CN112256888A - Geographic knowledge acquisition method - Google Patents
Geographic knowledge acquisition method Download PDFInfo
- Publication number
- CN112256888A CN112256888A CN202011189625.9A CN202011189625A CN112256888A CN 112256888 A CN112256888 A CN 112256888A CN 202011189625 A CN202011189625 A CN 202011189625A CN 112256888 A CN112256888 A CN 112256888A
- Authority
- CN
- China
- Prior art keywords
- geographic
- knowledge
- map
- information
- text
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
- G06N5/025—Extracting rules from data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/027—Frames
Abstract
The invention discloses a geographic knowledge acquisition method, which comprises the following steps: step 1, analyzing the source of geographic knowledge, and dividing the obtained source of the geographic knowledge into two categories, namely natural language and graphic language; step 2, carrying out feature analysis and concept modeling on the acquired geographic knowledge; step 3, extracting geographic information from the acquired geographic knowledge, wherein the extraction at least comprises identification of a geographic entity, extraction of characteristic information and association between the geographic entity and the characteristic information; and 4, generating a geographical knowledge map. The invention aims to provide a geographic knowledge acquisition method, which can make up the vacancy of the existing geographic knowledge acquisition method and convert the geographic knowledge acquisition method into a knowledge map, essentially promotes the intellectualization, socialization and popularization of geographic information services, and further promotes the development of geographic information scientific theory and application.
Description
Technical Field
The invention belongs to the technical field of acquisition of knowledge maps, and particularly relates to a geographic knowledge acquisition method.
Background
Knowledge Graph (knowledgegraph) is essentially a large semantic network that describes concepts, entities and their interrelationships in the objective world. In 2012, google introduced semantic search engines based on knowledge graph, which promoted the leap-type development of internet from character search to things content. Knowledge maps have then rapidly attracted a great deal of attention in both academia and industry. At present, the knowledge graph is constructed into an effective carrier for expressing information and knowledge in the computer world in a form close to human thinking, becomes an important infrastructure for artificial intelligence application, and highlights more and more important application values in the aspects of semantic search, intelligent question answering, prediction decision making and the like.
Geographic knowledge is the result of human cognition on the spatial distribution, evolution process and interaction laws of geographic objects or phenomena. Natural languages (such as text, voice, etc.) and graphical languages (such as maps, schematics, remote sensing images, videos, etc.) are the main forms of carriers for recording and propagating geographic knowledge. The geographical knowledge map is a knowledge system for formally describing geographical concepts, entities and mutual relations thereof, can provide systematic and deep structured geographical knowledge, and has huge application potential in the aspects of geographical knowledge understanding, geoscience problem solving, space-time prediction decision making and the like.
In recent years, the geographical knowledge graph gradually becomes a research hotspot in the field of geographic information science, but is still in the stages of concept discussion and preliminary experiments, and there is no method for converting text and graphic information in practical application scenes into a geographical knowledge graph which is easier to understand by the masses.
In view of the above-mentioned shortcomings of the existing method for acquiring geographical knowledge, the present inventors have been actively researched and innovated based on the practical experience and professional knowledge that are abundant for many years in designing and manufacturing such products, and by using the theory, in order to create a method for acquiring geographical knowledge, which is more practical. After continuous research and design and repeated trial production and improvement, the invention with practical value is finally created.
Disclosure of Invention
The invention aims to provide a geographic knowledge acquisition method, which can make up the vacancy of the existing geographic knowledge acquisition method and convert the geographic knowledge acquisition method into a knowledge map, essentially promotes the intellectualization, socialization and popularization of geographic information services, and further promotes the development of geographic information scientific theory and application.
In order to achieve the purpose, the invention provides the following technical scheme:
the geographic knowledge acquisition method comprises the following steps:
and 4, generating a geographical knowledge map.
As a preferred technical solution, the natural language in the step 1 at least includes a text;
the acquisition of geographic knowledge in the text at least comprises: time information extraction, geographic entity identification, attribute information extraction, geographic entity relationship extraction and event information extraction.
As a preferred technical solution, the graphic language at least includes a map, and a convolutional neural network is adopted to extract information in the map.
As a preferred technical solution, the extracting information in the map by using a convolutional neural network includes:
s1, constructing a map information labeling sample library;
s2, automatically identifying the information in the map;
and S3, constructing a deep convolutional neural network model according to the map symbols and the text annotation reference information, and realizing geographic entity identification and extraction of characteristic information and association relation.
