WO2022095256A1 - 一种地理编码方法、系统、终端以及存储介质 - Google Patents

一种地理编码方法、系统、终端以及存储介质 Download PDF

Info

Publication number
WO2022095256A1
WO2022095256A1 PCT/CN2020/139759 CN2020139759W WO2022095256A1 WO 2022095256 A1 WO2022095256 A1 WO 2022095256A1 CN 2020139759 W CN2020139759 W CN 2020139759W WO 2022095256 A1 WO2022095256 A1 WO 2022095256A1
Authority
WO
WIPO (PCT)
Prior art keywords
path
place name
node
geocoding
address
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.)
Ceased
Application number
PCT/CN2020/139759
Other languages
English (en)
French (fr)
Inventor
钱静
彭树宏
陈朝亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Publication of WO2022095256A1 publication Critical patent/WO2022095256A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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
    • G06F16/23—Updating
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06F16/28—Databases characterised by their database models, e.g. relational or object models
    • G06F16/284—Relational databases
    • G06F16/288—Entity relationship models

Definitions

  • the present application belongs to the technical field of geocoding, and in particular, relates to a geocoding method, system, terminal and storage medium.
  • geographic information system As the product of the combination of location service and information platform, geographic information system has a wider and wider range of applications. With the popularization and continuous maturity of geographic information technology, many enterprises, units and government departments have established business based on geographic information, such as pharmaceuticals, media, etc., and the demand for management and operation with the help of geographic information has become increasingly prominent. However, the naming methods of geographic information such as national place names and addresses have the characteristics of messy semantics and disordered word order, that is, there is no unified criterion to standardize them. In addition, the geographic information that can be collected by ordinary departmental units is only the textual description information (non-spatial information) of various disorganized place names and addresses, and the spatial coordinate information that can be used directly cannot be obtained.
  • the present application provides a geocoding method, system, terminal and storage medium, aiming to solve one of the above technical problems in the prior art at least to a certain extent.
  • a geocoding method that includes:
  • a geocoding library is established according to the place name and address model, and the geocoding library includes an administrative area entity data table, a street and alley entity data table, and a community entity data table;
  • the technical solution adopted in the embodiment of the present application further includes: before the establishment of the place-name-address model according to the place-name-address data, the following further includes:
  • the technical solution adopted in the embodiment of the present application further includes: the establishing of the geographic coding library according to the place name and address model includes:
  • the address dictionary-based, using the N-shortest path optimization algorithm to perform word segmentation and standardization on the place name address data includes:
  • the place name groups in the place name address data are matched, and a directed acyclic graph is constructed.
  • Each phrase is a node in the directed acyclic graph, and corresponds to a side given length;
  • All possible word edges of the directed acyclic graph are established according to preset rules, so that all words contained in the geographical name geographic data correspond to the edges of the directed acyclic graph one-to-one respectively, and solve the To the N-shortest path set from the start node to the end node in the acyclic graph, the place name address data is segmented according to the N-shortest path set.
  • the preset rules for establishing all possible word edges in the directed acyclic graph are: :
  • a directed edge ⁇ Vi-1, Vj> is established between nodes Vi-1 and Vj, the length of the edge is Lw, and the word corresponding to the edge is w(0 ⁇ i ⁇ j ⁇ n).
  • the technical solutions adopted in the embodiments of the present application further include: the solving of the set of N-shortest paths from the start node to the end node in the directed acyclic graph includes:
  • Path(i,j) is the set of all paths from node Vi to node Vj;
  • Length(path) is the length of the path path, and the value of Length(path) is equal to the sum of the lengths of all edges in the path;
  • LS is directed acyclic
  • NLS be the set of N-shortest path lengths from V0 to Vn
  • NSP be the set of N-shortest path lengths from V0 to Vn
  • RS is the final N-shortest path rough division result set
  • min(
  • NSP ⁇ path
  • path ⁇ Path(0,n),Length(path) ⁇ NLS ⁇ RS ⁇ w1w2...wm
  • , wi is path The word corresponding to the i-th edge of , i 1,2,...,m, where path ⁇ NSP ⁇ , n is the number of shortest paths.
  • the said address dictionary-based, using the N-shortest path optimization algorithm to perform word segmentation and standardization on the place name address data further includes:
  • a geographic coding system comprising:
  • Place name address model building module used to build place name address model based on place name address data
  • Geographical coding library building module used to establish a geographic coding library according to the place name and address model, and the geographic coding library includes an administrative area entity data table, a street and lane entity data table and a community entity data table;
  • Word segmentation and standardization processing module used to perform word segmentation and standardization processing on the place name address data based on the address dictionary, using the N-shortest path optimization algorithm, and divide the place name address data into at least one phrase;
  • Coordinate matching module used to convert the at least one phrase into a character string in a predetermined format according to the level element in the place name address model, match the character string with the corresponding geographic coordinates in the geocoding library, and The geographic coordinates matched by the character string are used as the standard geographic coordinates of the address corresponding to the place name.
  • a terminal includes a processor and a memory coupled to the processor, wherein,
  • the memory stores program instructions for implementing the geocoding method
  • the processor is configured to execute the program instructions stored in the memory to control geocoding.
  • a storage medium storing program instructions executable by a processor, where the program instructions are used to execute the geocoding method.
  • the beneficial effects of the embodiments of the present application are: the geocoding method, system, terminal and storage medium of the embodiments of the present application perform word segmentation and standardization processing on place names and addresses based on the N-shortest path optimization algorithm. Results After the place name address is segmented, the segmented place name address is converted into a string that can be recognized by the computer according to the level elements in the place name address model, and finally the string is matched with the corresponding geographic coordinates in the geocoding library. Matching results assign standard geographic coordinates to place-name addresses.
  • the present application improves the disadvantages of word-by-word traversal, increases the practicability, and inherits the advantages of the full segmentation idea, which can not only reduce the number of segmented phrases as much as possible, but also include All results that need to be retained can effectively avoid wasting resources and increase search efficiency.
  • Fig. 1 is the flow chart of the geocoding method of the first embodiment of the present application
  • FIG. 2 is a schematic diagram of a representation of a place name and address according to an embodiment of the application
  • FIG. 3 is a schematic structural diagram of a directed acyclic graph according to an embodiment of the present application.
  • FIG. 4 is a schematic diagram of a directed acyclic graph solution process according to an embodiment of the application.
  • FIG. 5 is a schematic diagram of a precursor record table in the process of solving a directed acyclic graph according to an embodiment of the present application
  • Fig. 6 is the flow chart of the geocoding method of the second embodiment of the present application.
  • FIG. 7 is a schematic diagram of an N-shortest path improved word segmentation algorithm according to an embodiment of the application.
  • FIG. 8 is a schematic structural diagram of a geocoding system according to an embodiment of the present application.
  • FIG. 9 is a schematic structural diagram of a terminal according to an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of a storage medium according to an embodiment of the present application.
  • the geocoding method of the embodiment of the present application first performs data cleaning on the initial place name address data to prevent problems such as excessive typos, spelling mistakes, and text repetition in the input text; and then establishes a place name address model to enable it to Reflect the different representations of geographic names in a country or region, and then build a geographic coding library including a place name data table, a building data table and a door (building) plate data table according to the place name address model, and use the N-shortest path optimization algorithm.
