An interactive geocoding parsing system based on a multi-source geocoder

Abstract

The present invention discloses an interactive geocoding parsing system based on a multi-source geocoder, which includes a structured place name database, a geocoding module, an overlapping point checking module, an artificial verification and fixed-point module, and a result output module. The present invention utilizes multiple geocoders for geocoding and optimizes the coding results from two directions: the coding quality of the geocoders and spatial clustering. It not only fully considers the coding quality of each geocoder and performs quality control on the coding results, but also performs clustering and fusion on the coding results based on the objective spatial position relationship. The artificial verification and fixed-point module of the present invention can further solve problems such as overlapping coding results of different address names in the synchronous coding of multiple geocoders, large differences in the results of the same address name in different geocoders, and the inability of some encoders to parse coordinate results due to different address databases through a human-computer interaction method.

Description

Technical Field

[0001] The present invention belongs to the field of geographic information technology, and specifically relates to an interactive geocoding and parsing system based on a multi-source geocoder. Background Art

[0002] Geocoding, also known as address coding, is the process of converting the description of an address text into spatial coordinates. With the rapid development of Internet maps, online geocoding services have become an important means for ordinary users to obtain spatial location information. Ordinary users can use a structured place name database and combine it with online geocoding services to quickly obtain a large amount of POI (Point of Interest) coordinate information, and then establish their own required geographic coordinate databases.

[0003] Although online geocoding services have the advantages of high efficiency and convenience, there may still be a large gap between the coding results and the actual location pointed to by the address text, and the location accuracy is not high. Moreover, the quality of geocoding services provided by major online map platforms (such as Baidu Map, Gaode Map, Tencent Map, etc.) varies, and the parsed coordinate information may vary greatly. The parsing results of a single geocoder may not be accurate. In addition, due to the different address databases used by major map platforms, using only a single geocoder for parsing may result in coding failures due to incomplete address database data.

[0004] At the present stage, relevant technical personnel have used the reverse geocoding method to clean and mark the coding results, and used the system clustering method to optimize the coding results of different geocoders. However, they did not start from the geocoder coding results, lacked filtering and screening of coding quality, did not fully consider the spatial location relationship of each coding result. In addition, this integrated coding method lacks a certain degree of human-computer interaction, the coding results are uncontrollable, and the flexibility is poor.

[0005] When geocoding a large number of structured addresses, relying only on the geocoding method with specified rules may result in problems such as overlapping coding results for different address names, large differences in the results of the same address name in different geocoders (that is, the spatial coordinates parsed by different geocoders are very different), and some encoders not parsing out coordinate results due to different address databases. All of the above problems should be regarded as doubtful coding results and require further manual verification and correction. Summary of the Invention

[0006] The object of the present invention is to overcome the deficiencies of the prior art and propose an interactive geocoding and parsing system based on a multi-source geocoder, so as to solve the limitations in the prior art that the results are not necessarily accurate when performing geocoordinate parsing based on a single geocoder, as well as problems such as overlapping coding results of different address names, large differences in coding results of the same address name in different geocoders, and the inability of some encoders to parse coordinate results due to different address libraries when performing geocoding on a large number of structured addresses.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] An interactive geocoding and parsing system based on a multi-source geocoder, including a structured place name library, a geocoding module, an overlapping point checking module, an artificial verification and positioning module, and a result output module. Among them, the place name data in the structured place name library will be used as a data source and input into the geocoding module for preliminary geocoding and parsing, and doubtful points and normal points will be output; for the normal points, further overlapping checks will be performed through the overlapping point checking module, and overlapping points and non-overlapping points will be output; the overlapping points output by the overlapping point checking module and the doubtful points output by the geocoding module are regarded as doubtful results, and the artificial verification and positioning module is used for interactive point position correction. Finally, the corrected results and the non-overlapping points output by the overlapping point checking module will be input into the result output module for output and storage.

[0009] In the geocoding module, first, a multi-source online map service geocoding module is used for geocoding and parsing. For the parsing results, integration optimization will be performed respectively through a geocoding optimization module based on coding quality and encoder level and a geocoding optimization module based on spatial clustering. The doubtful results after optimization will be marked as doubtful by the doubtful marking module and output as doubtful points, while the normal results will be output as normal points by the result preliminary screening module.

[0010] The multi-source online map service geocoding module can use any number of publicly available online geocoding service interfaces as geocoders.