As a preferred technical solution, in S1, marking content and marking specifications of geographic information in a map are formulated, and map samples of different types, different contents, and different compiling forms are selected to construct the map information marking sample library.
As a preferable technical solution, in S2, a rule model of an auxiliary element, a supplementary explanation, and a graphic/image element in a map is created based on a geographic theory, and an algorithm is designed to realize automatic recognition of information in the map.
As a preferred technical solution, fusing the information in the text and the map, including:
1) carrying out concept mapping on the text and the geographic knowledge described in the map according to a unified concept classification system, and solving the problem of inconsistent granularity and grade of the geographic knowledge;
2) establishing geographic entity links of all concept levels according to the geographic entity types, the text similarity and the attribute characteristics, and solving the problems of the semantic ambiguity and the synonymy of the geographic entities in the text and the map;
3) processing the text and the relevant characteristic information of the geographic entity in the map by using a conflict detection and/or truth value discovery technology by taking the geographic entity as a unit, wherein the processing at least comprises any one of deduplication, association and combination;
4) and constructing a constraint rule set of the geographic entity relationship, and reconstructing the relevant relationship between the text and the geographic concept, example and feature in the map.
As a preferred technical solution, the geographical knowledge graph in the step 4 is based on a convolutional neural network technology, and includes:
i, constructing a geographical knowledge subgraph by referring to the composition and the relation of knowledge units in a geographical knowledge representation model according to geographical knowledge segments obtained from the text and the map;
II, performing link processing on the nodes of each group of knowledge subgraphs under the constraint of space-time characteristics;
III, judging different states and relations of the single geographic entity through model iteration and normalization processing, and outputting a state conversion process of the geographic entity according to a time change and space change sequence;
and IV, judging the incidence relation among different geographic entities and among different geographic features to generate geographic knowledge maps with different granularities and different levels.
By adopting the technical scheme, the following technical effects can be realized:
1. the invention solves the basic theoretical problem of the geographical knowledge map field and provides common key technology, in particular to a geographical representation model, a geographical knowledge acquisition and geographical knowledge map generation method based on deep learning by introducing advanced research results of related subject fields.
2. The invention starts from basic questions answered in geography, summarizes and abstracts the connotation and the extension of geographic knowledge, and provides a geographic knowledge representation model which takes space and time as a frame and takes a geographic entity as a core, thereby providing theoretical support for conversion of geographic data-geographic information-geographic knowledge;
3. the theory and method of related subjects such as linguistics, geographic information science, cartography, artificial intelligence and the like are comprehensively applied, text and maps are used as data sources, the method for acquiring geographic knowledge and generating the geographic knowledge map based on the deep learning model is provided, and technical support is provided for systematic, intelligent and engineered construction of the geographic knowledge map.
Drawings
FIG. 1 is a sample geographical knowledge map;
FIG. 2 is a schematic diagram of a geographic knowledge acquisition technique in a map;
FIG. 3 is a sample annotation of partial geographic knowledge in a map;
FIG. 4 is a path map of typhoon "mangosteen" in the embodiment;
FIG. 5 is a diagram of an embodiment of a sub-graph of geographic knowledge;
FIG. 6 is a schematic diagram of link prediction of a geographic knowledge sub-graph in an embodiment;
FIG. 7 is a sample geographical knowledge map generation result in an embodiment;
Detailed Description
To further illustrate the technical means and effects of the present invention for achieving the object of the invention, the detailed description of the embodiments, features and effects of the method for obtaining geographical knowledge according to the present invention is provided below.
The invention discloses a geographic knowledge acquisition method, which comprises the following steps:
and 4, generating a geographical knowledge map.
In the computer environment, there is a need to represent geographical knowledge as computer-acceptable symbols and as storable structures. In the information world, the subject object of geographic knowledge, a "geographic phenomenon or thing", is abstracted as a geographic entity, and various features of it are abstracted as information such as time, space, attributes, states, relationships, and the like. A geographic entity is an materialized abstraction of a human being for the purpose of describing and expressing a geographic phenomenon or thing in the objective world with a particular spatial extent, form, process, relationship, and attribute. Thus, a geographic entity is a subject object and core element of a representation of geographic knowledge in a computer.
