spatial representation of the original place name, the minimum bounding rectangle of the original place name is calculated on the boundary corresponding to the directional words of the original place name, and the minimum bounding rectangle is determined as the first prediction position; that is, if the original place name contains directional words (such as east, south, west, north), then according to the geometry of the original place name entity (i.e. the spatial representation of the original place name), a search is performed on the outside or boundary of its corresponding orientation, and the geometric calculation of the minimum bounding rectangle (MBR) can be used to determine the approximate direction area.

[0101] Distance buffer inference: Based on the spatial coordinates of the original place name, a buffer with a fixed radius is determined, and the buffer with the fixed radius is determined as the second predicted location. This method assumes that the candidate derived place name is located near the original place name if the original place name entity does not have a specific directional word. A buffer can be created for the original place name, and the area within the buffer is used as the possible location range of the unregistered place name.

[0102] To further pinpoint the spatial location of unregistered place names, a combined reasoning method of directional reasoning and distance buffer reasoning can be employed (e.g., searching within a 50-meter buffer zone on the western boundary of the base place name). This involves determining the spatial location of the unregistered place name based on both the first and second predicted locations. An example of an unregistered place name list with spatial information is provided. Figure 4 As shown.

[0103] Furthermore, the method for discovering unregistered place names and inferring spatial location also includes:

[0104] The discovered unregistered place names and their spatial locations are compared with acquired multi-source verification data. The accuracy of discovery and the plausibility of spatial location inference are calculated through manual evaluation, simulation verification, or multi-source data comparison. Multi-source verification data includes high-precision remote sensing imagery (such as Google Earth and Bing Mapsimagery) and local knowledge obtained through crowdsourcing or expert judgment. The formulas for calculating the accuracy of discovery and the plausibility of spatial location inference are shown below:

[0105] ;

[0106] ;

[0107] This application also provides an application scenario in which the aforementioned method for discovering and spatially reasoning about unregistered place names is applied. Specifically, the method for discovering and spatially reasoning about unregistered place names provided in this embodiment can be applied in a scenario involving the discovery and spatial reasoning of unregistered place names. This scenario includes a stage for discovering unregistered place names and a stage for spatially reasoning about unregistered place names. The stage for discovering unregistered place names is used to determine unregistered place names based on an open-source geographic database and target place name text data. The stage for spatially reasoning about unregistered place names is used to determine the spatial location of unregistered place names based on their common names and a knowledge graph of place name spatiotemporal derivation relationships. The method for discovering and spatially reasoning about unregistered place names provided in this embodiment belongs to both the stage for discovering unregistered place names and the stage for spatially reasoning about unregistered place names.

[0108] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 5 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores and processes data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a method for discovering unregistered place names and spatial location reasoning.

[0109] Those skilled in the art will understand that Figure 5 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0110] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0111] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0112] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0113] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0114] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0115] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0116] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for discovering unregistered place names and inferring their spatial location, characterized in that, The method for discovering unregistered place names and inferring spatial locations includes: Based on the acquired open-source geographic database, a knowledge graph of place name spatiotemporal derivation relationships is constructed. The open-source geographic database includes multiple original place name data, which at least include place names, geographic entity types, and spatial coordinates. The knowledge graph of place name spatiotemporal derivation relationships includes entities, attributes, and relationships. The entities are place names, the attributes are proper names, common names, and spatial coordinates, and the relationships are spatial relationships and semantic similarity. The acquired target place name text data is input into the sequence labeling model to obtain the labeling results. Based on the labeling results, candidate derived place names are extracted. The sequence labeling model is a combination of a trained bidirectional pre-trained language model and a conditional random field. The labeling results are the labeling of proper nouns, generic nouns and directional words in the place names. The candidate derived place names are composite place names that include multiple generic noun combinations or combinations of generic nouns and directional words. The candidate derived place names are segmented into common names to obtain potential basic place names. Based on the place name spatiotemporal derivation relationship knowledge graph, it is determined whether the potential basic place names meet the preset association conditions. If yes, the potential basic place names are excluded. If no, the potential basic place names are determined to be unregistered place names. Unregistered place names are potential basic place names that do not exist in the open source geographic database. Based on the generic name of the unregistered place name, determine the geographic entity type of the unregistered place name, and based on the geographic entity type of the unregistered place name and the place name spatiotemporal derivation relationship knowledge graph, use predefined mapping rules to determine the spatial representation and spatial coordinates of the corresponding original place name; Based on the spatial representation and spatial coordinates of the original place names, a geometric algorithm is used to determine the spatial location of unregistered place names.