  • the place name address is subjected to word segmentation and standardization processing.
  • the segmented place name address is converted into a string that can be recognized by the computer according to the level elements in the place name address model.
  • the corresponding geographic coordinates in the encoding library are matched.
  • FIG. 1 is a flowchart of the geocoding method according to the first embodiment of the present application.
  • the geocoding method of the first embodiment of the present application includes the following steps:
  • the present invention uses Trillum technology, adopts syntax analysis and fuzzy matching algorithm to perform data cleaning on place name addresses.
  • FIG. 2 it is a schematic diagram of the representation of place names and addresses.
  • the place name of the administrative region includes the provincial, city, county, township, street name, community place name, community name, gate building address, landmark name or alias, and unit name or its abbreviation;
  • the provincial level has priority over the city level
  • the city level has priority over the county level
  • the county level has priority over the township level
  • street and lane names have priority over community place names
  • community place names have priority over community names
  • gate building addresses have priority over landmark names or their aliases, followed by is the unit name or its abbreviation.
  • the street name and community name in a city are unique, so using the street name or community name can roughly lock a certain range of addresses, and using "street name or community name + door (floor)" Brand” can be accurately located to a location, and “administrative region place name + marker name” can basically be used to accurately locate a location. That is to say, when the content to be expressed in the text has a house number, use "street name or community name + door (building) number” to lock a location; when the local name address data contains a landmark name, use " Administrative region place name + landmark name” for precise positioning.
  • An example of the structure of place name address data according to the above description rules of granularity range is as follows:
  • the place name address data is "Huizhou College, No. 46 Yanda Avenue, Huizhou City", which can be simplified as “No. 46 Yanda Avenue” in the application of Huizhou City without any ambiguity at all; and if the place name address data is "Guangzhou City" "ICBC", at this time, multiple markers may be located, and the results obtained are difficult to filter, so it needs to be extended to the street name or community name for description, and then a certain ICBC can be accurately located.
  • S12 According to the place name and address model, establish a geocoding library including the entity data table of administrative areas, the entity data table of streets and alleys, and the entity data table of community;
  • the table structure of the administrative area entity data table, the street entity data table and the community entity data table can be defined according to the application scenario, and the establishment of the geocoding library follows the following principles:
  • Standardization principle The coding rules are adapted to the national standard system for data sharing.
  • each table enter all provinces, districts, counties, streets, communities, landmarks, house numbers and geographic coordinates in turn to construct the geocoding database. Select the field value in each data table as the place name address entry, and record it in the address dictionary together with the corresponding address level.
  • an address alias is used as a place-name address entry, it is also necessary to record the standard name so that the address elements can be normalized during address segmentation.
  • the implementation process of the N-shortest path optimization algorithm is as follows: the address dictionary records all geographical names addresses (including aliases and abbreviations, etc.) in different countries and regions, first, according to the address dictionary, place names that may appear in the geographical name address data are recorded. The phrases are matched in order, and then a directed acyclic graph is constructed. Each phrase is a node in the directed acyclic graph, and corresponds to a given edge length (ie, weight, in the non-statistical rough scoring model). , assuming that all words are equal, for the convenience of calculation, the side lengths of the corresponding sides of all words are set as the sides of 1). In all paths from the starting point to the end point in the directed acyclic graph, the path value from each node to the source node is obtained, and the corresponding path set is used as the path result set of each node.
  • edge length ie, weight, in the non-statistical rough scoring model
  • Path(i,j) be the set of all paths from the node Vi to the node Vj;
  • Length(path) is the length of the path path, and the value of Length(path) is equal to the sum of the lengths of all edges in the path;
  • LS is the directed and non-directional The set of lengths of all paths from V0 to Vn in the ring graph G; then there are:
  • NLS is the set of N-shortest path lengths from V0 to Vn
  • NSP is the set of N-shortest paths from V0 to Vn
  • RS is the final result set of N-shortest path rough division.
  • the definition of NLS is:
  • min(
  • ,N); a ⁇ LS-NLS,b ⁇ NLS ⁇ a ⁇ b NSP ⁇ path
  • path ⁇ Path(0,n),Length(path) ⁇ NLS ⁇ RS ⁇ w1w2...wm
  • wi is the word corresponding to the i-th edge of the path, i 1,2,...,m, where path ⁇ NSP ⁇ , n is the number of shortest paths.
  • the solution process of the text data is shown in Figure 4.
  • a greedy algorithm is used to obtain the local optimal solution of each node. Record the shortest path value at each node and the predecessor of the node. If a node includes more than two paths of the same length, record the predecessor of the node on each path separately.
  • the predecessor record table of the text data is shown in Figure 5.
  • the present invention uses the N-shortest path word segmentation algorithm to segment the place name address, which can not only greatly reduce the number of word segmentation, but also try to include all possible word segmentation results without loss, and avoid the algorithm itself. At the same time, it can reduce the search space as much as possible and improve the efficiency of word segmentation.
  • S14 Convert the segmented at least one phrase into a character string in a predetermined format (recognizable by the computer) according to the level element in the place name address model, and then match the converted character string with the corresponding geographic coordinates in the geocoding library;
  • FIG. 6 is a flowchart of the geocoding method according to the second embodiment of the present application.
  • the geocoding method of the second embodiment of the present application includes the following steps:
  • the present invention uses Trillum technology, adopts syntax analysis and fuzzy matching algorithm to perform data cleaning on place name addresses.
  • the place name address can be regarded as a hierarchically scalable place name address model.
  • S22 Establish a geocoding library including a place name data table, a building data table and a door (building) sign data table according to the place name address model;
  • the table structure of the place name data table, the building data table and the door (building) plate data table can be defined according to the application scenario, and all provinces, districts, counties, streets, communities, landmarks, house plates are sequentially entered according to the table structure. number and geographic coordinates for the construction of the geocoding library.
  • the N-shortest path improved word segmentation algorithm combining the dynamic deletion algorithm and the N-shortest path word segmentation algorithm is used to segment and standardize the irregular place name address data, and the place name address data is divided into at least one phrase ;
  • the present invention proposes an N-shortest path improved word segmentation algorithm that combines the dynamic deletion algorithm with the N-shortest path word segmentation algorithm.
  • the basic idea of the dynamic deletion algorithm is to construct the shortest path update queue. , used to store the child nodes of the deleted node; delete the node that should be deleted and all child nodes in the original shortest path tree; select the node closest to the root node for updating in the queue, and no longer Insert updated nodes into the queue to reduce the number of node updates.
  • the N-shortest path improved word segmentation algorithm is shown in Figure 7, and its solution process is as follows:
  • Step 1 First, based on the N-shortest path word segmentation algorithm, construct a directed acyclic graph G with words (or characters) as nodes; wherein, the directed acyclic graph construction process is the same as the first embodiment, and this embodiment will not repeat;
  • Lj is used to store the shortest path, where j is a dynamic variable, and the initial value of j can be set according to the length of the entire string, such as "what he said is true", the initial value of j is 8,
  • the initial value of j is 8
  • the nodes between the combinations are deleted, and the j value is updated.
  • the j value will become smaller and smaller until the sentence cannot be divided.