The knowledge graph adopts a unified triple < entity, attribute and relation >, takes an entity/concept as a node and a relation as a side, and establishes a knowledge graph structure based on a semantic network, so that knowledge acquisition, knowledge fusion and knowledge reasoning have remarkable advantages in the aspects of operability and computability. A knowledge graph is a knowledge system which represents knowledge at two levels, namely a knowledge definition (Schema) and a knowledge Instance (Instance), in a unified manner. Fig. 1 illustrates a sample geographical knowledge graph, in which the triplets "capital (china, beijing)", "capital" represents the relationship, "chinese" head entity, "beijing" is the tail entity, each entity also has attribute information such as population, area, longitude and latitude, and the above knowledge description form is not applicable to the geographical knowledge graph, and is mainly embodied in two aspects: firstly, the modeling of time and space characteristics is lacked, and the specific space-time characteristics and the geological mechanism characteristics of geographic knowledge cannot be embodied. In essence, knowledge is the expression of a subject about the state of things and their laws of variation. Thus, time and space should be the basic framework and important dimensions for knowledge description and expression, whether generic or domain-specific geographic. In addition, "entity" in an existing knowledge-graph refers to something that is distinguishable and exists independently, and attribute values correspond to attributes of particular concepts and entities. It is obvious that this "entity" does not directly correspond to a geographical phenomenon or thing that the geographical knowledge describes and expresses. The invention constructs a geographical knowledge representation model which takes space-time as a frame and takes a geographical entity as a core on the basis of the conventional knowledge map triple. Specifically, the geographic knowledge is characterized and expressed by octave < time, space, attribute, behavior, state, process, relationship, operation > to form semantic units with different granularities and different levels of geographic knowledge.
As a preferred technical solution, the natural language in step 1 at least includes a text;
the acquisition of geographical knowledge in the text at least comprises: time information extraction, geographic entity identification, attribute information extraction, geographic entity relationship extraction and event information extraction, wherein:
1. the time information may be divided into explicit time information and implicit time information. Wherein, the explicit time information refers to a general time expression with a relatively clear concept, such as "2018" and the like; implicit time is information hidden in the semantics, such as "when an earthquake occurs", without fixed lexical rules. The text characteristics of the explicit time information are obvious, the number of the special time nouns is limited, and the extraction of the explicit time information can be realized by adopting a rule model, a maximum entropy classifier, a conditional random field, pattern matching and the like. Implicit time information extraction can be divided into two phases: firstly, starting from a shallow semantic structure, formulating a grammar rule for extracting time words, and secondly, adding a characteristic for expressing long-distance context dependent information into a machine learning model;
2. the place name is the main expression form of geographic entities in natural language, and research shows that about 70% of texts contain place name information. Place name recognition refers to the extraction of place name from text by computer, mainly adopts a method based on rule model and a method based on machine learning model, the former mainly carries out place name recognition by inducing the language expression rule of place name expression, and has the advantages of easy realization and high accuracy, but the place name recognition depends on the completeness of a place name dictionary, and can not solve the problems of new place name recognition and semantic diversity; the latter takes a labeled corpus as training and testing data, and usually adopts models such as hidden Markov, support vector machine, maximum entropy, conditional random field and the like. Research shows that the conditional random field model has good performance, and knowledge bases such as place name feature words and context features are beneficial to improving the place name recognition effect. In recent years, deep learning methods such as a convolutional neural network and a deep belief network are gradually applied to place name recognition;
3. an attribute is a depiction of a certain characteristic of an entity, and an attribute value is a specific value assigned to an attribute. The attribute extraction needs to acquire the attribute type and the attribute value thereof related to the entity, and is an important means for mining valuable semantic units. At present, there are three main methods for extracting geographic attribute information: ontology semantic based method, rule matching based method, supervised learning method and weakly supervised learning method. The attribute information extracted from the text also needs to be normalized in attribute value: firstly, referring to guiding standards or using conventions in the geographic field, the method is converted into a uniform expression form; secondly, a knowledge base method, a corpus method and a co-occurrence frequency method are adopted, and similarity description of the same attribute is fused through similarity measurement;
4. the extraction of the geographic entity relationship is to distinguish the semantic relationship existing between two geographic entities from the text, and mainly focuses on the following three aspects: 1) the time relationship theory defines 13 types of time relationships such as "same", "overlapping", "previous", and the like. Because time is used as a natural and ordered concept and has the capabilities of reasoning and calculation, the information extraction and calculation of the time relation in the text can be realized by formulating a relevant rule according to common knowledge; 2) spatial relational expression is a basic function of human languages, and each language has a set of vocabulary systems capable of completely expressing spatial relations. The spatial relation vocabulary usually adopts a manual induction method and a Bootstrapping method, and a syntactic pattern can adopt a sequence alignment method for clustering and generalization. The construction of a spatial relationship labeling corpus is a complex system engineering, and the labeling quality and the data scale of the corpus have a decisive role in spatial relationship extraction. At present, a method of combining a rule model and a machine learning model is mainly adopted for extracting the spatial relationship, and the extraction performance of the direction relationship and the distance relationship is obviously superior to that of the topological relationship under the common condition; 3) the semantic relation extraction mainly comprises a supervised learning method, a semi-supervised learning method and an unsupervised learning method. In addition, high-quality entity relation examples are mapped into large-scale texts by virtue of an entity knowledge Base (such as DBPedia, YAGO, Open Cyc, Free Base and the like) irrelevant to the external field, so that the entity semantic relation extraction performance can be improved to a great extent;
5. the event information extraction is to present a text containing event information in a structured form, and not only the subject information of an event needs to be detected, but also relevant attributes such as time, place, role, behavior and the like need to be identified. At present, event information extraction mainly adopts a maximum entropy, a support vector machine, a conditional random field and a deep learning method.