[0011] The geocoding optimization module based on coding quality and encoder level consists of a coding result marking module and a parsing coordinate quality control module. Among them, the coding result marking module performs encoder level marking, confidence level marking, and parsing scale marking on the geocoding results of the multi-source online map service geocoding module, and outputs the marked results; the parsing coordinate quality control module performs quality control on the parsing scale and parsing confidence of the marked results, and the remaining results after filtering the results that do not meet the requirements are qualified results; finally, the geocoding optimization module based on coding quality and encoder level will select the results with a higher encoder level from the qualified results for output. If there are no qualified results, 0 will be returned.

[0012] The geocoding optimization module based on spatial clustering uses the DBSCAN clustering algorithm as the spatial clustering analysis method. The purpose of clustering analysis is to remove noise points in the parsing results and integrate and optimize multiple coding results from the perspective of spatial relationships.

[0013] The present invention includes five major modules: a structured place name database, a geocoding module, an overlapping point checking module, an artificial verification and positioning module, and a result output module.

[0014] Among them, the structured place name database is the data source and contains specified structured standard place name data.

[0015] The geocoding module is the core module of the present invention, which includes a multi-source online map service geocoding part, a coding result optimization part, and a parsing result comparison and marking part. The multi-source online map service geocoding part can perform geocoding on the place names in the structured place name database using multiple online map service geocoders to obtain preliminary coding results. The coding result optimization part optimizes the multi-source geocoding results from two directions: the coding quality of the geocoder and spatial clustering, and returns appropriate candidate coordinates. Finally, the candidate coordinates returned by the two methods are compared to obtain preliminary normal point and doubtful point data.

[0016] The function of the overlapping point checking module is to further check the overlapping nature of the normal points in the geocoding module. Two different place name points with a distance less than a certain threshold are regarded as overlapping points, and the overlapping points need to be further extracted for artificial verification and positioning. Using this method can effectively solve the problem of overlapping coding results of different addresses.

[0017] The function of the artificial verification and positioning module is to check for omissions and deficiencies in the doubtful points and overlapping points, complete the final manual correction operation, and improve the accuracy of the final point determination. Finally, the result output module is responsible for summarizing all the final results for output and storage.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] The present invention makes full use of multiple geocoders for geocoding, effectively breaking through the coding limitations of a single geocoder. Moreover, two different methods are adopted to integrate and optimize multiple coding results. It not only fully considers the coding quality of each geocoder, performs quality control on the coding results using the parsing scale and parsing confidence, but also performs clustering fusion on the parsing results based on the objective spatial position relationship. At the same time, the invention also introduces an artificial interaction module to further optimize the coding results under the condition of artificial intervention, improving the controllability of the coding results. For problems such as overlapping coding results for different address names, large differences in the results of the same address name in different geocoders, and the situation where some encoders fail to parse coordinate results due to different address databases, the artificial verification and fixed-point module can be used for further optimization and adjustment to ensure the integrity and reliability of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the flowchart of the operation of the present invention;

[0021] Figure 2 is the schematic diagram of the geocoding module in the present invention;

[0022] Figure 3 is Figure 2 the schematic diagram of a specific embodiment of the geocoding optimization module based on coding quality and encoder level in

[0023] Figure 4 is Figure 2 the schematic diagram of a specific embodiment of the geocoding optimization module based on spatial clustering in

[0024] Figure 5 is the schematic diagram of the operation interface of the artificial verification and fixed-point module in the embodiment of the present invention;

[0025] Figure 6 is the schematic diagram of the operation interface of the artificial verification and fixed-point module when only one coordinate point is parsed in the embodiment of the present invention;

[0026] Figure 7 is the schematic diagram of the overlapping points detected by the overlapping point inspection module in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] Embodiment

[0029] As Figure 1 shown, the embodiment of the present invention mainly consists of five major parts, namely a structured place name library a, a geographical coding module b, an overlapping point checking module c, an artificial verification and positioning module d, and a result output module e. Among them, the structured place name library a is composed of the full names of administrative villages at the fifth level in each province of our country. Each full name of an administrative village at the fifth level should be composed according to the following specifications:

[0030] Province name + Prefecture-level city name + County / district name + Township name + Administrative village name

[0031] Specific example: "xx Village, xx Town, xx District, xx City, xx Province".