  • Step 3 Starting from the first node in the current path, delete the first node with an in-degree greater than 1, and record the deleted node as Hm, and determine whether the descendant node of Hm is in the set E, if it is in the set In E, calculate the shortest path from the start node V0 to Hm, and record the end node of the shortest path as H'm; if it is not in the set E, delete the node Hm and all its all from the directed acyclic graph G Descendant node; wherein, the set E is the N-shortest path set (ie, NSP) from V0 to Vn, which is used here to determine whether the deleted node is in the shortest path.
  • Hm and H'm represent the end node in each cycle, and H'm will be the end marker of the next cycle.
  • n is the number of shortest paths after deleting nodes
  • the value of j should be moderate, neither too large nor too small, for the first j optimal paths to be reserved.
  • S24 Convert the segmented at least one phrase into a character string in a predetermined format (recognizable by the computer) according to the level element in the place name address model, and then match the converted character string with the corresponding geographic coordinates in the geocoding library;
  • the geocoding method of the embodiment of the present application uses the N-shortest path optimization algorithm to perform word segmentation and standardization processing on the place-name address, and after segmenting the place-name address according to the standardized processing result, according to the level element in the place-name address model.
  • the last place name address is converted into a character string that can be recognized by the computer, and finally the character string is matched with the corresponding geographic coordinates in the geocoding library, and the place name address is given standard geographic coordinates according to the matching result.
  • the present application improves the disadvantages of word-by-word traversal, increases the practicability, and inherits the advantages of the full segmentation idea, which can not only reduce the number of segmented phrases as much as possible, but also include All results that need to be retained can effectively avoid wasting resources and increase search efficiency.
  • FIG. 8 is a schematic structural diagram of a geocoding system according to an embodiment of the present application.
  • the geocoding system 41 of the embodiment of the present application includes:
  • Data cleaning module 41 used for data cleaning of the initial place name and address data; since the text data such as place name and address input by the user terminal may contain typos or repeated words, in order to avoid problems such as inconsistent character strings in the text data, spelling errors, etc. If the subsequent character string is incorrectly matched with the geographic coordinates, the embodiment of the present invention uses the Trillum technology, and uses the syntax analysis and fuzzy matching algorithm to clean the data of the place name address.
  • Place-name and address model building module 42 used to structure the cleaned place-name and address data to establish a place-name and address model; wherein, different countries or regions have different granularity and scope rules for the representation of place-names and addresses, and place-names and addresses can be regarded as a kind of Hierarchically scalable place-name address model.
  • Geocoding library building module 43 used to establish a geographic coding library including a place name data table, a building data table and a door (building) sign data table according to the place name address model; wherein, the place name data table, the building data table and the door (building) )
  • the table structure of the card data table can be defined according to the application scenario, and all provinces, districts, counties, streets, communities, landmarks, house numbers and geographic coordinates are entered in turn according to each table structure to construct a geocoding library.
  • the word segmentation and standardization processing module 44 is used to perform word segmentation and standardization processing on the irregular place name address data by using the N-shortest path optimization algorithm based on the address dictionary, and divide the place name address data into at least one phrase; wherein, the N-shortest path
  • the implementation process of the optimization algorithm is as follows: the address dictionary records all geographical names addresses (including aliases and abbreviations, etc.) in different countries and regions.
  • each phrase is a node in the directed acyclic graph, and corresponds to a given edge length (ie weight, in the non-statistical rough segmentation model, it is assumed that all words are correct. etc., for the convenience of calculation, the side lengths of the corresponding sides of all words are set as the sides of 1).
  • the path value from each node to the source node is obtained, and the corresponding path set is used as the path result set of each node.
  • Path(i,j) be the set of all paths from the node Vi to the node Vj;
  • Length(path) is the length of the path path, and the value of Length(path) is equal to the sum of the lengths of all edges in the path;
  • LS is the directed and non-directional The set of lengths of all paths from V0 to Vn in the ring graph G; then there are:
  • NLS is the set of N-shortest path lengths from V0 to Vn; NSP is the set of N-shortest paths from V0 to Vn; RS is the final N-shortest path rough division result set.
  • the definition of NLS is:
  • min(
  • ,N); a ⁇ LS-NLS,b ⁇ NLS ⁇ a ⁇ b NSP ⁇ path
  • path ⁇ Path(0,n),Length(path) ⁇ NLS ⁇ RS ⁇ w1w2...wm
  • wi is the word corresponding to the i-th edge of path, i 1,2,...,m, where path ⁇ NSP ⁇ .
  • the present invention uses the N-shortest path word segmentation algorithm to segment the place name address, which can not only greatly reduce the number of word segmentation, but also try to include all possible word segmentation results without loss, and avoid the algorithm itself. At the same time, it can reduce the search space as much as possible and improve the efficiency of word segmentation.
  • the word segmentation and standardization processing module 44 adopts the N-shortest path improved word segmentation algorithm that combines the dynamic deletion algorithm and the N-shortest path word segmentation algorithm to perform word segmentation, that is, standardized processing. for:
  • the first step First, based on the N-shortest path word segmentation algorithm, construct a directed acyclic graph G with words as nodes;
  • Step 3 Starting from the first node in the current path, delete the first node with an in-degree greater than 1, and record the deleted node as Hm, and determine whether the descendant node of Hm is in the set E, if it is in the set In E, calculate the shortest path from the start node V0 to Hm, and record the end node of the shortest path as H'm; if it is not in the set E, delete the node Hm and all its all from the directed acyclic graph G descendant node;
  • Coordinate matching module 45 used to convert at least one phrase into a character string in a predetermined format (recognizable by the computer) according to the level element in the place name address model, and then match the converted character string with the corresponding geographic coordinates in the geocoding library , take the geographic coordinates matched by the string as the standard geographic coordinates of the corresponding place name address.
  • FIG. 9 is a schematic structural diagram of a terminal according to an embodiment of the present application.
  • the terminal 50 includes a processor 51 and a memory 52 coupled to the processor 51 .
  • the memory 52 stores program instructions for implementing the above-described geocoding method.
  • the processor 51 is adapted to execute program instructions stored in the memory 52 to control the geocoding.
  • the processor 51 may also be referred to as a CPU (Central Processing Unit, central processing unit).
  • the processor 51 may be an integrated circuit chip with signal processing capability.
  • the processor 51 may also be a general purpose processor, digital signal processor (DSP), application specific integrated circuit (ASIC), off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components .
  • DSP digital signal processor
  • ASIC application specific integrated circuit
  • FPGA off-the-shelf programmable gate array
  • a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
  • FIG. 10 is a schematic structural diagram of a storage medium according to an embodiment of the present application.
  • the storage medium of this embodiment of the present application stores a program file 61 capable of implementing all the above methods, wherein the program file 61 may be stored in the above-mentioned storage medium in the form of a software product, and includes several instructions to enable a computer device (which may It is a personal computer, a server, or a network device, etc.) or a processor that executes all or part of the steps of the methods of the various embodiments of the present invention.
  • a computer device which may It is a personal computer, a server, or a network device, etc.
  • a processor that executes all or part of the steps of the methods of the various embodiments of the present invention.
  • the aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes , or terminal devices such as computers, servers, mobile phones, and tablets.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Remote Sensing (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种地理编码方法、系统、终端以及存储介质。所述方法包括：根据地名地址数据建立地名地址模型；根据地名地址模型建立地理编码库，地理编码库包括行政区域实体数据表、街巷实体数据表以及小区实体数据表；基于地址词典，运用N-最短路径优化算法对地名地址数据进行分词及标准化处理，将地名地址数据切分为至少一个词组；按照地名地址模型中的级别要素将至少一个词组转换成预定格式的字符串，将字符串与地理编码库中对应的地理坐标进行匹配，将字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。能够尽可能减少切分词组的数量，又能包含所有需要被保留的结果，能有效避免资源浪费，加大搜索效率。