As a preferred technical scheme, the graphic language at least comprises a map, and the information in the map is extracted by adopting a convolutional neural network. When map knowledge is acquired, the symbols and the notations in the map need to have better compatibility on content and type, and the method is suitable for the continuous change of processing objects. Therefore, the invention refers to the relevant theories of linguistics and cartography, and researches the method for acquiring the geographic knowledge in the map by combining the rule model and the deep learning model from three levels of grammar, semantics and pragmatics, and the specific technical process is shown in figure 2.
The convolutional neural network is a feedforward neural network containing convolutional calculation and having a deep structure, is one of representative algorithms for deep learning, is a core algorithm in the field of image recognition, and is widely applied to the fields of computer vision, natural language processing, remote sensing science, atmospheric science and the like. The most typical convolutional neural network consists of an input layer, a convolutional layer, a pooling layer, a fully-connected layer, and an output layer. The convolutional layers realize local feature extraction through filtering operation, each convolutional layer corresponds to a plurality of different convolutional kernels, and the features extracted by each convolutional kernel correspond to one type of features in the original input; the pooling layer reduces the dimension of the feature vector output by the convolutional layer by carrying out aggregation statistics on the features; the fully-connected layer is equivalent to a hidden layer in a multilayer perceptron in a traditional neural network, namely, each neuron of the previous layer is connected with the next layer, and all neurons between the two layers are connected in a weighted mode. The convolutional neural network can be used for supervised learning and unsupervised learning and is suitable for map symbol identification and information extraction.
As a preferred technical solution, the extracting information in the map by using the convolutional neural network includes:
s1, constructing a map information labeling sample library;
s2, automatically identifying the information in the map;
and S3, constructing a deep convolutional neural network model according to the map symbols and the text annotation reference information, and realizing geographic entity identification and extraction of characteristic information and association relation.
As a preferred technical solution, in S1, marking content and marking specifications of geographic information in a map are formulated, and as shown in fig. 3, map samples of different types, different contents, and different compiling forms are selected to construct a map information marking sample library.
As a preferable embodiment, in S2, a rule model of map auxiliary elements, supplementary explanation, and graphic/image elements is created based on the geographic theory, and an algorithm is designed to realize automatic recognition of information in the map.
As a preferred technical solution, fusing information in a text and a map, including:
1) carrying out concept mapping on the geographic knowledge described in the text and the map according to a unified concept classification system, and solving the problem of inconsistent granularity and grade of the geographic knowledge;
2) establishing geographic entity links of all concept levels according to the geographic entity types, the text similarity and the attribute characteristics, and solving the problems of the semantic ambiguity and the synonymy of the geographic entities in the text and the map;
3) processing relevant characteristic information of the geographic entities in the text and the map by using a conflict detection and/or truth value discovery technology by taking the geographic entities as a unit, wherein the relevant characteristic information at least comprises any one of duplication removal, association and combination;
4) and constructing a constraint rule set of the geographic entity relationship, and reconstructing the related relationship of the geographic concept, the example and the feature in the text and the map.
