WO2019165644A1

WO2019165644A1 - Address error correction method and terminal

Info

Publication number: WO2019165644A1
Application number: PCT/CN2018/077926
Authority: WO
Inventors: 李林贵; 吴卫东; 周涛
Original assignee: 福建联迪商用设备有限公司
Priority date: 2018-03-02
Filing date: 2018-03-02
Publication date: 2019-09-06
Also published as: CN108369582B; CN108369582A

Abstract

The present invention relates to the field of data processing and in particular relates to an address error correction method and a terminal. acquiring an address to be error corrected; identifying, based on a first trie tree, the province names corresponding to the address to be error corrected to obtain a primary name; the first trie tree is used to store the province names and city names; acquiring a second trie tree corresponding to the primary name; the second trie tree is used to store the city names, county names and district names corresponding to the current province names; identifying, based on the second trie tree, the county names or district names corresponding to the address to be error corrected to obtain a secondary name; acquiring a third trie tree corresponding to the secondary name; the third trie tree is used to store the village town names, village names and street names corresponding to the secondary name; and acquiring, based on the third trie tree, more than one candidate addresses corresponding to the address to be error corrected to obtain a candidate address set. The space occupied during address error correction is reduced.

Description

Address correction method and terminal

Technical field

The present invention relates to the field of data processing, and in particular, to an address error correction method and a terminal.

Background technique

The methods of post-processing to identify address information by OCR technology mainly include constructing vocabulary method, statistical language model, syntax tree, similar words, distance information and the like. More commonly used is the construction of vocabulary and statistical language models.

The statistical language model uses probability and statistics to obtain similar words and words or the relationship between words and words. According to the probability of occurrence of this relationship, the most likely result is obtained. The Markov model is commonly used. For example, given an address of "Lake x Changsha City", according to the statistical probability of the address, the conditional probability of "South" after the word "L" is N1, the conditional probability of "North" is M1; after the word "South" The conditional probability of “province” is N2, the conditional probability of “province” after “North” is M2, the probability of “Hunan Province” is N1*N2, and the probability of “Hubei Province” is M1*M2. According to the word "长" after the word "province", the probability of being "Hunan Province" is greater than that of "Hubei Province", and the address is "Changsha City, Hunan Province". According to the characteristics of the address, an address data can usually be divided into multiple words, and the relationship between words is greater than the relationship between words and words. Therefore, the statistical language model based on words is more suitable for address error correction. Using the word-based statistical language model for address error correction, generally by collecting address data, an address database training language model is constructed to obtain the conditional probability between different address names, which is saved as a parameter; then according to a certain word segmentation rule The address is divided into multiple words; finally, the search algorithm is used to find the optimal solution of the language model, that is, the address with the highest probability of occurrence.

However, the shortcoming of word-based statistical language models is the need to calculate the probability of occurrence of words, using the search algorithm to derive the final address. When training the statistical language model, the parameter space is huge, and a large corpus is needed. If the corpus data is insufficient, it is easy to have a conditional probability of 0, resulting in poor model effect. There are approximate place names in the address, which may not be distinguished according to statistical probability. If the order of the Markov model is increased, the parameter space will increase sharply.

The construction of the vocabulary method uses a certain data structure to save the classified words, query according to the vocabulary, and obtain possible words to correct the current wrong words. Data structures can be linear or tree-like. In general, linear structures are less efficient in time and space. Commonly used are tree structures, such as dictionary trees applied in search engines. The dictionary tree is constructed with the word common root nodes with the same prefix, such as add, and, andy stored as a tree structure as shown in Figure 1. Saving data as a dictionary tree can share nodes and reduce redundancy. However, due to the variety of Chinese characters, each node stores a Chinese character and a pointer. The resulting dictionary tree is very large and takes up a lot of space. When querying, go down from the root node, enter different branches, and finally connect all the nodes that have entered, and get the address.

However, the disadvantage of the dictionary tree is that the dictionary tree that constructs the address data is too large and takes up too much space.

Summary of the invention

The technical problem to be solved by the present invention is how to reduce the space occupied by the address error correction process.

In order to solve the above technical problems, the technical solution adopted by the present invention is:

The invention provides an address error correction method, comprising:

S1, obtaining an address to be corrected;

S2: Identify, according to the first dictionary tree, a province name corresponding to the to-be-corrected address, to obtain a first-level name; the first dictionary tree is configured to store a province name and a city name;

S3: Obtain a second dictionary tree corresponding to the first-level name; the second dictionary tree is configured to store a city name, a county name, and a district name corresponding to the current province name;

S4. Identify, according to the second dictionary tree, a county name or a zone name corresponding to the to-be-corrected address, and obtain a secondary name;

S5: Obtain a third dictionary tree corresponding to the second-level name; the third dictionary tree is configured to store a township name, a village name, and a street name corresponding to the second-level name;

S6. Acquire one or more candidate addresses corresponding to the to-be-corrected address according to the third dictionary tree, to obtain a candidate address set.