Description

一种地理编码方法、系统、终端以及存储介质 技术领域
本申请属于地理编码技术领域，特别涉及一种地理编码方法、系统、终端以及存储介质。
背景技术
地理信息系统作为位置服务与信息化平台结合的产物，其应用范围越来越广泛。随着地理信息技术的普及与不断成熟，众多企业、单位及政府部门都纷纷建立了基于地理信息的业务，例如药业、传媒等，借助地理信息进行管理运作的需求也日益凸显。然而我们国家地名地址等地理信息的命名方式具有语义凌乱、语序混乱等特点，即没有一个统一的准则将它们规范化。另外，通常普通部门单位所能够采集到的地理信息只是各种杂乱无章的地名地址类文字性描述信息(非空间信息)，而无法获取到可以直接使用的空间坐标信息。如果不能将这些非空间信息成功转化为空间坐标信息，相关企业就无法将相关专题数据和地理信息结合起来，间接影响到对GIS软件的可视化、功能分析等应用。因此，如何将与地理位置相关的非空间信息转换成能够被计算机识别的GIS系统地理坐标，实现非空间信息与实体地理坐标的匹配，才能够发挥地理信息系统的最大作用。
发明内容
本申请提供了一种地理编码方法、系统、终端以及存储介质，旨在至少在一定程度上解决现有技术中的上述技术问题之一。
为了解决上述问题，本申请提供了如下技术方案：
一种地理编码方法，包括：
根据地名地址数据建立地名地址模型；
根据所述地名地址模型建立地理编码库，所述地理编码库包括行政区域实体数据表、街巷实体数据表以及小区实体数据表；
基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理，将所述地名地址数据切分为至少一个词组；
按照所述地名地址模型中的级别要素将所述至少一个词组转换成预定格式的字符串，将所述字符串与所述地理编码库中对应的地理坐标进行匹配，将所述字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
本申请实施例采取的技术方案还包括：所述根据地名地址数据建立地名地址模型前还包括：
对所述地名地址数据进行数据清洗。
本申请实施例采取的技术方案还包括：所述根据所述地名地址模型建立地理编码库包括：
定义所述行政区域实体数据表、街巷实体数据表以及小区实体数据表的表结构，按照所述表结构依次录入省、区县、街道、小区、标志物、门牌号和地理坐标进行所述地理编码库的构建。
本申请实施例采取的技术方案还包括：所述基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理包括：
根据所述地址词典顺序匹配出所述地名地址数据中的地名词组，并构建一个有向无环图，每个词组分别是所述有向无环图中的一个节点，且分别对应着一条被赋予边长的边；
按照预设规则建立所述有向无环图所有可能的词边，使得所述地名地理数据中包含的所有词分别与所述有向无环图中的边一一对应，并求解所述有向无环图中从开始节点到结束节点的N-最短路径集合，根据所述N-最短路径集合对所述地名地址数据进行分词。
本申请实施例采取的技术方案还包括：假设所述地名地理数据S＝c1c2……cn，其中ci(i＝1,2,…n)为单个的字，n为串的长度，n≥1，建立的有向无环图G的节点数为n+1，各节点编号依次为V0，V1，V2，…，Vn，所述建立有向无环图所有可能的词边的预设规则为：
相邻节点Vk-1,Vk之间建立有向边<Vk-1,Vk>，边的长度值为Lk，边对应的词默认为ck(k＝1,2,…n)；
若w＝ci ci+1……cj是一个词，则节点Vi-1,Vj之间建立有向边<Vi-1,Vj>，边的长度值为Lw，边对应的词为w(0<i<j≤n)。
本申请实施例采取的技术方案还包括：所述求解所述有向无环图中从开始节点到结束节点的N-最短路径集合包括：
假设Path(i,j)为所有从节点Vi到节点Vj的路径集合；Length(path)为路径path的长度，Length(path)值等于path中所有边的长度之和；LS为有向无环图G中所有从V0到Vn路径的长度集合，则有：
LS＝{len|len＝Length(path),path∈Path(0,n)}
设NLS为V0到Vn的N-最短路径长度集合，NSP为V0到Vn的N-最短路径集合，RS是最终求出的N-最短路径粗分结果集，|NLS|＝min(|LS|,N)；a∈LS-NLS,b∈NLS→a<b，NSP＝{path|path∈Path(0,n),Length(path)∈NLS}RS＝{w1w2…wm|，wi是path的第i条边对应的词，i＝1,2,…,m,其中path∈NSP}，n为最短路径数。
本申请实施例采取的技术方案还包括：所述基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理还包括：
计算从开始节点到结束节点的最短路径为Lj＝1，如果j小于最短路径数并且存在其他候选路径，则更新当前路径L为Lj，反之结束；
从当前路径中的第一个节点开始，将入度大于1的第一个节点删除，并将被删除节点记为Hm，判断Hm的子孙节点是否在集合E中，如果在集合E中，则计算从开始节点到Hm的最短路径，并将该最短路径的结束节点记为H’m；如果不在集合E中，则从有向无环图G中删除节点Hm及其所有子孙节点；其中，集合E为V0到Vn的N-最短路径集合，Hm与H’m在每一个循环当中均代表结束节点，H’m作为下一次循环的结束标记；
重复所述节点删除过程，直到m≮n，更新当前路径，求得开始节点V0到所有节点H’m的最短路径j＝j+1；n为删除节点后的最短路径数，m为j循环构造后的最短路径，在每一次循环当中，m的值为m＝j+1。
本申请实施例采取的另一技术方案为：一种地理编码系统，包括：
地名地址模型构建模块：用于根据地名地址数据建立地名地址模型；
地理编码库构建模块：用于根据所述地名地址模型建立地理编码库，所述地理编码库包括行政区域实体数据表、街巷实体数据表以及小区实体数据表；
分词及标准化处理模块：用于基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理，将所述地名地址数据切分为至少一个词组；
坐标匹配模块：用于按照所述地名地址模型中的级别要素将所述至少一个词组转换成预定格式的字符串，将所述字符串与所述地理编码库中对应的地理坐标进行匹配，将所述字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
本申请实施例采取的又一技术方案为：一种终端，所述终端包括处理器、与所述处理器耦接的存储器，其中，
所述存储器存储有用于实现所述地理编码方法的程序指令；
所述处理器用于执行所述存储器存储的所述程序指令以控制地理编码。
本申请实施例采取的又一技术方案为：一种存储介质，存储有处理器可运行的程序指令，所述程序指令用于执行所述地理编码方法。
相对于现有技术，本申请实施例产生的有益效果在于：本申请实施例的地理编码方法、系统、终端及存储介质基于N-最短路径优化算法对地名地址进行分词及标准化处理，根据标准化处理结果对地名地址进行切分后，按照地名地址模型中的级别要素将切分后的地名地址转换成计算机能够识别的字符串，最后将字符串与地理编码库中相应的地理坐标进行匹配，根据匹配结果为地名地址赋予标准地理坐标。本申请通过在算法中加入辅助的语法与语义规则，改进逐词遍历的弊端，增加了实用性，秉承了全切分思想的优势，既能够尽可能减少切分词组的数量，同时又能包含所有需要被保留的结果，能有效避免资源浪费，加大搜索效率。
附图说明
图1是本申请第一实施例的地理编码方法的流程图；
图2为本申请实施例的地名地址表述方式示意图；
图3为本申请实施例的有向无环图结构示意图；
图4为本申请实施例的有向无环图求解过程示意图；
图5为本申请实施例对有向无环图求解过程中的前驱记录表示意图；
图6是本申请第二实施例的地理编码方法的流程图；
图7为本申请实施例的N-最短路径改进分词算法示意图；
图8为本申请实施例的地理编码系统结构示意图；
图9为本申请实施例的终端结构示意图；
图10为本申请实施例的存储介质的结构示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。
针对现有技术的不足，本申请实施例的地理编码方法首先对初始地名地址数据进行数据清洗，防止输入的文本错别字过多、拼写错误、文本重复等问题；然后建立地名地址模型，使其能够反映一个国家或地区对地理名称的不同表示，再根据该地名地址模型建立包含地名数据表、建筑物数据表以及门(楼)牌数据表的地理编码库，并运用N-最短路径优化算法对地名地址进行分词及标准化处理，根据标准化处理结果对地名地址进行切分后，按照地名地址模型中的级别要素将切分后的地名地址转换成计算机能够识别的字符串，最后将字符串与地理编码库中相应的地理坐标进行匹配。本申请实施例秉承了全切分思想的优势，既能够尽可能减少切分词组的数量，同时又能包含所有需要被保留的结果，能有效避免资源浪费，加大搜索效率。
具体的，请参阅图1，是本申请第一实施例的地理编码方法的流程图。本申请第一实施例的地理编码方法包括以下步骤：
S10：对初始地名地址数据进行数据清洗；
本步骤中，由于用户端输入的地名地址等文本数据中可能会包含错别字或重复字，为了避免由于文本数据中字符串不一致、拼写错误等问题导致后续的字符串与地理坐标匹配错误，本发明实施例使用Trillum技术，采用语法分析和模糊匹配算法对地名地址进行数据清洗。
