Spatial fuzzy position determination method and device based on place name space-time derivation relation network

Abstract

The invention discloses a spatial fuzzy position determination method and device based on a place name space-time derivation relation network, and relates to the field of data information processing. The method comprises the following steps: acquiring to-be-reasoned ground feature information; based on a place name space-time derivation relation network, determining a space similarity set by adopting semantic similarity according to the to-be-reasoned ground feature information; the place name space-time derivation relation network is a knowledge graph database which is determined by adopting a place name space-time derivation relation identification method and is used for representing the place name space-time derivation relation; the place name space-time derivation relationship is determined according to a process of naming a new place name according to a space-time relationship of a place feature referred by a known place name; determining space fuzzy position information according to the space similarity set by adopting a place name space-time derivation relationship; the space fuzzy position information comprises orientation and distance; the spatial fuzzy position information is used for providing a basis for searching geographic information. The invention aims to fully apply the place name space-time derivation relationship to realize the reasoning determination of the space fuzzy position.

Description

Technical Field

[0001] The present application relates to the field of data information processing, and in particular to a method and device for determining spatial fuzzy positions based on a spatiotemporal derivation relationship network of place names. Background Art

[0002] A place name is the name of a natural or human geographic entity at a specific spatial location. Simply put, place names originate from the human conception and naming of geographic features, entities, or places. Place names play a vital role in geographic information systems (GIS). They not only serve as a core reference for spatial data positioning but also carry rich semantic information. Place names are a representative and relatively special type of geographic data in GIS. They provide an intuitive way to identify and access specific geographic locations, thereby enhancing the retrieval, analysis, and visualization capabilities of GIS data.

[0003] The spatiotemporal derivation relationship of place names refers to the process of naming newly discovered place names, where new place names are generated through derivation based on the place name's attributes. This derivational relationship between existing and new place names is known as the "spatiotemporal derivation relationship of place names." It not only reflects the textual similarity between two place names but also links the locations of geographic entities, indicating the spatial proximity between the two place names. This plays an important role in semantic expression and geographic information retrieval. Although the spatiotemporal derivation relationship of place names has potential application value in the expression and retrieval of geographic information, existing research has insufficiently explored this spatiotemporal derivation relationship and failed to fully utilize this relationship to enhance the relevance between place names, thereby improving the expression of place name semantics and information retrieval. Summary of the Invention

[0004] The purpose of this application is to provide a method and device for determining spatial fuzzy positions based on a spatiotemporal derivation relationship network of place names, which can fully utilize the spatiotemporal derivation relationship of place names to realize the reasoning and determination of spatial fuzzy positions.

[0005] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a method for determining a spatially fuzzy position based on a spatiotemporal derivation relationship network of place names, comprising: Obtain information about the object to be inferred; Based on the spatiotemporal derivation relationship network of place names, a spatially similar set is determined based on the information of the feature to be inferred using semantic similarity; the spatiotemporal derivation relationship network of place names is determined using a method for identifying spatiotemporal derivation relationships of place names, and is used to represent a knowledge graph library of spatiotemporal derivation relationships of place names; the spatiotemporal derivation relationships of place names are determined by the process of naming new place names based on the spatiotemporal relationships of the features referred to by known place names; The spatiotemporal derivation relationship of place names is adopted to determine spatial fuzzy position information according to the spatial similarity set; the spatial fuzzy position information includes: direction and distance; the spatial fuzzy position information is used to provide a basis for searching geographic information.

[0006] In one embodiment, the method for identifying the spatiotemporal derivation relationship of place names specifically includes: a text feature recognition method, a spatial feature recognition method, and a temporal feature recognition method.

[0007] In one embodiment, the text feature recognition method uses a sequence comparison method to calculate text similarity; the expression corresponding to the text similarity is: ; in, is the text similarity; is the longest common subsequence of two place names; It is a native place name; It is a derived place name; Calculates the length of the sequence.

[0008] In one embodiment, the spatial feature recognition method uses a spatial proximity relationship classification model to recognize spatial features and obtain recognition results; the recognition results include: spatial proximity relationships and non-spatial proximity relationships; The method for determining the spatial proximity relationship classification model specifically includes: Determine the topological relationship based on the nine-intersection model with dimensional expansion; Performing feature screening based on the topological relationship to obtain screening features; the screening features include: spatial distances between features and categories of primary features and derived features; Constructing a data set based on the screening features and the corresponding recognition results; Dividing the data set into a training set and a test set; The K-fold cross-validation method is used to train and optimize the data indicator parameters of the classification model according to the training set to obtain a trained classification model; the data indicator parameters include: precision, recall rate and F1 score; the classification model adopts a CART decision tree; Testing the trained classification model according to the test set, and performing generalization performance evaluation based on the confusion matrix to obtain a tested classification model; The tested classification model is used as the spatial proximity relationship classification model.