As a preferred technical solution, the geographical knowledge graph in step 4 is based on a convolutional neural network technology, and includes:
i, constructing a geographical knowledge subgraph by referring to the composition and the relation of knowledge units in a geographical knowledge representation model according to geographical knowledge segments acquired from texts and maps;
II, carrying out link processing on the nodes of each group of knowledge subgraphs under the constraint of space-time characteristics;
III, judging different states and relations of the single geographic entity through model iteration and normalization processing, and outputting a state conversion process of the geographic entity according to a time change and space change sequence;
and IV, judging the incidence relation among different geographic entities and among different geographic features to generate geographic knowledge maps with different granularities and different levels.
The following simulation was performed using a specific embodiment for acquisition of geographical knowledge and map generation:
example (b):
typhoon is a grade of tropical cyclones north of the equator, west of the sun border, asian pacific countries or regions. In 2018, 9, 7, and 20 days, typhoon "mangosteen" is generated on the pacific surface in the northwest of the west; 15 days 9 months later, the typhoon "mangosteen" logged in from the north of the philippines; at 18 days, the Guangdong province defense always decides to promote the windproof II-level emergency response to I level; the wind was landed on the banquet town of Guangdong Taishan sea on day 16, at which the maximum wind power near the center was 14 th and the minimum air pressure at the center was 955 hectopascal. By 17 th 18 th 9 th 2018, nearly 300 thousands of people in Guangdong, Guangxi, Hainan, Hunan and Guizhou provinces (areas) suffer from disasters, 5 people die, 1 person is lost, and 160.1 thousands of people are transferred and placed in an emergency risk avoidance way; according to the introduction of related responsible persons of the emergency management department, more than 1200 houses in 5 provinces (regions) are collapsed due to typhoon 'mangosteen', more than 800 houses are seriously damaged, and nearly 3500 houses are generally damaged; the disaster area of crops is 174.4 kilo hectares, wherein 3.3 kilo hectares are absolutely harvested; direct economic loss of 52 billion yuan. (Note: Green represents time information and yellow represents spatial information.)
[ map sample ] A typhoon "mangosteen" route map published by the weather station in Dongguan city is detailed in FIG. 4.
The process of geographic knowledge acquisition for this embodiment is as follows:
after extracting and integrating the geographic information in the text and map samples, forming corresponding geographic knowledge subgraphs, as shown in fig. 5; referring to the concept types and space-time characteristics of the knowledge units represented by the vertexes and other nodes of the subgraph, and performing vertex link prediction on the subgraph by using a graph convolution neural network, as shown in FIG. 6; through operations such as merging and association of subgraph vertices, a knowledge graph for a typhoon mangosteen event is generated, as shown in fig. 7. Therefore, the geographical knowledge graph shown in the invention can express geographical knowledge of different abstract levels of things, events and phenomena, and related characteristic knowledge can be refined on different granularities, so that the application requirements of different users on common sense and professional geographical knowledge can be met.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (8)
1. The geographic knowledge acquisition method is characterized by comprising the following steps:
step 1, analyzing the source of geographic knowledge, and dividing the obtained source of the geographic knowledge into two categories, namely natural language and graphic language;
step 2, carrying out feature analysis and concept modeling on the acquired geographic knowledge;
step 3, extracting geographic information from the acquired geographic knowledge, wherein the extraction at least comprises identification of a geographic entity, extraction of characteristic information and association between the geographic entity and the characteristic information;
and 4, generating a geographical knowledge map.
2. The method according to claim 1, wherein the natural language in step 1 includes at least text;
the acquisition of geographic knowledge in the text at least comprises: time information extraction, geographic entity identification, attribute information extraction, geographic entity relationship extraction and event information extraction.
3. The method according to claim 1 or 2, wherein the graphic language at least includes a map, and the information in the map is extracted by using a convolutional neural network.
4. The method of claim 3, wherein extracting information from the map using a convolutional neural network comprises:
s1, constructing a map information labeling sample library;
s2, automatically identifying the information in the map;
and S3, constructing a deep convolutional neural network model according to the map symbols and the text annotation reference information, and realizing geographic entity identification and extraction of characteristic information and association relation.
5. The method for acquiring geographic knowledge according to claim 4, wherein in S1, marking content and marking specifications of geographic information in a map are formulated, and map samples of different types, different contents and different compiling forms are selected to construct the map information marking sample library.
6. The method for acquiring geographic knowledge according to claim 4, wherein in step S2, a regular model of map auxiliary elements, supplementary description, graphic/image elements is established based on the theory of cartography, and an algorithm is designed to realize automatic recognition of information in the map.