[0032] Specifically, in the embodiment of the present invention, the place name data in the place name library a is first transmitted to the geographical coding module b for preliminary geographical coding analysis and the doubtful points and normal points are output. Among them, the doubtful points will be further manually corrected in the artificial verification and positioning module d; while the normal points need to be checked for overlapping points c. In principle, each place name point should be separately parsed into a unique decimal longitude and latitude coordinate. In this embodiment, place name points with a distance between two or more longitude and latitude coordinates less than 0.0001 degrees (about 11 meters) are regarded as a group of overlapping points, and the rest are regarded as non-overlapping points. For overlapping points, further manual correction is required in the artificial verification and positioning module d; for non-overlapping points, they can be directly transmitted to the result output module e.

[0033] Specifically, in order to simplify the calculation in this embodiment, the Euclidean distance is uniformly used as the distance calculation formula, and the formula form is as follows:

[0034]

[0035] Where lon α is the longitude of the coordinate of point α, lat α is the latitude of the coordinate of point α, lon β is the longitude of the coordinate of point β, lat β is the latitude of the coordinate of point β.

[0036] Finally, the results output by the artificial verification and positioning module d and the non-overlapping points will be saved as the final positioning results in the result output module e.

[0037] Furthermore, as Figure 2As shown in the figure, in the geocoding module b, multi-source online map service geocoding analysis will be performed on each full structured address a1 in the structured place name database a. Specifically, the multi-source online map service geocoding module b1 in this embodiment uses Tianditu API, Baidu Map API, Tencent Map API, and Amap API as geocoders respectively. After being parsed by the multi-source online map service geocoding module b1, if the number of output coding results is 0, it is regarded as the failure of this place name parsing, and the result is directly passed to the suspicious marking module b4 for suspicious marking; if the number of output coding results is not 0, then the geocoding optimization module b2 based on coding quality and encoder level (abbreviated as method one) and the geocoding optimization module b3 based on spatial clustering (abbreviated as method two) will be used for further geocoding optimization processing. After the geocoding optimization processing, if the return values of the two optimization methods are 0, that is, only one parsing result or no parsing results are obtained by the two optimization methods, it is regarded as the result being suspicious and needs to be passed to the suspicious marking module b4 for suspicious marking; if the return results of the two optimization methods are not 0, then the similarity of the two parsing results is compared, that is, whether the spatial distance between the two parsing coordinates is less than the specified distance threshold. Specifically, this embodiment uses 0.0001 degrees (about 11 meters) as the distance threshold. If it is greater than the threshold, it is regarded as the parsing results of the two methods being unqualified and needs to be passed to the suspicious marking module b4 for suspicious marking; if it is less than the threshold, the average value of the two is taken as the output result by the result preliminary screening module b5 and marked as a normal point.

[0038] Furthermore, as Figure 3Method 1 is shown as follows. The encoding result marking module b201 and the parsed coordinate quality control module b202 are used to screen and filter multiple encoding results, and finally the optimal result is selected according to the specified encoder level. Specifically, the encoding result marking module b201 mainly marks attributes such as the encoder level, confidence level, and parsing scale based on the geocoding result, so that the qualified coordinate points can be preferably selected as candidate points by the parsed coordinate quality control module b202 according to the marking result. More specifically, Method 1 adopts a geocoder priority decision scheme. Since "Tianditu" is a comprehensive geographic information service website built by the National Administration of Surveying, Mapping and Geoinformation, integrating the public geographic information service resources from the surveying and mapping geographic information departments at the national, provincial, and city (county) levels, as well as relevant government departments, enterprises, institutions, social organizations, and the public, the online comprehensive geographic information service provided by this platform to various users is more authoritative and the data quality is higher. Therefore, in this embodiment, the Tianditu api is used as the main encoder, and the encoder level is set to level 4. The other three map apis are used as auxiliary encoders, and the levels are as follows: Baidu Map - level 3, Tencent Map - level 2, Amap - level 1. The coordinate results parsed by the multi-source online map service geocoding module b1 will be marked by the encoder level marking module b2011.

[0039] More specifically, the confidence level marking module b2012 determines the parsing confidence according to the attributes related to the confidence level in the return values of each encoder. For example, the "score" attribute in the geocoding result of the Tianditu api represents the reliability of the parsing result, and the value range is 0 - 100. The larger the value, the higher the reliability. To simplify and unify the evaluation standard of the confidence level, in this embodiment, a rounding mapping method is adopted to uniformly set the confidence level to ten levels from level 1 to level 10. In particular, in this embodiment, since the encoding results of the Amap api lack the attributes related to the confidence level, their confidence levels are all set to level 6, that is, it is default that the parsing results of the Amap are all reliable.