[0009] In one embodiment, the time feature recognition method is to convert time into a timestamp, compare the sizes of any two timestamps, and determine the time feature recognition result based on the comparison result; the time feature recognition result is used to determine whether the generation time of the original place name is earlier than the generation time of the derived place name.

[0010] In one embodiment, the spatiotemporal derivation relationship of place names is adopted to determine spatially ambiguous location information according to the spatial similarity set, specifically including: Determining the orientation of the ground object information to be inferred based on the directional relationship between the ground objects in the spatially similar set; Based on the spatial constraint distance of the native feature, the distance between the feature information to be inferred and the native feature is determined.

[0011] In one embodiment, the spatiotemporal derivation relationship network of place names is stored in a place name database.

[0012] In one embodiment, it further includes: The spatially ambiguous position information is verified to determine accuracy.

[0013] In one embodiment, the accuracy is determined using a success rate function; the mathematical expression of the success rate function is: ; in, The correct amount for the location; Quantities that are located for spatial and temporal derivations; For the success rate.

[0014] In a second aspect, the present application provides a spatial fuzzy position determination device based on a spatiotemporal derivation relationship network of place names, comprising: Information acquisition module, used to obtain information about the object to be inferred; A spatially similar set determination module is configured to determine a spatially similar set based on the spatial-temporal derivation relationship network of place names and the information of the ground feature to be inferred using semantic similarity; the spatial-temporal derivation relationship network of place names is determined using a method for identifying spatial-temporal derivation relationships of place names and is used to represent a knowledge graph library of spatial-temporal derivation relationships of place names; the spatial-temporal derivation relationships of place names are determined by the process of naming new place names based on the spatial-temporal relationships of the ground features referred to by known place names; The spatial fuzzy location information determination module is used to adopt the spatiotemporal derivation relationship of the place name and determine the spatial fuzzy location information according to the spatial similarity set; the spatial fuzzy location information includes: direction and distance; the spatial fuzzy location information is used to provide a basis for retrieving geographic information.

[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a method and device for determining spatial fuzzy positions based on a spatiotemporal derivation relationship network of place names. Semantic similarity is used to obtain a spatially similar set of features to be inferred from the spatiotemporal derivation relationship network of place names. The spatiotemporal derivation relationship network of place names is determined using a method for identifying spatiotemporal derivation relationships of place names, and is used to represent a knowledge graph library of spatiotemporal derivation relationships of place names. The spatiotemporal derivation relationships of place names are determined by the process of naming new place names based on the spatiotemporal relationships of features referred to by known place names. The spatiotemporal derivation relationships are then used to perform spatial fuzzy position inference and determine spatial fuzzy position information to provide a basis for searching for geographic information. Therefore, the present application can fully utilize the spatiotemporal derivation relationships of place names to realize the reasoning and determination of spatial fuzzy positions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 A flowchart of a method for determining spatial fuzzy locations based on a spatiotemporal derivation relationship network of place names; Figure 2 Flowchart for identifying spatiotemporal derivation relationships of place names; Figure 3 Construct a process diagram for the knowledge graph of spatiotemporal derivation relationships of place names; Figure 4 This is a schematic diagram of the spatiotemporal derivation relationship network; Figure 5 This is a schematic diagram of spatial fuzzy position reasoning; Figure 6 This is a schematic diagram of the place name reasoning process; Figure 7 This is a structural diagram of a spatial fuzzy position determination device based on a spatiotemporal derivation relationship network of place names. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0019] The core goal of this application is to address the underappreciated application of spatiotemporal derivational relationships in geographic information science. Therefore, a specific method for identifying and expressing spatiotemporal derivational relationships in place names is proposed. Based on the construction of a knowledge graph of spatiotemporal derivational relationships in place names, this network is used to conduct research on reasoning about spatially ambiguous locations.

[0020] The significance of this application lies in that research on the spatiotemporal derivation of place names not only enriches the theoretical framework for this approach but also provides new perspectives and tools for practical applications in geographic information science. By incorporating spatiotemporal derivation into inference systems, this research can support areas such as semantic reasoning about place names.

[0021] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0022] In an exemplary embodiment, Figure 1 As shown, a method for determining spatial fuzzy positions based on a spatiotemporal derivation relationship network of place names is provided, including: Step 100: Obtain information about the object to be inferred.