7. The method for acquiring geographical knowledge according to claim 4, wherein the fusing the information in the text and the map comprises:
1) carrying out concept mapping on the text and the geographic knowledge described in the map according to a unified concept classification system;
2) establishing geographic entity links of each concept level according to the geographic entity types, the text similarity and the attribute characteristics;
3) processing the text and the relevant characteristic information of the geographic entity in the map by using a conflict detection and/or truth value discovery technology by taking the geographic entity as a unit, wherein the processing at least comprises any one of deduplication, association and combination;
4) and constructing a constraint rule set of the geographic entity relationship, and reconstructing the relevant relationship between the text and the geographic concept, example and feature in the map.
8. The method for acquiring geographical knowledge according to claim 1, wherein the geographical knowledge graph in the step 4 is based on a convolutional neural network technology, and comprises:
i, constructing a geographical knowledge subgraph by referring to the composition and the relation of knowledge units in a geographical knowledge representation model according to geographical knowledge segments obtained from the text and the map;
II, performing link processing on the nodes of each group of knowledge subgraphs under the constraint of space-time characteristics;
III, judging different states and relations of the single geographic entity through model iteration and normalization processing, and outputting a state conversion process of the geographic entity according to a time change and space change sequence;
and IV, judging the incidence relation among different geographic entities and among different geographic features to generate geographic knowledge maps with different granularities and different levels.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011189625.9A CN112256888A (en) | 2020-10-30 | 2020-10-30 | Geographic knowledge acquisition method |
| PCT/CN2021/072793 WO2022088526A1 (en) | 2020-10-30 | 2021-01-20 | Geographic knowledge acquisition method |
| JP2022505247A JP7468929B2 (en) | 2020-10-30 | 2021-01-20 | How to acquire geographical knowledge |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011189625.9A CN112256888A (en) | 2020-10-30 | 2020-10-30 | Geographic knowledge acquisition method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112256888A true CN112256888A (en) | 2021-01-22 |
Family
ID=74269085
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011189625.9A Pending CN112256888A (en) | 2020-10-30 | 2020-10-30 | Geographic knowledge acquisition method |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP7468929B2 (en) |
| CN (1) | CN112256888A (en) |
| WO (1) | WO2022088526A1 (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112988946A (en) * | 2021-05-07 | 2021-06-18 | 泰瑞数创科技(北京)有限公司 | Geographic entity database user customization method |
| CN113065000A (en) * | 2021-03-29 | 2021-07-02 | 泰瑞数创科技(北京)有限公司 | Multisource heterogeneous data fusion method based on geographic entity |
| CN113139065A (en) * | 2021-04-25 | 2021-07-20 | 杭州电子科技大学 | Mixed knowledge map construction method fusing geographical knowledge |
| CN113297395A (en) * | 2021-07-08 | 2021-08-24 | 中国人民解放军国防科技大学 | Spatio-temporal multi-modal mixed data processing method, correlation method and indexing method |
| CN113486136A (en) * | 2021-08-04 | 2021-10-08 | 泰瑞数创科技(北京)有限公司 | Method and system for assembling geographic entity service on demand |
| CN113505234A (en) * | 2021-06-07 | 2021-10-15 | 中国科学院地理科学与资源研究所 | Construction method of ecological civilization geographical knowledge map |
| CN114138923A (en) * | 2021-12-03 | 2022-03-04 | 吉林大学 | Method for constructing geological map knowledge graph |
| CN114153928A (en) * | 2021-12-03 | 2022-03-08 | 中国电信股份有限公司 | Method, system, equipment and medium for constructing urban geographic semantic knowledge network |
| CN114564966A (en) * | 2022-03-04 | 2022-05-31 | 中国科学院地理科学与资源研究所 | Spatial relation semantic analysis method based on knowledge graph |
| CN115048478A (en) * | 2022-08-12 | 2022-09-13 | 深圳市其域创新科技有限公司 | Construction method, equipment and system of geographic information map of intelligent equipment |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115269751B (en) * | 2022-05-10 | 2023-05-30 | 泰瑞数创科技(北京)股份有限公司 | Method for constructing geographic entity space-time knowledge graph ontology library |