2. The method for discovering unregistered place names and inferring spatial location according to claim 1, characterized in that, Based on the acquired open-source geographic database, a knowledge graph of spatiotemporal derivation relationships of place names is constructed, specifically including: Based on the acquired open-source geographic database, a knowledge graph of place name spatiotemporal derivation relationships is constructed using top-down and bottom-up methods. The top-down method defines the entities, attributes, and relationships in the knowledge graph. The bottom-up method extracts entities, attributes, and relationships from the open-source geographic database according to its data type, and then cleans and merges these entities, attributes, and relationships before filling them into the knowledge graph.

3. The method for discovering unregistered place names and inferring spatial location according to claim 1, characterized in that, Based on the data types of the open-source geodatabase, entity, attribute, and relationship extraction is performed, specifically including: When the open-source geodatabase is structured data, spatial database or geographic information system tools are used to automatically extract entities and spatial relationships between entities, and text similarity algorithms are used to calculate the semantic similarity between entities. When the open-source geodatabase contains unstructured data, natural language processing techniques are used to extract entities and the spatial relationships between them.

4. The method for discovering unregistered place names and inferring spatial location according to claim 1, characterized in that, The candidate derived place names are segmented into generic names to obtain potential base place names, specifically including: Candidate derived place names are segmented into generic terms according to predefined derivation patterns to obtain potential base place names; the derivation patterns include "locative term + proper name + derived generic term", "derived generic term + proper name + locative term", "numeral + proper name + derived generic term", "derived generic term + proper name + numeral", "proper name + original generic term + derived generic term" and "derived generic term + original generic term + proper name".

5. The method for discovering unregistered place names and inferring spatial location according to claim 1, characterized in that, The preset association conditions specifically include: The correlation between the geographic entity type of the potential base place name and the geographic entity type of the corresponding original place name in the place name spatiotemporal derivation relationship knowledge graph is greater than the first threshold. The common name similarity between the potential base place name and the original place name is greater than the second threshold; The proper name similarity between the potential base place name and the original place name is greater than the third threshold; The similarity of geographic entity types between the potential base place name and the original place name is greater than the fourth threshold.

6. The method for discovering unregistered place names and inferring spatial location according to claim 1, characterized in that, Based on the spatial representation and spatial coordinates of the original place names, a geometric algorithm is used to determine the spatial location of unregistered place names, specifically including: Based on the spatial representation of the original place name, the minimum bounding rectangle of the original place name is calculated on the boundary corresponding to the directional word of the original place name, and the minimum bounding rectangle is determined as the first predicted position. Based on the spatial coordinates of the original place name, a buffer zone with a fixed radius is determined, and the buffer zone with the fixed radius is determined as the second predicted location; Based on the first predicted location and the second predicted location, the spatial location of the unregistered place name is determined.

7. The method for discovering unregistered place names and inferring spatial location according to claim 1, characterized in that, The method for discovering unregistered place names and inferring spatial location also includes: The discovered unregistered place names and their spatial locations are compared with the acquired multi-source verification data. The accuracy rate of unregistered place name discovery and the reasonableness rate of spatial location inference are calculated through manual evaluation, simulation verification, or multi-source data comparison.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the method for discovering unregistered place names and inferring spatial locations as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for discovering unregistered place names and inferring spatial locations as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for discovering unregistered place names and inferring spatial locations as described in any one of claims 1-7.

Images