[0023] Step 200: Based on the spatiotemporal derivation relationship network of place names, semantic similarity is used to determine spatially similar sets based on the feature information to be inferred. The spatiotemporal derivation relationship network of place names is determined using a method for identifying spatiotemporal derivation relationships of place names, and is used to represent the knowledge graph of spatiotemporal derivation relationships of place names. Spatiotemporal derivation relationships of place names are determined by the process of naming new place names based on the spatiotemporal relationships of the features referred to by known place names.

[0024] The spatiotemporal derivation relationship network of place names is stored in the place name database.

[0025] In one embodiment, the method for identifying the spatiotemporal derivation relationship of place names specifically includes: a text feature recognition method, a spatial feature recognition method, and a temporal feature recognition method.

[0026] The text feature recognition method uses the sequence comparison method to calculate text similarity; the expression corresponding to text similarity is: .

[0027] in, is the text similarity; is the longest common subsequence of two place names; It is a native place name; It is a derived place name; Calculates the length of the sequence.

[0028] The spatial feature recognition method uses a spatial proximity relationship classification model to recognize spatial features and obtain recognition results; the recognition results include: spatial proximity relationships and non-spatial proximity relationships.

[0029] The method for determining the spatial proximity relationship classification model specifically includes: The topological relationship is determined based on the nine-intersection model with dimensional expansion.

[0030] Feature screening is performed based on topological relationships to obtain screening features; the screening features include: spatial distances between features and categories of primary features and derived features.

[0031] Construct a data set based on the screening features and the corresponding recognition results; divide the data set into a training set and a test set.

[0032] The K-fold cross-validation method is used to train and optimize the data indicator parameters of the classification model based on the training set to obtain the trained classification model; the data indicator parameters include: precision, recall rate and F1 score; the classification model adopts the CART decision tree.

[0033] The trained classification model is tested according to the test set, and the generalization performance is evaluated based on the confusion matrix to obtain the tested classification model; the tested classification model is used as the spatial proximity relationship classification model.

[0034] The time feature recognition method is to convert time into a timestamp, compare the sizes of any two timestamps, and determine the time feature recognition result based on the comparison result; the time feature recognition result is used to determine whether the generation time of the original place name is earlier than the generation time of the derived place name.

[0035] Step 300: Using the spatiotemporal derivation relationship of place names, determine spatially fuzzy location information based on spatial similarity sets. Spatially fuzzy location information includes: direction and distance; spatially fuzzy location information is used to provide a basis for searching geographic information.

[0036] In one embodiment, the temporal and spatial derivation relationship of place names is used to determine spatially ambiguous location information based on spatial similarity sets, specifically including: According to the directional relationship between the features in the spatial similarity set, the orientation of the feature information to be inferred is determined; based on the spatial constraint distance of the native feature, the distance between the feature information to be inferred and the native feature is determined.

[0037] As an optional implementation, the method for determining spatially fuzzy positions based on a spatiotemporal derivation relationship network of place names mentioned in the present application further includes: verifying the spatially fuzzy position information to determine the accuracy.

[0038] The accuracy is determined using a success rate function; the mathematical expression of the success rate function is: .

[0039] in, The correct amount for the location; Quantities that are located for spatial and temporal derivations; For the success rate.

[0040] This application retrieves place-name data from a place-name database that has a spatiotemporal derivation relationship with the feature to be inferred. It then uses semantic similarity to obtain a spatially similar set of the feature to be inferred from the spatiotemporal derivation relationship network. This approach leverages spatiotemporal derivation relationships to perform spatial fuzzy position inference (determination). This method can effectively increase the utilization of unregistered place names and provides an innovative solution for place-name data management.

[0041] The research focus of this application is to use the semantic and spatial relationships contained in the constructed spatiotemporal derivation relationship network of place names to perform spatial fuzzy position reasoning on the missing geographic entities in the place name database, and to study and discuss the influencing factors that need to be considered in the reasoning process.

[0042] This application method includes the following three steps: The first step is to obtain place name data that have a spatiotemporal derivation relationship with the feature to be inferred from the place name database.

[0043] Through the recognition method of the spatiotemporal derivation relationship of place names, including text feature recognition method, spatial feature recognition method and time feature recognition method, the place name data of the spatiotemporal derivation relationship is determined and entered into the place name database.

[0044] Specifically, based on the definition of the spatiotemporal derivation relationship of place names, the characteristics and identification methods of the spatiotemporal derivation relationship of place names in terms of text, space and time are determined.