S11：将清洗后的地名地址数据结构化后，建立地名地址模型；
本步骤中，不同国家或地区对地名地址的表述方式具有不同粒度范围的描述规则，本申请实施例根据不同粒度范围的描述规则建立可伸缩的地名地址模型。具体如图2所示，为地名地址表述方式示意图。在该地名地址表述方式中，行政区域地名所包含省级、市级、县级、乡级、街巷名、小区地名、社区名、门楼址、标志物名或别名以及单位名或其简称；其中，省级优先于市级，市级优先于县级，县级优先于乡级；街巷名优先于小区地名，小区地名优先于社区名，门楼址优先于标志物名或其别名，其次是单位名或其简称。通常情况下，在一个城市内的街巷名以及小区名是唯一的，因此采用街巷名或小区名就可以大致锁定某一个范围地址，而采用“街巷名或小区名+门(楼)牌号”则可以精确到一个地点，采用“行政区域地名+标志物名”基本上也可以准确定位到一个位置。也就是说，当文本中所要表述的内容有门牌号时，就用“街巷名或小区名+门(楼)牌号”来锁定一个位置；当地名地址数据中含有标志物名时，使用“行政区域地名+标志物名”进行精确定位。按照上述的粒度范围的描述规则进行的地名地址数据结构化示例如下：
(1)广东省广州市天河区体育西路111号，地名地址数据结构化为：行政区域名+街巷名+门(楼)牌号；
(2)广东省广州市天河区体育西路建和中心，地名地址数据结构化为：行政区域名+街巷名+标志物名。
当遇到包含多个标志物名时，可以根据当前行政区域的粒度进行延伸，直到能够确定唯一的地点。例如，地名地址数据是“惠州市演达大道46号惠州学院”，在惠州市的应用中可以简化为“演达大道46号”完全不会产生任何歧义；而如果地名地址数据是“广州市工商银行”，此时可能会定位到多个标志物，得出的结果很难筛选，则需要将其延伸至街巷名或小区名等进行描述，即可准确定位到某一家工商银行。
S12：根据地名地址模型建立包含行政区域实体数据表、街巷实体数据表以及小区实体数据表的地理编码库；
本步骤中，行政区域实体数据表、街巷实体数据表以及小区实体数据表的表结构可根据应用场景进行定义，地理编码库的建立遵循以下原则：
唯一性原则：任何地理实体都只能被唯一标识；
透明性原则：从编码中可以识别结构间的从属关系；
灵活性原则：应当适应对象的发展变化；
标准性原则：编码规则适应国家标准体系，以便实现数据共享。
行政区域实体数据表、街巷实体数据表以及小区实体数据表结构分别如下表1、2、3所示：
表1 行政区域实体数据表
Figure PCTCN2020139759-appb-000001
表2 街巷实体数据表
Figure PCTCN2020139759-appb-000002
表3 小区实体数据表
Figure PCTCN2020139759-appb-000003
按照各表结构依次录入所有省、区县、街道、小区、标志物、门牌号和地理坐标进行地理编码库的构建。选取各数据表中的字段值作为地名地址词条，连同相应的地址级别记录在地址词典中。当是地址别名作为地名地址词条时，还需记录标准的名称，以便在地址分词时进行地址要素的规范化。
S13：基于地址词典，运用N-最短路径优化算法对不规则的地名地址数据进行分词及标准化处理，将地名地址数据切分为至少一个词组；
本步骤中，N-最短路径优化算法实现过程具体为：地址词典中记录有不同国家不同地区的所有地名地址(包括别名以及简称等)，首先根据地址词典将地名地址数据中有可能出现的地名词组顺序匹配出来，然后构建一个有向无环图，每个词组分别是有向无环图中的一个节点，且分别对应着一条被赋予边长(即权值，在非统计粗分模型中，假定所有的词都是对等的，为便于计算，将所有词对应边的边长均设为1)的边。有向无环图中起点到终点的所有路径中，求出每个节点到源节点的路径值，对应路径集合，作为每个节点的路径结果集。
例如，假设待分字串S＝c1c2……cn，其中ci(i＝1,2,…n)为单个的字，n为串的长度，n≥1。建立一个节点数为n+1的有向无环图G，各节点编号依次为V0，V1，V2，…，Vn。通过以下两种规则建立G所有可能的词边：
(1)相邻节点Vk-1,Vk之间建立有向边<Vk-1,Vk>，边的长度值为Lk，边对应的词默认为ck(k＝1,2,…n)；
(2)若w＝ci ci+1……cj是一个词，则节点Vi-1,Vj之间建立有向边<Vi-1,Vj>，边的长度值为Lw，边对应的词为w(0<i<j≤n)。
依照上述规则，使得待分字串S中包含的所有词与有向无环图G中的边一一对应，具体如图3所示，为本申请实施例的有向无环图结构示意图。N-最短路径优化算法的词语粗切问题即为求解有向无环图G的集合NSP。有向无环图结构的求解过程具体为：
设：Path(i,j)为所有从节点Vi到节点Vj的路径集合；Length(path)为路径path的长度，Length(path)值等于path中所有边的长度之和；LS为有向无环图G中所有从V0到Vn路径的长度集合；则有：
LS＝{len|len＝Length(path),path∈Path(0,n)}(1)
NLS为V0到Vn的N-最短路径长度集合，NSP为V0到Vn的N-最短路径集合；RS是最终求出的N-最短路径粗分结果集。NLS的定义为：|NLS|＝min(|LS|,N)；a∈LS-NLS,b∈ NLS→a<b NSP＝{path|path∈Path(0,n),Length(path)∈NLS}RS＝{w1w2…wm|wi是path的第i条边对应的词，i＝1,2,…,m,其中path∈NSP}，n为最短路径数。
以文本数据“他说的确实在理”构建有向无环图的求解为例，该文本数据的求解过程如图4所示。首先，采用贪心算法求得每个节点的局部最优解。记录每个节点处的最短路径值以及该节点的前驱，如果某节点包括两条以上相同长度的路径，则分别记录每条路径上该节点的前驱，该文本数据的前驱记录表如图5所示，其中，在(a)中，前驱(2,1)他、(3,1)他，长度分别为3和4，对应节点分别为012,0123；在(b)中，前驱(4,1)他和(4,2)他说，长度分别为4和5，对应节点分别为0123,01234；在(c)中，前驱(4,1)他、(5,1)他((4,2)他)和(5,2)他说，长度分别为4、5和6，对应节点分别为0123,01234,012345；在(d)中，前驱(6,1)他((5,1)他)、(6,2)他说((5,2))和(6,3)他说的，长度分别为5、6和7，对应节点分别为01234,012345,0123456。然后通过回溯算法，往前搜索更加优选的结果，最终求解出文本数据“他说的确实在理”的最优分词结果为“他|说|的|确实|在理|”。
基于上述，本发明通过采用N-最短路径分词算法对地名地址进行分词，不仅可以大大减少分词数量，还能尽量将所有可能的分词结果包含不流失，避免算法本身因素可能造成的舍弃正确结果的同时，又能尽可能的缩小搜索空间，提升分词效率。
S14：按照地名地址模型中的级别要素将切分后的至少一个词组转换为预定格式(计算机可识别)的字符串，然后将转换后的字符串与地理编码库中对应的地理坐标进行匹配；
S15：将字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
请参阅图6，是本申请第二实施例的地理编码方法的流程图。本申请第二实施例的地理编码方法包括以下步骤：
S20：对初始地名地址数据进行数据清洗；
本步骤中，由于用户端输入的地名地址等文本数据中可能会包含错别字或重复字，为了避免由于文本数据中字符串不一致、拼写错误等问题导致后续的字符串与地理坐标匹配错误，本发明实施例使用Trillum技术，采用语法分析和模糊匹配算法对地名地址进行数据清洗。
S21：将清洗后的地名地址数据结构化后，建立地名地址模型；
本步骤中，不同国家或地区对地名地址的表述方式具有不同粒度范围规则，可以把地名地址看成一种在层次上可伸缩的地名地址模型。
S22：根据地名地址模型建立包含地名数据表、建筑物数据表以及门(楼)牌数据表的地理编码库；
本步骤中，地名数据表、建筑物数据表以及门(楼)牌数据表的表结构可根据应用场景进行定义，按照各表结构依次录入所有省、区县、街道、小区、标志物、门牌号和地理坐标进行地理编码库的构建。
S23：基于地址词典，采用动态删除算法与N-最短路径分词算法相结合的N-最短路径改进分词算法对不规则的地名地址数据进行分词及标准化处理，将地名地址数据切分为至少一个词组；
本步骤中，为了进一步提高分词效率，本发明提出将动态删除算法与N-最短路径分词算法相结合的N-最短路径改进分词算法，其中，动态删除算法的基本思想是：构建最短路径更新队列，用于存放被删除的节点的子节点们；而在原来的最短路径树中删除应当被删除的节点以及所有的子节点；选择最接近根节点的节点用于在队列中更新，并且不再将更新过的节点插入队列中以减少节点更新的次数。N-最短路径改进分词算法如图7所示，其求解过程具体为：
第一步：首先基于N-最短路径分词算法，以词(或字)为节点构造有向无环图G；其中，有向无环图构造过程与第一实施例相同，本实施例将不再赘述；
第二步：计算从开始节点到结束节点的最短路径为Lj＝1，如果j小于最短路径数(即j＜n)并且存在其他候选路径，则更新当前路径L为Lj，反之结束；其中，候选路径是指不同的组词切割方式产生的不同路径。
其中，Lj用于存储最短路径，其中的j是一个动态变量，j的初始值可根据整个字符串的长度进行设定，如“他说的的确是在理”，j的初始值为8，当完成词语组合，删除组合间的节点，再更新j值，随着句子分割的持续进行，j值会越来越小，直到句子无法切割为止。