| CN114707004B (en) * | 2022-05-24 | 2022-08-16 | 国网浙江省电力有限公司信息通信分公司 | Method and system for extracting and processing case-affair relation based on image model and language model |
| CN116167440B (en) * | 2023-04-26 | 2023-09-05 | 北京大学 | Space-time knowledge rule judging method based on grid space-time knowledge graph and related equipment |
| CN116450765B (en) * | 2023-06-16 | 2023-08-25 | 山东省国土测绘院 | Polymorphic geographic entity coding consistency processing method and system |
| CN117332091B (en) * | 2023-08-29 | 2024-03-29 | 泰瑞数创科技(北京)股份有限公司 | Geographic entity space-time knowledge graph construction method based on semantic relation |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107133220A (en) * | 2017-06-07 | 2017-09-05 | 东南大学 | Name entity recognition method in a kind of Geography field |
| US10496678B1 (en) * | 2016-05-12 | 2019-12-03 | Federal Home Loan Mortgage Corporation (Freddie Mac) | Systems and methods for generating and implementing knowledge graphs for knowledge representation and analysis |
| CN111488467A (en) * | 2020-04-30 | 2020-08-04 | 北京建筑大学 | Construction method and device of geographical knowledge graph, storage medium and computer equipment |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1967524B (en) | 2005-11-15 | 2010-07-21 | 日电（中国）有限公司 | Collecting and inquiry system of traffic information and method thereof |
| JP5637073B2 (en) | 2011-05-31 | 2014-12-10 | 富士通株式会社 | Information processing apparatus, information processing method, and program |
| US9348815B1 (en) * | 2013-06-28 | 2016-05-24 | Digital Reasoning Systems, Inc. | Systems and methods for construction, maintenance, and improvement of knowledge representations |
| CN106547880B (en) * | 2016-10-26 | 2020-05-12 | 重庆邮电大学 | Multi-dimensional geographic scene identification method fusing geographic area knowledge |
| US10824675B2 (en) | 2017-11-17 | 2020-11-03 | Microsoft Technology Licensing, Llc | Resource-efficient generation of a knowledge graph |
| US11625620B2 (en) | 2018-08-16 | 2023-04-11 | Oracle International Corporation | Techniques for building a knowledge graph in limited knowledge domains |
| CN110472066B (en) * | 2019-08-07 | 2022-03-25 | 北京大学 | Construction method of urban geographic semantic knowledge map |
| CN111428048A (en) * | 2020-03-20 | 2020-07-17 | 厦门渊亭信息科技有限公司 | Cross-domain knowledge graph construction method and device based on artificial intelligence |
-
2020
- 2020-10-30 CN CN202011189625.9A patent/CN112256888A/en active Pending
-
2021
- 2021-01-20 WO PCT/CN2021/072793 patent/WO2022088526A1/en active Application Filing
- 2021-01-20 JP JP2022505247A patent/JP7468929B2/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10496678B1 (en) * | 2016-05-12 | 2019-12-03 | Federal Home Loan Mortgage Corporation (Freddie Mac) | Systems and methods for generating and implementing knowledge graphs for knowledge representation and analysis |
| CN107133220A (en) * | 2017-06-07 | 2017-09-05 | 东南大学 | Name entity recognition method in a kind of Geography field |
| CN111488467A (en) * | 2020-04-30 | 2020-08-04 | 北京建筑大学 | Construction method and device of geographical knowledge graph, storage medium and computer equipment |
Non-Patent Citations (2)
| Title |
|---|
| 张雪英，张春菊: "顾及时空特征的地理知识图谱构建方法", 《中国科学:信息科学》 * |
| 杨丹: "基于神经网络的地图符号识别的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113065000B (en) * | 2021-03-29 | 2021-10-22 | 泰瑞数创科技(北京)有限公司 | Multisource heterogeneous data fusion method based on geographic entity |
| CN113065000A (en) * | 2021-03-29 | 2021-07-02 | 泰瑞数创科技(北京)有限公司 | Multisource heterogeneous data fusion method based on geographic entity |
| CN113139065A (en) * | 2021-04-25 | 2021-07-20 | 杭州电子科技大学 | Mixed knowledge map construction method fusing geographical knowledge |
| CN113139065B (en) * | 2021-04-25 | 2022-07-22 | 杭州电子科技大学 | Mixed knowledge graph construction method fusing geographical knowledge |
| CN112988946B (en) * | 2021-05-07 | 2021-08-31 | 泰瑞数创科技(北京)有限公司 | Geographic entity database user customization method |
| CN112988946A (en) * | 2021-05-07 | 2021-06-18 | 泰瑞数创科技(北京)有限公司 | Geographic entity database user customization method |
| CN113505234A (en) * | 2021-06-07 | 2021-10-15 | 中国科学院地理科学与资源研究所 | Construction method of ecological civilization geographical knowledge map |