[0045] Spatiotemporal derivation of place names: The process of creating new place names based on their spatiotemporal relationship to the features designated by existing place names is called "place name derivation." Existing place names are called "primitive place names," while the newly created place names are called "derived place names." The relationship between the derived and derived place names is known as the "spatiotemporal derivation of place names."

[0046] In terms of text feature recognition, text similarity is used to identify whether place names are similar; in terms of spatial feature recognition, topological relationships and decision trees are used to identify the spatial proximity relationships between land features; in terms of temporal feature recognition, timestamps are used to identify the chronological order of the generation of two place names.

[0047] The method for identifying the spatiotemporal derivation relationship of place names can more accurately understand the semantic association and spatial proximity between place names, and enrich the semantic information of place names so that they not only reflect the geographical location, but also the historical, cultural and geographical connections between place names.

[0048] The method for identifying the spatiotemporal derivative relationship of place names first uses text similarity to determine whether two place names are similar in naming; then calculates the topological relationship between geographic entities, and uses the trained decision tree model to determine the spatial proximity relationship of land features with separate topological relationships; finally, uses timestamps to determine the order of generation time of the original place name and the derived place name. Figure 2 Flowchart for identifying spatiotemporal derivation relationships of place names.

[0049] 1. Text feature recognition method.

[0050] Based on the text features of the spatiotemporal derivation relationship of place names, it can be seen that the original place names and the derived place names have similarities in their names. Therefore, this application uses text similarity to calculate the similarity between the two place names and performs text feature recognition based on the similarity scores of the place names.

[0051] Text similarity is a measure of the semantic or meaning-related closeness between two texts. Smaller values ​​indicate greater semantic differences between the texts, meaning lower semantic similarity. Conversely, larger values ​​indicate greater semantic similarity between the texts. Given that place names are expressed as strings, this application uses a sequence comparison-based method to calculate the similarity between two place names. This method calculates the similarity between the two strings by identifying the longest common subsequence (LCS) of the two strings. This method is fast, takes sequence length into account, and normalizes the results, allowing for direct comparison.

[0052] .

[0053] The similarity score (result) is normalized to a range of 0 to 1. A score of 0 indicates that the two strings have no common subsequences, meaning that the two place names are ordinary place names. A score of 1 indicates that the two strings are identical, meaning that the two place names have a complete spatiotemporal derivational relationship. Other scores indicate that the two strings are only partially identical, meaning that the two place names have a partial spatiotemporal derivational relationship.

[0054] 2. Spatial feature recognition method.

[0055] Spatial feature identification primarily involves determining whether the geographical features referred to by place names with a derived-derived relationship are spatially proximal. First, the topological relationships between geographic entities are calculated. Then, spatially constrained distances are used to determine whether geographical features with disjoint topological relationships are spatially proximal. This application considers this process a classification process for spatial proximity, using feature type and distance as characteristic factors for classification using a decision tree to identify spatial features.

[0056] Topological relationship: Using the nine-intersection model based on dimension expansion proposed by ClementiniE et al., by analyzing geographic entities a and geographic entities b The relationship matrix is ​​constructed by the intersection dimension (DIM) of the interior (I), boundary (B), and exterior (E) to extract the topological relationship of geographic entities. .

[0057] .

[0058] I 、 B 、 E Represent the interior, boundary and exterior of a geographic entity respectively. DIM Represents dimension.

[0059] Spatial Proximity Classification Model: For topologically separated place names, the final step of determining spatial proximity is based on the spatially constrained distance between the two features. The threshold for this distance depends on the feature category. Therefore, the spatial proximity between the two place names is determined based on the categories of the original and derived features and the spatial distance between them. This process can be considered a text classification process, categorizing the relationships between features into spatial proximity and non-spatial proximity based on the two characteristic criteria of feature category and distance.

[0060] Text classification typically requires attention to issues such as the training sample set used to train the classifier, feature selection, classification algorithm, and performance evaluation metrics of the classification system. These are described below.

[0061] (1) Feature screening

[0062] Feature screening, also known as feature subset selection (FSS), involves selecting the most influential features from the original features for the research classification. This increases the versatility and robustness of the model while also strengthening the connection between features and feature sets, thereby improving the accuracy and interpretability of the model. Based on the spatial characteristics of the spatiotemporal derivation of place names and the above analysis, the features used in this application are the spatial distance between features and the categories of native and derived features.

[0063] Spatial distance: To accurately quantify the spatial relationship between geographic entities, this application uses numerical values ​​to represent the quantitative distance between entities. Considering that the geographic entities referred to by place names have three types of elements: points, lines, and surfaces, when calculating distance, we first extract the center points of the surface elements and line elements, and then calculate the distance between the center points and the point elements. The distance formula between geographic entities is: .