The present invention also provides an address correction terminal comprising one or more processors and a memory, the memory storing a program, and being configured to perform the following steps by the one or more processors:

S1, obtaining an address to be corrected;

S2: Identify, according to the first dictionary tree, a province name corresponding to the to-be-corrected address, and obtain a first-level name; the first dictionary tree is configured to store a province name and a city name;

The invention has the beneficial effects that: different from the prior art, when the error correction address is needed, a complete dictionary tree corresponding to the national address needs to be called, and the occupied space is large. The present invention classifies the national address according to the streets of provinces, cities, counties, and towns and villages. Save, in turn check the province information, city and county information and township and village information in the address to be corrected, and dynamically retrieve the dictionary tree corresponding to the next-level address according to each verification result, which greatly reduces the The memory space is occupied during address error correction and has high accuracy.

DRAWINGS

Figure 1 is a schematic diagram of a dictionary tree;

2 is a flow chart of a specific implementation manner of an address error correction method according to the present invention;

3 is a structural block diagram of a specific implementation manner of an address error correction terminal according to the present invention;

4 is a schematic diagram of a first dictionary tree;

Figure 5 is a schematic diagram of a second dictionary tree;

6 is a schematic diagram of a third dictionary tree;

7 is a schematic diagram of a dictionary tree corresponding to an address to be corrected;

Label description:

1, the processor; 2, memory.

Detailed ways

The most important technical idea of the invention is that the national address is stored hierarchically according to the province, city, county, township and village streets, and the province information, city and county information, and township and village information in the address to be corrected are sequentially checked, and according to each The verification result of the second time dynamically retrieves the dictionary tree corresponding to the next-level address, which reduces the occupation of the memory space in the address error correction process.

Please refer to Figure 2 to Figure 7,

As shown in FIG. 2, the present invention provides an address error correction method, including:

S1, obtaining an address to be corrected;

Further, the S2 is specifically:

When there is no province name adapted to the error correction address in the first dictionary tree, acquiring a city name adapted to the error correction address, obtaining a current city name; acquiring the current city name The corresponding province name gives the first-level name.

It can be seen from the above description that in the case that the province name of the address to be error-corrected is more serious, the name of the province corresponding to the address to be corrected can be confirmed by the city name, which is advantageous for improving the accuracy of error correction.

Further, it also includes:

A node in the first dictionary tree represents a province name or a city name;

A node in the second dictionary tree represents a city name, a county name or a zone name;

A node in the third dictionary tree represents one of a township name, a village name, or a street name.

It can be seen from the above description that the general province, city, and county names are less likely to be duplicated, and the whole word can be saved as one node, and the county level may be township, village, or street in the future, and the possibility of repeated occurrence is relatively large, sharing. The same prefix can effectively reduce redundancy and reduce the space required.

Further, the S5 is specifically:

Obtaining a dictionary tree corresponding to the second-level name to obtain a third dictionary tree;

Obtaining a character corresponding to the preset order after the second-level name from the to-be-corrected address, to obtain a current character;

And a third dictionary tree to be constructed according to the branch of the third dictionary tree adapted to the current character; the root node of the third dictionary tree is the second-level name.

It can be seen from the above description that by specifying a character at a specific location and selecting branch information adapted to a character at a specific location as a candidate address, the capacity of the third dictionary tree is reduced, that is, the occupation required to check the street address of the township village is reduced. Space.

Further, it also includes:

The character corresponding to the preset order is a first character after the second-level name and a fourth character after the second-level name.

It can be known from the above description that the first character after the second-level name is generally the first character of the town name, and the fourth character after the second-level name is generally the first character of the village name. Generally, the town and village after the county name can be screened. Can effectively reduce the dictionary tree nodes that need to be generated.

Further, after the S6, the method further includes:

S71. Obtain a candidate address from the candidate address set to obtain a current candidate address.

S72. Count the number of characters in the same position of the current candidate address and the same location of the to-be-corrected address, and obtain a matching number.

S73. Repeat performing the S71 to the S72 until the candidate address set is traversed;

S74. Obtain a candidate address having the largest matching number in the candidate address set, to obtain an optimal address.

S75. Update the to-be-corrected address according to the optimal address to obtain a correct address.

Further, the S75 is specifically:

If there are more than two consecutive characters in the optimal address that are not adapted to the address to be corrected, then:

Obtaining, from the optimal address, a character string located before two or more consecutive characters that are not adapted to the error correction address;

Updating the to-be-corrected address according to the string to obtain a correct address;

Otherwise, set the best address to the correct address.

As can be seen from the above description, the correct rate of selecting the address with the highest similarity to the error-correction address from the one or more candidate addresses as the correct address is improved.