第三步：从当前路径中的第一个节点开始，将入度大于1的第一个节点删除，并将被删除节点记为Hm，判断Hm的子孙节点是否在集合E中，如果在集合E中，则计算从开始节点V0到Hm的最短路径，并将该最短路径的结束节点记为H’m；如果不在集合E中，则从有向无环图G中删除节点Hm及其所有子孙节点；其中，集合E为V0到Vn的N-最短路径集合(即NSP)，此处用于判断被删除节点是否在最短路径中。Hm与H’m在每一个循环当中均代表结束节点，H’m会作为下一次循环的结束标记。
第四步：重复上述过程，直到m≮n，更新当前路径，求得开始节点V0到所有节点H’m的最短路径j＝j+1。其中，n为删除节点后的最短路径数，m为j循环构造后的最短路径，在每一次循环当中，m的值为m＝j+1，且在进入下一个循环后，Hm会随m值更新，而H’m不会改变。
在上述求解过程中，为了避免影响搜索效率和准确性，对于保留的前j个最优路径，j的值应当适中，不能过大也不能过小。
S24：按照地名地址模型中的级别要素将切分后的至少一个词组转换为预定格式(计算机可识别)的字符串，然后将转换后的字符串与地理编码库中对应的地理坐标进行匹配；
S25：将字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
为了验证本申请实施例的可行性和有效性，通过借助ArcGIS api for Javascript平台对本方案进行了实验，并与传统算法的精确度进行了对比，对比结果如表1所示：
表1 算法精确度对比
Figure PCTCN2020139759-appb-000004
Figure PCTCN2020139759-appb-000005
实验结果表明，采用本申请实施例的地理坐标匹配正确率超过了传统算法，且分词速度加快了两倍以上。
基于上述，本申请实施例的地理编码方法运用N-最短路径优化算法对地名地址进行分词及标准化处理，根据标准化处理结果对地名地址进行切分后，按照地名地址模型中的级别要素将切分后的地名地址转换成计算机能够识别的字符串，最后将字符串与地理编码库中相应的地理坐标进行匹配，根据匹配结果为地名地址赋予标准地理坐标。本申请通过在算法中加入辅助的语法与语义规则，改进逐词遍历的弊端，增加了实用性，秉承了全切分思想的优势，既能够尽可能减少切分词组的数量，同时又能包含所有需要被保留的结果，能有效避免资源浪费，加大搜索效率。
请参阅图8，是本申请实施例的地理编码系统的结构示意图。本申请实施例的地理编码系统41包括：
数据清洗模块41：用于对初始地名地址数据进行数据清洗；由于用户端输入的地名地址等文本数据中可能会包含错别字或重复字，为了避免由于文本数据中字符串不一致、拼写错误等问题导致后续的字符串与地理坐标匹配错误，本发明实施例使用Trillum技术，采用语法分析和模糊匹配算法对地名地址进行数据清洗。
地名地址模型构建模块42：用于将清洗后的地名地址数据结构化后，建立地名地址模型；其中，不同国家或地区对地名地址的表述方式具有不同粒度范围规则，可以把地名地址看成一种在层次上可伸缩的地名地址模型。
地理编码库构建模块43：用于根据地名地址模型建立包含地名数据表、建筑物数据表以及门(楼)牌数据表的地理编码库；其中，地名数据表、建筑物数据表以及门(楼)牌数据表的表结构可根据应用场景进行定义，按照各表结构依次录入所有省、区县、街道、小区、标志物、门牌号和地理坐标进行地理编码库的构建。
分词及标准化处理模块44：用于基于地址词典，运用N-最短路径优化算法对不规则的地名地址数据进行分词及标准化处理，将地名地址数据切分为至少一个词组；其中，N-最短路径优化算法实现过程具体为：地址词典中记录有不同国家不同地区的所有地名地址(包括别名以及简称等)，首先根据地址词典将地名地址数据中有可能出现的地名词组顺序匹配出来，然后构建一个有向无环图，每个词组分别是有向无环图中的一个节点，且分别对应着一条被赋予边长(即权值，在非统计粗分模型中，假定所有的词都是对等的，为便于计算，将所有词对应边的边长均设为1)的边。有向无环图中起点到终点的所有路径中，求出每个节点到源节点的路径值，对应路径集合，作为每个节点的路径结果集。
例如，假设待分字串S＝c1c2……cn，其中ci(i＝1,2,…n)为单个的字，n为串的长度，n≥1。建立一个节点数为n+1的有向无环图G，各节点编号依次为V0，V1，V2，…，Vn。通过以下两种规则建立G所有可能的词边：
(1)相邻节点Vk-1,Vk之间建立有向边<Vk-1,Vk>，边的长度值为Lk，边对应的词默认为ck(k＝1,2,…n)；
(2)若w＝ci ci+1……cj是一个词，则节点Vi-1,Vj之间建立有向边<Vi-1,Vj>，边的长度值为Lw，边对应的词为w(0<i<j≤n)。
依照上述规则，使得待分字串S中包含的所有词与有向无环图G中的边一一对应，具体如图3所示，为本申请实施例的有向无环图结构示意图。N-最短路径优化算法的词语粗切问题即为求解有向无环图G的集合NSP。有向无环图结构的求解过程具体为：
设：Path(i,j)为所有从节点Vi到节点Vj的路径集合；Length(path)为路径path的长度，Length(path)值等于path中所有边的长度之和；LS为有向无环图G中所有从V0到Vn路径的长度集合；则有：
LS＝{len|len＝Length(path),path∈Path(0,n)}  (1)
NLS为V0到Vn的N-最短路径长度集合；NSP为V0到Vn的N-最短路径集合；RS是最终求出的N-最短路径粗分结果集。NLS的定义为：|NLS|＝min(|LS|,N)；a∈LS-NLS,b∈NLS→a<b NSP＝{path|path∈Path(0,n),Length(path)∈NLS}RS＝{w1w2…wm|wi是path的第i条边对应的词，i＝1,2,…,m,其中path∈NSP}。
以文本数据“他说的确实在理”构建有向无环图的求解为例，求解过程如图4所示。首先，采用贪心算法求得每个节点的局部最优解。记录每个节点处的最短路径值以及该节点的前驱，如果某节点包括两条以上相同长度的路径，则分别记录每条路径上该节点的前驱(前驱记录表如图4所示)，然后通过回溯算法，往前搜索更加优选的结果，最终求解出文本数据“他说的确实在理”的最优分词结果为“他|说|的|确实|在理|”。
基于上述，本发明通过采用N-最短路径分词算法对地名地址进行分词，不仅可以大大减少分词数量，还能尽量将所有可能的分词结果包含不流失，避免算法本身因素可能造成的舍弃正确结果的同时，又能尽可能的缩小搜索空间，提升分词效率。
在本申请另一实施例中，为了进一步提高分词效率，分词及标准化处理模块44采用将动态删除算法与N-最短路径分词算法相结合的N-最短路径改进分词算法进行分词即标准化处理，具体为：
第一步：首先基于N-最短路径分词算法，以词为节点构造有向无环图G；
第二步：计算从开始节点到结束节点的最短路径为Lj＝1，如果j小于最短路径数并且存在其他候选路径，则更新当前路径L为Lj，反之结束；
第三步：从当前路径中的第一个节点开始，将入度大于1的第一个节点删除，并将被删除节点记为Hm，判断Hm的子孙节点是否在集合E中，如果在集合E中，则计算从开始节点V0到Hm的最短路径，并将该最短路径的结束节点记为H’m；如果不在集合E中，则从有向无环图G中删除节点Hm及其所有子孙节点；
第四步：重复上述过程，直到m≮n，更新当前路径，求得开始节点V0到所有节点H’m的最短路径j＝j+1。
坐标匹配模块45：用于按照地名地址模型中的级别要素将至少一个词组转换为预定格式(计算机可识别)的字符串，然后将转换后的字符串与地理编码库中对应的地理坐标进行匹配，将字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
请参阅图9，为本申请实施例的终端结构示意图。该终端50包括处理器51、与处理器51耦接的存储器52。
存储器52存储有用于实现上述地理编码方法的程序指令。
处理器51用于执行存储器52存储的程序指令以控制地理编码。
其中，处理器51还可以称为CPU(Central Processing Unit，中央处理单元)。处理器51可能是一种集成电路芯片，具有信号的处理能力。处理器51还可以是通用处理器、数字信号处理器(DSP)、专用集成电路(ASIC)、现成可编程门阵列(FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
请参阅图10，为本申请实施例的存储介质的结构示意图。本申请实施例的存储介质存储有能够实现上述所有方法的程序文件61，其中，该程序文件61可以以软件产品的形式存储在上述存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本发明各个实施方式方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质，或者是计算机、服务器、手机、平板等终端设备。
对所公开的实施例的上述说明，使本领域专业技术人员能够实现或使用本申请。对这些实施例的多种修改对本领域的专业技术人员来说将是显而易见的，本申请中所定义的一般原理可以在不脱离本申请的精神或范围的情况下，在其它实施例中实现。因此，本申请将不会被限制于本申请所示的这些实施例，而是要符合与本申请所公开的原理和新颖特点相一致的最宽的范围。