[0064] in, and Respectively represent the spatial coordinate values ​​of the geographical entities referred to by the two place names, Represents the distance between two entities.

[0065] Category: Determine the category of a place name based on its attribute information.

[0066] (2) Classification algorithm

[0067] Text classification algorithms mainly include association rule-based classification algorithms, logistic regression, support vector machine (SVM), Bayesian classification, decision tree and neural network algorithms.

[0068] Classification algorithms based on association rules are called association classification. Classification based on association rules (CBA) constructs a classifier in two steps. The first step uses a standard association rule mining algorithm to mine relevant association rules, namely, Classification Association Rules (CARs) whose right-hand side is a class attribute value. The second step is to select high-priority rules from the discovered CARs to cover the training set. If multiple association rules have the same left-hand side but different right-hand sides, the rule with the highest confidence is selected as the possible rule. The CBA algorithm primarily constructs a classifier by discovering association rules in the training set. The classic algorithm for discovering association rules is the Apriori algorithm. The advantage of this algorithm is its high classification accuracy. Its main disadvantage is that when the size of the potentially frequent two-item set is large, the algorithm is constrained by hardware memory, resulting in excessive system I / O load and low efficiency. Secondly, it is time-consuming and expensive.

[0069] Logistic regression is a classification algorithm primarily used for binary classification problems. The output variable of logistic regression has only two possible values, representing either one or the other of the two categories. In classification tasks, logistic regression uses the logistic function (also known as the sigmoid function) to construct a prediction function. This function maps a linear combination of inputs to a probability value between 0 and 1, representing the probability that a sample belongs to a certain class. In binary classification problems, the goal is to ensure that the predicted probabilities are as close as possible to the actual observed values. The loss function measures the difference between the model's predictions and the actual observations. Its form can be derived from the log-likelihood function. Using gradient descent, the model parameters are iteratively adjusted to continuously reduce the loss function, thereby finding the parameter configuration that optimizes model performance. However, logistic regression struggles with imbalanced data and suffers from low accuracy.

[0070] Support Vector Machine (SVM) is a new development in statistical learning theory. Compared with traditional statistics, SVM is not based on the traditional principle of empirical risk minimization, but is built on the principle of structural risk minimization, and has developed into a new type of structured learning method. It can effectively solve the problem of constructing high-dimensional models with a limited number of samples, and the constructed models have good predictive performance. As a powerful machine learning algorithm, SVM has efficient nonlinear classification capabilities, the ability to process high-dimensional data, and strong interpretability and robustness. However, SVM is slow when processing large-scale data sets, is sensitive to parameter selection, cannot directly handle multi-class problems, and is relatively sensitive to missing data.

[0071] Bayesian classification primarily uses Bayes' theorem to predict the probability of a sample of unknown class belonging to various categories, selecting the most likely category as the final category for the sample. In many cases, the basic Bayesian (NaiveBayes) classification algorithm is comparable to decision tree and neural network classification algorithms. This algorithm can be applied to large databases and features simplicity, high classification accuracy, and high speed. The Bayesian classifier assumes that all features are independent, meaning that there are no correlations between features. In practical applications, this assumption does not always hold true, as dependencies between features can affect classification accuracy. When this assumption is not met, classification accuracy decreases. The accuracy of the Bayesian classifier depends on the training data. In particular, when the feature dimension is high or the data is noisy, sufficient training samples are required to obtain reliable probability estimates.

[0072] The Decision Tree (DT) model is a supervised learning algorithm that builds a tree model through a top-down recursive construction process. The goal of this model is to learn decision rules through training data to predict the label value of the target variable. The CART decision tree classification method is convenient, easy to understand, and efficient, and is one of the mainstream classification methods. This method uses the Gini coefficient method to select features at each node of the decision tree and builds the tree through a recursive method. Starting from the root node, the Gini coefficient values ​​of all features are calculated, and the minimum value is selected as the feature of the node. The Gini coefficient of the feature value is then calculated, and the minimum value is selected as the partition node of the feature. One of the advantages of decision trees is that they are easy to interpret and visualize because each node corresponds to a simple decision made by a certain feature. The classification method of the decision tree model is concise and direct, easy to understand and explain. As a very fast learning and prediction algorithm, it can provide high text classification efficiency.

[0073] The advantages of neural network methods lie in their ability to approximate any function with arbitrary precision, and classification itself is the process of determining a discriminant function. Neural network methods are inherently nonlinear models, making them adaptable to complex real-world data relationships. Neural networks possess a strong learning ability, automatically adjusting their internal weights through learning to gradually adapt to different tasks and environments. This enables them to demonstrate good generalization when faced with new tasks. However, neural networks are often considered "black box" models, and their outputs are often difficult to interpret. This can lead to concerns about their results. Furthermore, neural networks typically require a large number of parameters to train. These parameters, including weights and biases, can be extremely numerous, increasing their complexity and difficulty in training.