Further, the S1 is specifically:

The address information in the identity card is identified by an optical character recognition technology to obtain the address to be corrected.

As shown in FIG. 3, the present invention also provides an address error correction terminal including one or more processors 1 and a memory 2, the memory 2 storing a program and configured to be configured by the one or more processors 1 Perform the following steps:

S1, obtaining an address to be corrected;

Further, the S2 is specifically:

Further, it also includes:

A node in the first dictionary tree represents a province name or a city name;

Further, the S5 is specifically:

Further, it also includes:

Further, after the S6, the method further includes:

Further, the S75 is specifically:

Otherwise, set the best address to the correct address.

Further, the S1 is specifically:

Embodiment 1 of the present invention is:

This embodiment provides an address error correction method, including:

S1: Obtain an address to be corrected.

Optionally, the address information in the identity card is identified by an optical character recognition technology to obtain the to-be-corrected address.

For example, the address to be corrected is "Hongshan, Hongshan, Gulou District, Fuchuan City, Fujian Province".

S2: Identify, by the first dictionary tree, a province name corresponding to the to-be-corrected address, to obtain a first-level name; the first dictionary tree is configured to store a province name and a city name. Specifically:

As shown in FIG. 4, a node in the first dictionary tree represents a province name or a city name; the province name is located in the first layer, and the city name corresponding to the province name is located in the second layer.

For example, the province to which the error correction address belongs is Fujian Province, and the first-level name is Fujian Province.

S3. Obtain a second dictionary tree corresponding to the first-level name. The second dictionary tree is configured to store a city name, a county name, and a zone name corresponding to the current province name.

The node in the second dictionary tree represents a city name, a county name, or a zone name. The root node of the second dictionary tree is the first-level name.

For example, FIG. 5 is a second dictionary tree corresponding to Fujian Province.

S4. Identify a county name or a zone name corresponding to the to-be-corrected address according to the second dictionary tree, and obtain a secondary name.

For example, the area to which the error correction address belongs is the Gulou area, and the second level name is the Gulou area.

S5: Obtain a third dictionary tree corresponding to the second-level name; the third dictionary tree is configured to store a township name, a village name, and a street name corresponding to the second-level name; specifically:

The node in the third dictionary tree represents one character in a township name, a village name, or a street name.

The character corresponding to the preset order is the first character after the second-level name and the fourth character after the second-level name.

For example, FIG. 6 is a third dictionary tree corresponding to the Gulou area. The third dictionary tree is saved according to the word formation node. The input address when querying is, for example, “Hongshan Bridge, Hongshan Town, Gulou District, Fuzhou City, Fujian Province”. The address of the district is “Hongshan Bridge Hongshan Town”, according to the first word “Hong”. You can filter out the first node of the dictionary tree that needs to be restored as a "Hong" branch. Other branches such as "Wufeng Street" do not need to be restored to reduce memory usage.

The first character and the fourth character are generally the first words of the town and the village. Considering the general situation to reduce the branch of the dictionary tree that needs to be restored, the case of non-conformity cannot be reduced, and only the original third dictionary tree can be restored. A tailored third dictionary tree is constructed, and each word after the county name of the address to be corrected is queried. If the query is not available, the child nodes of all nodes of the current branch in the third dictionary tree are used as candidate nodes, and the next word is queried in the candidate nodes. For example, the address to be corrected is “Five North Road, Gulou District, Fuzhou City, Fujian Province”. In Figure 6, “Gulou District” can be found after “five”, after “five”, if “x” is not found, then The child nodes of the nodes "four", "one", and "phoen" are used as candidate nodes. If the next word is queried, the word "north" can be queried.

Among them, as the province, city, county, township and village streets are matched step by step, the dictionary tree corresponding to each level is dynamically acquired, and a complete dictionary tree corresponding to the address to be corrected is constructed, as shown in FIG. 7 .

At the end of the query, you can get a node with the lowest level. According to the pointer of this node, you can find the only parent node of the upper layer, and this parent node can find its only parent node. This process is called backtracking. The lowest level node is backtracked to get the first node, and the first node is connected to the lowest node to get a string. Returns all addresses prefixed by this string as a candidate address based on this string. For example, in Figure 7, the lowest node is the "bridge" word. According to the node pointer, the last parent node can be obtained as the "mountain". Repeat this process to get the last parent node, that is, the first node after the zone name. "flood". Connect the first node to the last node to get the Hongshan Bridge. The string corresponding from the root node to the lowest node is "Hongshan Bridge, Hongshan Town, Gulou District, Fuzhou City, Fujian Province".

S7. Select an optimal address according to the candidate address set. Specifically:

S73. Perform S71 to S72 repeatedly until the candidate address set is traversed.

S74. Acquire a candidate address having the largest matching number in the candidate address set to obtain an optimal address.