| CN113505234B (en) * | 2021-06-07 | 2023-11-21 | 中国科学院地理科学与资源研究所 | Construction method of ecological civilized geographic knowledge graph |
| CN113297395A (en) * | 2021-07-08 | 2021-08-24 | 中国人民解放军国防科技大学 | Spatio-temporal multi-modal mixed data processing method, correlation method and indexing method |
| CN113486136A (en) * | 2021-08-04 | 2021-10-08 | 泰瑞数创科技(北京)有限公司 | Method and system for assembling geographic entity service on demand |
| CN113486136B (en) * | 2021-08-04 | 2022-06-17 | 泰瑞数创科技(北京)有限公司 | Method and system for assembling geographic entity service on demand |
| CN114138923A (en) * | 2021-12-03 | 2022-03-04 | 吉林大学 | Method for constructing geological map knowledge graph |
| CN114153928A (en) * | 2021-12-03 | 2022-03-08 | 中国电信股份有限公司 | Method, system, equipment and medium for constructing urban geographic semantic knowledge network |
| CN114564966A (en) * | 2022-03-04 | 2022-05-31 | 中国科学院地理科学与资源研究所 | Spatial relation semantic analysis method based on knowledge graph |
| CN115048478A (en) * | 2022-08-12 | 2022-09-13 | 深圳市其域创新科技有限公司 | Construction method, equipment and system of geographic information map of intelligent equipment |
| CN115048478B (en) * | 2022-08-12 | 2022-10-21 | 深圳市其域创新科技有限公司 | Construction method, equipment and system of geographic information map of intelligent equipment |
| WO2024032717A1 (en) * | 2022-08-12 | 2024-02-15 | 深圳市其域创新科技有限公司 | Geographic information graph constructing method and system for intelligent devices, and device |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2023502827A (en) | 2023-01-26 |
| JP7468929B2 (en) | 2024-04-16 |
| WO2022088526A1 (en) | 2022-05-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112256888A (en) | Geographic knowledge acquisition method | |
| CN108563653B (en) | Method and system for constructing knowledge acquisition model in knowledge graph | |
| CN105631479B (en) | Depth convolutional network image labeling method and device based on non-equilibrium study | |
| CN107480125B (en) | Relation linking method based on knowledge graph | |
| CN109934261B (en) | Knowledge-driven parameter propagation model and few-sample learning method thereof | |
| CN111753024B (en) | Multi-source heterogeneous data entity alignment method oriented to public safety field | |
| CN112200317A (en) | Multi-modal knowledge graph construction method | |
| CN110232186A (en) | The knowledge mapping for merging entity description, stratification type and text relation information indicates learning method | |
| CN112860908A (en) | Knowledge graph automatic construction method based on multi-source heterogeneous power equipment data | |
| CN107871158A (en) | A kind of knowledge mapping of binding sequence text message represents learning method and device | |
| CN114064918A (en) | Multi-modal event knowledge graph construction method | |
| WO2023108991A1 (en) | Model training method and apparatus, knowledge classification method and apparatus, and device and medium | |
| WO2017193471A1 (en) | Digital global sharing platform for preserving dongba ancient texts | |
| CN113947161A (en) | Attention mechanism-based multi-label text classification method and system | |
| CN112949281A (en) | Incremental social event detection method for graph neural network | |
| CN113779220A (en) | Mongolian multi-hop question-answering method based on three-channel cognitive map and graph attention network | |
| Yin et al. | Sentiment lexical-augmented convolutional neural networks for sentiment analysis | |
| CN114897167A (en) | Method and device for constructing knowledge graph in biological field | |
| CN115496072A (en) | Relation extraction method based on comparison learning | |
| CN116303996A (en) | Theme event extraction method based on multifocal graph neural network | |
| CN116796744A (en) | Entity relation extraction method and system based on deep learning | |
| CN111967267A (en) | XLNET-based news text region extraction method and system | |
| CN116383352A (en) | Knowledge graph-based method for constructing field intelligent question-answering system by using zero samples | |
| CN112784017A (en) | Archive cross-modal data feature fusion method based on main affinity expression | |
| CN111259106A (en) | Relation extraction method combining neural network and feature calculation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210122 |
|
| RJ01 | Rejection of invention patent application after publication |