Claims (10)

  1. 一种地理编码方法，其特征在于，包括：
    根据地名地址数据建立地名地址模型；
    根据所述地名地址模型建立地理编码库，所述地理编码库包括行政区域实体数据表、街巷实体数据表以及小区实体数据表；
    基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理，将所述地名地址数据切分为至少一个词组；
    按照所述地名地址模型中的级别要素将所述至少一个词组转换成预定格式的字符串，将所述字符串与所述地理编码库中对应的地理坐标进行匹配，将所述字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
  2. 根据权利要求1所述的地理编码方法，其特征在于，所述根据地名地址数据建立地名地址模型前还包括：
    对所述地名地址数据进行数据清洗。
  3. 根据权利要求1所述的地理编码方法，其特征在于，所述根据所述地名地址模型建立地理编码库包括：
    定义所述行政区域实体数据表、街巷实体数据表以及小区实体数据表的表结构，按照所述表结构依次录入省、区县、街道、小区、标志物、门牌号和地理坐标进行所述地理编码库的构建。
  4. 根据权利要求3所述的地理编码方法，其特征在于，所述基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理包括：
    根据所述地址词典顺序匹配出所述地名地址数据中的地名词组，并构建一个有向无环图，每个词组分别是所述有向无环图中的一个节点，且分别对应着一条被赋予边长的边；
    按照预设规则建立所述有向无环图所有可能的词边，使得所述地名地理数据中包含的所有词分别与所述有向无环图中的边一一对应，并求解所述有向无环图中从开始节点到结束节点的N-最短路径集合，根据所述N-最短路径集合对所述地名地址数据进行分词。
  5. 根据权利要求4所述的地理编码方法，其特征在于，假设所述地名地理数据S＝c1 c2……cn，其中ci(i＝1,2,…n)为单个的字，n为串的长度，n≥1，建立的有向无环图G的节点数为n+1，各节点编号依次为V0，V1，V2，…，Vn，所述建立有向无环图所有可能的词边的预设规则为：
    相邻节点Vk-1,Vk之间建立有向边<Vk-1,Vk>，边的长度值为Lk，边对应的词默认为ck(k＝1,2,…n)；
    若w＝ci ci+1……cj是一个词，则节点Vi-1,Vj之间建立有向边<Vi-1,Vj>，边的长度值为Lw，边对应的词为w(0<i<j≤n)。
  6. 根据权利要求5所述的地理编码方法，其特征在于，所述求解所述有向无环图中从开始节点到结束节点的N-最短路径集合包括：
    假设Path(i,j)为所有从节点Vi到节点Vj的路径集合；Length(path)为路径path的长度，Length(path)值等于path中所有边的长度之和；LS为有向无环图G中所有从V0到Vn路径的长度集合，则有：
    LS＝{len|len＝Length(path),path∈Path(0,n)}
    设NLS为V0到Vn的N-最短路径长度集合，NSP为V0到Vn的N-最短路径集合，RS是最终求出的N-最短路径粗分结果集，|NLS|＝min(|LS|,N)；a∈LS-NLS,b∈NLS→a<b，NSP＝{path|path∈Path(0,n),Length(path)∈NLS}RS＝{w1w2…wm|，wi是path的第i条边对应的词，i＝1,2,…,m,其中path∈NSP}，n为最短路径数。
  7. 根据权利要求6所述的地理编码方法，其特征在于，所述基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理还包括：
    计算从开始节点到结束节点的最短路径为Lj＝1，如果j小于最短路径数并且存在其他候选路径，则更新当前路径L为Lj，反之结束；
    从当前路径中的第一个节点开始，将入度大于1的第一个节点删除，并将被删除节点记为Hm，判断Hm的子孙节点是否在集合E中，如果在集合E中，则计算从开始节点到Hm的最短路径，并将该最短路径的结束节点记为H’m；如果不在集合E中，则从有向无环图G中删除节点Hm及其所有子孙节点；其中，集合E为V0到Vn的N-最短路径集合，Hm与H’m在每一个循环当中均代表结束节点，H’m作为下一次循环的结束标记；
    重复所述节点删除过程，直到m≮n，更新当前路径，求得开始节点V0到所有节点H’m的最短路径j＝j+1；n为删除节点后的最短路径数，m为j循环构造后的最短路径，在每一次循环当中，m的值为m＝j+1。
  8. 一种地理编码系统，其特征在于，包括：
    地名地址模型构建模块：用于根据地名地址数据建立地名地址模型；
    地理编码库构建模块：用于根据所述地名地址模型建立地理编码库，所述地理编码库包括行政区域实体数据表、街巷实体数据表以及小区实体数据表；
    分词及标准化处理模块：用于基于地址词典，运用N-最短路径优化算法对所述地名地址数据进行分词及标准化处理，将所述地名地址数据切分为至少一个词组；
    坐标匹配模块：用于按照所述地名地址模型中的级别要素将所述至少一个词组转换成预定格式的字符串，将所述字符串与所述地理编码库中对应的地理坐标进行匹配，将所述字符串匹配到的地理坐标作为对应地名地址的标准地理坐标。
  9. 一种终端，其特征在于，所述终端包括处理器、与所述处理器耦接的存储器，其中，
    所述存储器存储有用于实现权利要求1-7任一项所述的地理编码方法的程序指令；
    所述处理器用于执行所述存储器存储的所述程序指令以控制地理编码。
  10. 一种存储介质，其特征在于，存储有处理器可运行的程序指令，所述程序指令用于执行权利要求1至7任一项所述地理编码方法。
PCT/CN2020/139759 2020-11-05 2020-12-26 一种地理编码方法、系统、终端以及存储介质 Ceased WO2022095256A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011222303.X 2020-11-05
CN202011222303.XA CN112256817A (zh) 2020-11-05 2020-11-05 一种地理编码方法、系统、终端以及存储介质