[0040] More specifically, in this embodiment, the parsing scale marking module b2013 determines the parsing scale of the encoding result according to the attributes related to the parsing scale in the return values of each encoder. For example, each encoder's parsing result has a level attribute representing the parsing scale. To simplify and unify the evaluation standard, in this embodiment, the place names are roughly divided into five scale levels, and the specific classification is as follows:

[0041] {Provincial level: level 1, Municipal level: level 2, County level: level 3, Town level: level 4, Village level and others: level 5}

[0042] Specifically, since the level attribute in the return value of the Tianditu API has a low level of refinement and cannot distinguish the scale differences between townships and villages, the parsing scale for townships and below is determined by comparing the parsing results of the previous-level place names. The scale above townships is still determined using the level attribute. The specific method is as follows:

[0043] (1) Parse village-level place names simultaneously: Province name + City name + County name + Township name + Village name

[0044] Township-level place names: Province name + City name + County name + Township name

[0045] (2) Compare the parsing results of village-level place names with those of township-level place names. If the parsing results of village-level place names are exactly the same as those of township-level place names, it means that the positioning coordinates of this village-level place name are not encoded at the village level but are parsed to the positioning coordinates at the township level. The parsing level needs to be set to {township level: 4th level}, otherwise it is parsed as {village level and others: 5th level}.

[0046] Furthermore, as Figure 3 shown, the parsing coordinate quality control module b202 in the embodiment of the present invention first uses the parsing scale filtering module b2021 to filter the marked parsing results by parsing scale. The user can filter the parsing results of non-desired scales according to the set scale level. In this embodiment, the parsing scale results other than "village level and others" are mainly filtered, that is, the points with a scale level <5 are regarded as unqualified. If there are remaining coordinates after parsing scale filtering, they are passed to the parsing confidence filtering module b2022 for further filtering operations; if all coordinates are filtered, it means that all parsing coordinates are unqualified, and 0 is returned. The parsing confidence filtering module b2022 will further filter the remaining coordinates according to the specified confidence level threshold. In this embodiment, the confidence level threshold is set to 6th level, and the remaining coordinates after filtering are regarded as qualified coordinates. Finally, the coordinate with the highest encoder level is selected from the remaining qualified coordinates as the final coordinate; if all coordinates are filtered after confidence filtering, it also means that all parsing coordinates are unqualified, and 0 is returned.

[0047] Furthermore, as Figure 4As shown, method 2 starts from the perspective of spatial relationships and uses the spatial clustering analysis module b301 to perform spatial clustering analysis on the parsing results of the multi-source online map service geocoding module b1, so as to further eliminate noise points and aggregate and optimize the encoding results of multiple geocoders. Specifically, in this embodiment, the DBSCAN clustering algorithm is used as the main clustering method. The DBSCAN clustering algorithm is a density clustering algorithm that can divide areas with sufficiently high density into clusters and can find clusters of any shape in noisy spatial data, which helps to further identify and filter noise data. More specifically, in this embodiment, the minimum number of clustering points of DBSCAN clustering is set to 2, and the neighborhood radius is set to 0.001° (about 111 meters).

[0048] More specifically, after clustering, Figure 4 As shown, three clustering results can be roughly obtained:

[0049] (1) Clustered into class 0, this means that the distance between all point coordinates exceeds the neighborhood radius, that is, clustering fails, and all points are regarded as noise points, represented by the "x" symbol. The coordinate points are discretely distributed and need to be marked as doubtful, and 0 is returned.

[0050] (2) Clustering into multiple categories. In this case, the coordinates of the four points may be clustered into two or more categories. In this embodiment, the coordinates of the four points can be clustered into two categories at most, with "*" and "o" representing one of the categories respectively. Multiple clustering results also mean that the coordinate point distribution is discrete, which needs to be marked as doubtful and returned as 0.

[0051] (3) Clustered into 1 category, this means that the spatial distance between at least two analytical coordinate points is relatively close, and only one cluster is formed. It can be preliminarily determined that the target point is also located in the cluster area, and the coordinates of the center point of the cluster are returned as candidate coordinates.