[0074] Compared to other classification models like neural networks or Bayesian classification, decision trees are simple to understand and generate understandable rules, making them easier to understand and accept. Furthermore, decision tree classification generally does not require manual parameter setting, is fast to build, and can handle both continuous and discrete values. Furthermore, decision trees use the information principle to analyze the information content of the attributes of a large number of samples, calculating the information content of each attribute and clearly showing which features are important for classification. Based on the characteristics of the research data, this application selected the CART decision tree (Classification and Regression Tree) as the spatial proximity classifier.

[0075] (3) Model training and evaluation.

[0076] The constructed data set is divided into a training set and a test set, and then the training set data is used to train the decision tree model, and parameters are set to control the depth of the tree to prevent overfitting. The trained model is used to predict the test set to achieve the classification of the spatial proximity relationship between objects. For the trained decision tree classification model, three parameters, precision, recall and F1 score, are usually used as data indicator parameters to evaluate the prediction accuracy of the model. However, in the case of data imbalance or significantly different error costs, simple prediction accuracy may not meet all requirements. Therefore, this application combines the generalization performance evaluation technology of the confusion matrix to comprehensively evaluate the learning ability of the model.

[0077] Accuracy , recall rate The calculation formula of F1 score is as follows: .

[0078] TP, or True Positive, represents the number of positive samples predicted as true; FP, or False Positive, represents the number of negative samples predicted as true; and FN, or False Negative, represents the number of positive samples predicted as false. In terms of metrics, precision measures the accuracy of positive predictions; recall measures the proportion of samples that are actually positive that are correctly predicted by the model; and the F1 score considers both precision and recall, providing a comprehensive assessment of model performance in the form of a harmonic mean. A larger F1 value indicates better classification results.

[0079] The confusion matrix is ​​a tool for evaluating the performance of classification models. The diagonal elements of the confusion matrix represent the number of correct predictions made by the classification model, and the off-diagonal elements represent the number of incorrect predictions made by the classification model. Therefore, comparing the predicted values ​​with the true values ​​in the confusion matrix can reveal the distribution of incorrect predictions in different categories. In order to reduce the risk of underfitting or overfitting the model, the decision tree model also needs to be generalized. K-fold cross validation can evenly fit the data distribution, that is, multiple divisions are made between the training set and the test set, and the training set and test set obtained from each division are used to train and test the model. A balance can be established between the training and test sets, which helps to evaluate the consistency level of different randomly divided data sets to improve the accuracy of the proposed model. Therefore, this application uses K-fold cross validation for model tuning. In addition, this application also uses an independent test set to classify objects with spatial proximity to evaluate the generalization ability of the model.

[0080] For the optimized decision tree model, feature category and spatial distance are used as feature parameters to determine whether there is a spatial proximity relationship between two features.

[0081] 3. Time feature recognition.

[0082] Convert the time to a timestamp, and then directly compare the sizes of the two timestamps to determine whether the original place name was generated earlier than the derived place name.

[0083] In the second step, semantic similarity is used to obtain the spatially similar set of features to be inferred from the spatiotemporal derived relationship network.

[0084] The spatiotemporal derivation relationship network is an established knowledge graph library. The semantic similarity is used to calculate the place names to be inferred and the place names in the spatiotemporal derivation relationship network, and the place names with higher similarity scores are selected to form a similar set. Figure 3 A process for building a knowledge graph of spatiotemporal derivation of place names. Data layers are constructed based on the open geographic database and GIS platform (GeoNames) and the Open Street Map (OSM).

[0085] The similarity formula is the same as the expression corresponding to the text similarity described above.

[0086] The third step is to use the temporal and spatial derivative relationship to predict the location (determination of spatially fuzzy location information).

[0087] Check the directional relationship between spatial similarity sets, and infer the approximate direction of the feature to be judged by inferring the directional relationship between the features in the similarity sets. At the same time, the distance is determined by judging the maximum distance between the feature to be judged and the native feature based on the spatial constraint distance of the native feature; the prediction result gives the approximate direction and distance.

[0088] The fourth step is to verify the proposed spatial fuzzy position reasoning method and use the success rate to evaluate the accuracy of the spatial position reasoning results proposed in this application using the spatiotemporal derivative relationship network.