Publications (1)

Publication Number Publication Date
WO2022095256A1 true WO2022095256A1 (zh) 2022-05-12

Family

ID=74268299

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/139759 Ceased WO2022095256A1 (zh) 2020-11-05 2020-12-26 一种地理编码方法、系统、终端以及存储介质

Country Status (2)

Country Link
CN (1) CN112256817A (zh)
WO (1) WO2022095256A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115455018A (zh) * 2022-09-01 2022-12-09 浪潮软件科技有限公司 一种基于地理空间网格的地名地址可视化方法及系统
CN115809315A (zh) * 2022-11-24 2023-03-17 中科星图智慧科技安徽有限公司 一种地名地址标准化匹配算法
CN116910386A (zh) * 2023-09-14 2023-10-20 深圳市智慧城市科技发展集团有限公司 地址补全方法、终端设备及计算机可读存储介质
CN117874013A (zh) * 2023-12-23 2024-04-12 武汉科信云图信息技术有限公司 一种业务地名地址数据标准化治理效率提升方法及装置
CN119598385A (zh) * 2024-10-23 2025-03-11 国家基础地理信息中心 一种多源地名数据的处理方法及装置
CN119719245A (zh) * 2024-11-29 2025-03-28 天津市城市规划设计研究总院有限公司 多人协同采集专题地理信息数据的方法、设备以及产品
CN119739806A (zh) * 2025-03-05 2025-04-01 安徽青际科技有限公司 一种地区匹配方法、系统、电子设备及计算机存储介质
CN120832864A (zh) * 2025-09-17 2025-10-24 河海大学 一种基于地址分布密度的编码平台的编码结果优化方法
CN120850953A (zh) * 2025-09-24 2025-10-28 河海大学 基于道路约束多源网络地理编码平台的地理坐标优化方法

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112949260A (zh) * 2021-03-05 2021-06-11 浪潮云信息技术股份公司 一种加快非结构化企业地址转换为经纬度的方法
CN112699640B (zh) * 2021-03-23 2021-06-11 城云科技（中国）有限公司 基于PostgreSQL的地理编码方法和系统
CN113723681A (zh) * 2021-08-30 2021-11-30 平安国际智慧城市科技股份有限公司 路径选择方法、装置、电子设备及可读存储介质
CN114970518B (zh) * 2022-02-15 2022-12-16 北京青萌数海科技有限公司 一种修正地址数据的方法和装置
CN115062108B (zh) * 2022-06-10 2025-05-09 上海估家网络科技有限公司 一种获取标准化房产地址的方法
CN118820490A (zh) * 2024-06-25 2024-10-22 中国地质图书馆（中国地质调查局地学文献中心） 地质文献资源空间化处理方法及基于gis的文献空间搜索系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150127324A1 (en) * 2007-09-28 2015-05-07 Telogis, Inc. Natural language parsers to normalize addresses for geocoding
CN108763215A (zh) * 2018-05-30 2018-11-06 中智诚征信有限公司 一种基于地址分词的地址存储方法、装置及计算机设备
CN109145169A (zh) * 2018-07-26 2019-01-04 浙江省测绘科学技术研究院 一种基于统计分词的地址匹配方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150127324A1 (en) * 2007-09-28 2015-05-07 Telogis, Inc. Natural language parsers to normalize addresses for geocoding
CN108763215A (zh) * 2018-05-30 2018-11-06 中智诚征信有限公司 一种基于地址分词的地址存储方法、装置及计算机设备
CN109145169A (zh) * 2018-07-26 2019-01-04 浙江省测绘科学技术研究院 一种基于统计分词的地址匹配方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WU XIAO-QIAN, HU XUE-GANG: "Research on Chinese Word Segmentation Based on N-shortest Path", JOURNAL OF ANHUI UNIVERSITY OF SCIENCE AND TECHNOLOGY(NATURAL SCIENCE), vol. 34, no. 1, 1 March 2014 (2014-03-01), pages 72 - 75, XP055931102, ISSN: 1672-1098 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115455018A (zh) * 2022-09-01 2022-12-09 浪潮软件科技有限公司 一种基于地理空间网格的地名地址可视化方法及系统
CN115809315A (zh) * 2022-11-24 2023-03-17 中科星图智慧科技安徽有限公司 一种地名地址标准化匹配算法
CN116910386A (zh) * 2023-09-14 2023-10-20 深圳市智慧城市科技发展集团有限公司 地址补全方法、终端设备及计算机可读存储介质
CN116910386B (zh) * 2023-09-14 2024-02-02 深圳市智慧城市科技发展集团有限公司 地址补全方法、终端设备及计算机可读存储介质
CN117874013A (zh) * 2023-12-23 2024-04-12 武汉科信云图信息技术有限公司 一种业务地名地址数据标准化治理效率提升方法及装置
CN119598385A (zh) * 2024-10-23 2025-03-11 国家基础地理信息中心 一种多源地名数据的处理方法及装置
CN119719245A (zh) * 2024-11-29 2025-03-28 天津市城市规划设计研究总院有限公司 多人协同采集专题地理信息数据的方法、设备以及产品
CN119739806A (zh) * 2025-03-05 2025-04-01 安徽青际科技有限公司 一种地区匹配方法、系统、电子设备及计算机存储介质
CN120832864A (zh) * 2025-09-17 2025-10-24 河海大学 一种基于地址分布密度的编码平台的编码结果优化方法
CN120850953A (zh) * 2025-09-24 2025-10-28 河海大学 基于道路约束多源网络地理编码平台的地理坐标优化方法

Also Published As

Publication number Publication date
CN112256817A (zh) 2021-01-22

Similar Documents

Publication Publication Date Title
WO2022095256A1 (zh) 一种地理编码方法、系统、终端以及存储介质
CN108959244B (zh) 地址分词的方法和装置
CN110990520B (zh) 一种地址编码方法、装置、电子设备和存储介质
CN112528174B (zh) 基于知识图谱和多重匹配的地址修整补全方法及应用
CN113342912B (zh) 地理位置区域的编码方法、建立编码模型的方法及装置
US8949196B2 (en) Systems and methods for matching similar geographic objects
US20100250599A1 (en) Method and apparatus for integration of community-provided place data
WO2015027836A1 (zh) 一种地名实体识别的方法及系统
US9710485B2 (en) Density-based dynamic geohash
CN106462624A (zh) 基于图块的地理编码器
CN102147795A (zh) 兴趣点检索方法、兴趣点检索装置和导航系统
CN102169498A (zh) 一种地址模型的构建方法，以及地址匹配的方法和系统
CN111291099B (zh) 一种地址模糊匹配方法、系统及计算机设备
CN114595302A (zh) 空间要素的多层级空间关系构建方法、装置、介质及设备
CN106033460A (zh) 地址数据处理方法及装置
CN107368471B (zh) 一种网页文本中地名地址的提取方法
CN113139032B (zh) 地理位置的搜索方法、装置、电子设备及存储介质
CN116303854B (zh) 一种基于地址知识图谱的定位方法及装置
CN106777118B (zh) 一种基于模糊字典树的地理词汇快速抽取方法
CN116414823A (zh) 一种基于分词模型的地址定位方法和装置
CN106874384A (zh) 一种异构地址标准转换及匹配方法
CN115544197A (zh) 一种细粒度地址匹配方法及系统
Moura et al. Reference data enhancement for geographic information retrieval using linked data
CN112069824B (zh) 基于上下文概率和引证的地域识别方法、装置及介质
CN111966768B (zh) 地址描述方法、装置、电子设备和存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20960713

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20960713

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20960713

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 12/12/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20960713

Country of ref document: EP

Kind code of ref document: A1