[0052] Furthermore, if Figure 5 As shown, the manual verification module d is mainly used to perform manual verification operations on doubtful points. The embodiment in the figure is a doubtful phenomenon caused when the spatial distance of the coordinates of method 1 and method 2 is greater than the threshold. The manual verification module d can be used to combine the high-resolution remote sensing image base map and place name annotations to further judge the two analysis results. More specifically, Figure 5 The example shown in the figure is the different spatial coordinate results of the point "Tongzhu Village, Huixian Town, Lingui District, Guilin City, Guangxi Zhuang Autonomous Region" analyzed by the two methods. The red icon is the result obtained by method one, and the yellow icon is the result obtained by method two. Through manual comparison, it can be found that the yellow icon obviously locates the location of "Muhuantou Village", while the red icon accurately locates the "Tongzhu Village Villagers Committee", so the red icon should be selected as the final positioning coordinate.

[0053] Furthermore, if Figure 6As shown, the interface shows the parsing result of "Xiaoxiao Village, Luojin Town, Yongfu County, Guilin City, Guangxi Zhuang Autonomous Region". At this time, only one return result is obtained for the two optimization methods, so the result is in doubt. The user can decide whether to select this point as the final result according to the actual situation, or select a suitable coordinate point in the image map manually as the final coordinate. Further, as Figure 7 shown, after being checked by the overlapping point checking module c, two points that are relatively close or overlapping among the normal points can be found. In the figure, "Tailan, Lancheng Township, Pingtan County, Fuzhou City, Fujian Province" and "Zhonghu Village, Lancheng Township, Pingtan County, Fuzhou City, Fujian Province" have "overlapped". Such points are regarded as overlapping points in the embodiments of the present invention. Similarly, the point coordinates need to be further corrected through manual verification. The manual verification and fixed-point module d fully reflects the human-computer interaction of the invention. By using the means of manual verification and fixed-point, the preliminary parsing result can be checked for omissions and deficiencies, and the accuracy of the final positioning can be improved.

[0054] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An interactive geocoding parsing system based on a multi-source geocoder, characterized in that, It includes a structured place name database (a), a geographic coding module (b), an overlapping point checking module (c), an artificial verification and fixed-point module (d), and a result output module (e). Among them, the place name data in the structured place name database (a) will be used as a data source and passed into the geographic coding module (b) for preliminary geographic coding analysis, and doubtful points and normal points will be output. For the normal points, further overlapping checks will be performed through the overlapping point checking module (c), and overlapping points and non-overlapping points will be output. The overlapping points output by the overlapping point checking module (c) and the doubtful points output by the geographic coding module (b) are regarded as doubtful results. The artificial verification and fixed-point module (d) is used for interactive point position correction. Finally, the corrected results and the non-overlapping points output by the overlapping point checking module (c) will be passed into the result output module (e) for output and storage. Among them: In the geographic coding module (b), first, the multi-source online map service geographic coding module (b1) is used for geographic coding analysis. For the analysis results, integration optimization will be performed respectively through the geographic coding optimization module (b2) based on coding quality and encoder level and the geographic coding optimization module (b3) based on spatial clustering. The doubtful results after optimization will be marked as doubtful by the doubtful marking module (b4) and output as doubtful points, while the normal results will be output as normal points by the result preliminary screening module (b5). The geographic coding optimization module (b2) based on coding quality and encoder level consists of a coding result marking module (b201) and an analysis coordinate quality control module (b202). Among them, the coding result marking module (b201) marks the encoder level, confidence level, and analysis scale of the geographic coding results of the multi-source online map service geographic coding module (b1), and outputs the marked results. The analysis coordinate quality control module (b202) performs quality control on the analysis scale and analysis confidence of the marked results, and the remaining results after filtering the results that do not meet the requirements are qualified results. Finally, the geographic coding optimization module (b2) based on coding quality and encoder level will select the results with a higher encoder level from the qualified results for output. If there are no qualified results, 0 will be returned.

2. The interactive geocoding parsing system based on a multi-source geocoder according to claim 1, wherein The multi-source online map service geographic coding module (b1) uses any number of publicly available online geographic coding service interfaces as geographic encoders.

3. An interactive geocoding parsing system based on a multi-source geocoder according to claim 1, wherein, The geographic coding optimization module (b3) based on spatial clustering uses the DBSCAN clustering algorithm as the spatial clustering analysis method to remove the noise points in the analysis results and perform integration optimization on multiple coding results from the perspective of spatial relationships.

Images