[0089] The mathematical expression of the success rate function is: ; in, The correct amount for the location; Quantities that are located for spatial and temporal derivations; For the success rate.

[0090] For example: Figure 4 As shown, the place name "Somewhere University Plot 5" does not exist in the place name database, so the geographical location of the geographical entity referred to by the place name cannot be obtained. Using the spatial fuzzy position reasoning method of the present application, first look for place name data in the place name database that has a place name spatiotemporal derivation relationship with the place name. First identify the place name with a derived relationship and put it into the database for marking. Through keyword query, find the place name data that has a place name spatiotemporal derivation relationship with the place name) and then obtain the spatial similarity set of the land object: Somewhere University, Somewhere University Plot 1, Somewhere University Plot 2, Somewhere University Plot 3, Somewhere University Plot 4; judge the directional relationship between the names in the spatial similarity set; then according to the definition of the derived relationship, it can be determined that Somewhere University Plot 5 is located within the constraint range of Somewhere University, and then according to the distribution mode of Somewhere University Plot 1, Somewhere University Plot 2, Somewhere University Plot 3, Somewhere University Plot 4 being due west, determine that Somewhere University Plot 5 is located due west of Somewhere University Plot 4.

[0091] like Figure 5 As shown, for spatial location reasoning, this application uses a method based on the existing derivative relationship in the database to infer the geographical location of place names that are not logged in the database. Location is a basic and important feature among the many feature dimensions of space. Cognitive spatial location can answer the question of "where" among the six major questions in geography, and provide a spatial reference for the answers to other geographical questions. In urban space, the main forms of location information include: coordinates, postal codes, telephone numbers, IP, place names and addresses. Among them, except for the coordinate type, other location data need to be spatially positioned before they can be converted into urban space. Among these types of data that need to be converted, place name data is a relatively standardized and very rich data in the form of natural language, which is widely included in the data of citizens' lives, government management and corporate institutions. Figure 5 The derived relationship query in Figure 4 The spatiotemporal derived relationship network in .

[0092] like Figure 6As shown, the feasibility of location reasoning is verified using a map. To perform location reasoning on a particular place name (place name 1), the first step is to search the place name database for place names that share a spatiotemporal derivation relationship with the place name. Then, a spatial similarity set for the feature is obtained. Finally, based on the derivation relationship definition, it is determined that place name 1 is within the constraint of the place name "Red Deer." By searching the map for the location of place name 1, it is found to be approximately 1,760 meters away from "Red Deer."

[0093] Evaluation of spatial fuzzy position reasoning methods.

[0094] The proposed spatial fuzzy location inference method was verified, and the success rate was used to evaluate the accuracy of the spatial location inference results using the spatiotemporal derivative relationship network proposed in this application. The specific results are shown in Table 1.

[0095] Table 1 Detailed result information

[0096] Because spatiotemporal derivation is dependent on category, this analysis randomly sampled 300 place names from a specific category from a dataset for location inference. A location threshold of 2 kilometers was set. If the location inference result was within 2 kilometers, the inference was considered correct; if it exceeded 2 kilometers, it was considered incorrect. The 300 place names contained 32 pairs of derived place names, of which 25 were correctly inferred. Therefore, the success rate of spatial location inference using spatiotemporal derivation of place names was 78%. Limitations of spatial fuzzy location inference include the need for further optimization of the location threshold setting and location orientation inference.

[0097] Spatiotemporal derivation of place names, a special type of place name relationship, can simultaneously characterize the semantic and spatiotemporal connection between two place names. Therefore, identifying derivational relationships between place names not only enriches the semantic expression of place names but also enables more accurate geographic information retrieval. The success rate of spatial location reasoning using spatiotemporal derivation of place names was 78%. This result demonstrates the high success rate of the proposed method for spatial location reasoning, providing new perspectives and tools for practical applications in geographic information science.

[0098] In an exemplary embodiment, Figure 7 As shown, a spatial fuzzy position determination device based on a spatiotemporal derivation relationship network of place names is provided, comprising: The information acquisition module is used to obtain the information of the ground object to be inferred.

[0099] The spatial similarity set determination module is used to determine the spatial similarity set based on the spatial-temporal derivation relationship network of place names and the information of the land features to be inferred by using semantic similarity; the spatial-temporal derivation relationship network of place names is determined by using the recognition method of the spatial-temporal derivation relationship of place names, and is used to represent the knowledge graph library of the spatial-temporal derivation relationship of place names; the spatial-temporal derivation relationship of place names is determined by the process of naming new place names based on the spatial-temporal relationship of the land features referred to by known place names.

[0100] The spatial fuzzy location information determination module is used to determine the spatial fuzzy location information based on the spatial similarity set by adopting the spatiotemporal derivation relationship of place names; the spatial fuzzy location information includes: direction and distance; the spatial fuzzy location information is used to provide a basis for the retrieval of geographic information.

[0101] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, 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.

[0102] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for determining spatial fuzzy position based on a spatiotemporal derivation relationship network of place names, characterized in that: include: Obtain information about the object to be inferred; Based on the spatiotemporal derivation relationship network of place names, a spatially similar set is determined according to the ground feature information to be inferred using semantic similarity; The spatiotemporal derivation relationship network of place names is determined by using a method for identifying spatiotemporal derivation relationships of place names, and is used to represent a knowledge graph library of spatiotemporal derivation relationships of place names. The spatiotemporal derivation relationships of place names are determined by the process of naming new place names based on the spatiotemporal relationships of the features referred to by known place names. Using the spatiotemporal derivation relationship of the place name, determining spatial fuzzy location information according to the spatial similarity set; The spatially fuzzy position information includes: direction and distance; The spatially fuzzy position information is used to provide a basis for retrieving geographic information.

2. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 1, characterized in that: The method for identifying the spatiotemporal derivation relationship of place names specifically includes: text feature recognition method, spatial feature recognition method and temporal feature recognition method.

3. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 2, characterized in that: The text feature recognition method uses a sequence comparison method to calculate text similarity; the expression corresponding to the text similarity is: ； in, is the text similarity; is the longest common subsequence of two place names; It is a native place name; It is a derived place name; Calculates the length of the sequence.

4. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 2 is characterized in that: The spatial feature recognition method adopts a spatial proximity relationship classification model to recognize spatial features and obtain recognition results; the recognition results include: spatial proximity relationships and non-spatial proximity relationships; The method for determining the spatial proximity relationship classification model specifically includes: Determine the topological relationship based on the nine-intersection model with dimensional expansion; Performing feature screening based on the topological relationship to obtain screening features; the screening features include: spatial distances between features and categories of primary features and derived features; Constructing a data set based on the screening features and the corresponding recognition results; Dividing the data set into a training set and a test set; The K-fold cross-validation method is used to train and optimize the data indicator parameters of the classification model according to the training set to obtain a trained classification model; the data indicator parameters include: precision, recall rate and F1 score; the classification model adopts a CART decision tree; Testing the trained classification model according to the test set, and performing generalization performance evaluation based on the confusion matrix to obtain a tested classification model; The tested classification model is used as the spatial proximity relationship classification model.

5. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 2 is characterized in that: The time feature recognition method is to convert time into a timestamp, compare the sizes of any two timestamps, and determine the time feature recognition result based on the comparison result; the time feature recognition result is used to determine whether the generation time of the original place name is earlier than the generation time of the derived place name.

6. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 1, characterized in that: Adopting the spatiotemporal derivation relationship of the place name and determining the spatial fuzzy location information according to the spatial similarity set specifically includes: Determining the orientation of the ground object information to be inferred based on the directional relationship between the ground objects in the spatially similar set; Based on the spatial constraint distance of the native feature, the distance between the feature information to be inferred and the native feature is determined.

7. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 1 is characterized in that: The place name spatiotemporal derivation relationship network is stored in a place name database.

8. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 1, characterized in that: Also includes: The spatially ambiguous position information is verified to determine accuracy.

9. The method for determining spatial fuzzy position based on the spatiotemporal derivation relationship network of place names according to claim 8, characterized in that: The accuracy is determined using a success rate function; the mathematical expression of the success rate function is: ； in, The correct amount for the location; Quantities that are located for spatial and temporal derivations; For the success rate.

10. A spatial fuzzy position determination device based on a spatiotemporal derivation relationship network of place names, characterized in that: include: Information acquisition module, used to obtain information about the object to be inferred; A spatial similarity set determination module is used to determine a spatial similarity set based on the spatial-temporal derivation relationship network of place names and the semantic similarity according to the ground feature information to be inferred; The spatiotemporal derivation relationship network of place names is determined by using a method for identifying spatiotemporal derivation relationships of place names, and is used to represent a knowledge graph library of spatiotemporal derivation relationships of place names. The spatiotemporal derivation relationships of place names are determined by the process of naming new place names based on the spatiotemporal relationships of the features referred to by known place names. a spatially fuzzy location information determination module, configured to determine spatially fuzzy location information based on the spatial similarity set using the spatiotemporal derivation relationship of the place name; The spatially fuzzy position information includes: direction and distance; The spatially fuzzy position information is used to provide a basis for retrieving geographic information.

Images