S75. Update the to-be-corrected address according to the optimal address to obtain a correct address. Specifically:

Obtaining, from the optimal address, a character string located before two or more consecutive characters that are not adapted to the error correction address; updating the error correction address according to the character string to obtain a correct address;

Otherwise, set the best address to the correct address.

The address identified by OCR (Optical Character Recognition Technology) is called an error-correction address, and the error-correction address may have an error. The candidate address can be obtained by querying the error correction address in the dictionary tree. The address most similar to the address to be error corrected among the candidate addresses is selected as the best address, and the degree of similarity is evaluated according to the same number of Chinese characters in the same position. Then compare the best address with the address to be corrected. If the number of consecutive different Chinese characters is within two words, the best address is used as the correct address; if the number of consecutive different words is two or more, then this part will be followed. The address to be corrected is the correct address. According to the above error correction principle, the best address and the error correction address are combined as the last correct address.

For example, the address to be corrected is “Hongshan, Hongshan, Gulou District, Fuchuan City, Fujian Province”. The candidate address is “Hongshan Bridge, Hongshan Town, Gulou District, Fuzhou City, Fujian Province”. In the city level, since there are consecutive different Chinese characters, "Fuzhou City" is taken as the correct address instead of "Fuchuan City" in the address to be corrected. In the same way, after the district and county level, there is only one consecutive Chinese character in the "Hongshan Bridge of Hongshan Town" (two different words are discontinuous), so the final address after correction is "Gulou District, Fuzhou City, Fujian Province". Hongshan Bridge in Hongshan Town." Because the query is carried out according to the province, city, county and county level, the error correction also compares the recognition result and the query result according to the classification, and selects whether to perform error correction according to the above error correction principle.

It can be seen from the above description that the national address is stored hierarchically according to the provinces, cities, counties, towns and villages, and the province name is saved as the first dictionary tree for querying the province to which the error correction address belongs. The provincial, city, and county addresses of the province are saved as a second dictionary tree for querying the district and county names. Finally, the street-level address of the township village is built according to the word structure to reduce the redundancy, and the dictionary tree is restored. The dictionary tree needs to be restored according to the address to be corrected, and the number of nodes is reduced. According to the provincial, city, district, township and village street level query, as long as there are not many typos in the error correction address, the correct address name can be obtained according to the similarity at the provincial, city, district and county levels. A candidate address that is closest to the correct address can be obtained. Finally, according to the error correction principle, the error correction address and the candidate address are compared to obtain an error corrected address. For example, the address to be corrected is “Hongshan and Hongshan Overseas Chinese in Gulou District, Fuchuan City, Fujian Province”. At the provincial, city, district and county levels, according to the similarity, it can be concluded that “Fujian Province” has the next-level address as “Fux City”. "Fuzhou City", at the village level, according to the lowest level node "mountain" that can be queried, you can get "Hongshan Town Hongshan", so the candidate address is the address prefixed with "Hongshan Town, Fuzhou City, Fushan City, Fujian Province", such as "Fujian Province" Hongshan Bridge, Hongshan Town, Fuzhou City." According to the principle of error correction, there are no more than one consecutive Chinese characters, so the address to be corrected is corrected as "Hongshan Bridge, Hongshan Town, Fuzhou City, Fujian Province".

Compared with the word-based statistical language model, the present invention does not need to train the parameter model, and does not need to calculate the probability of occurrence of the word multiple times. The search algorithm is used to find the optimal path, and only after constructing the dictionary tree, the query can be performed. faster. Different cities may have counties or towns or villages with the same name. According to the statistical model, the first-order Markov may not be able to judge, and when the order is increased to judge, the calculation amount also increases. In the hierarchical query, the present invention enters different branch queries according to the constructed dictionary tree, and the address names below the county are stored as nodes, and the candidate addresses are obtained by backtracking from the lowest node.

Compared with the dictionary tree for constructing the national address, the present invention constructs the dictionary tree of the county for the provinces, cities and counties that need to be queried according to the information of the address to be corrected, and then performs the county dictionary tree according to the address to be corrected. Cropping greatly reduces the space required and query time. If the national address data is saved as text about 60M, the average address data of a province is about 2M. When building the entire province address dictionary tree during query, it takes at least a dozen M memory, and it takes 5s to query the address once. According to the county name, when querying the village-level address, only the address under the county needs to be restored. The general data amount of the dictionary tree restored after the cropping is only a few K, and it takes about 0.05s to query the address once. After the national address is constructed as a dictionary tree, the tree node is saved as a text by layer to about 10M, indicating that the dictionary tree structure effectively removes the redundancy of the village address. The nodes in the county dictionary tree use bidirectional pointers. The last node that is queried can be traced back to the first node, the connection gets the address prefix, and the candidate error correction address can be obtained according to the address prefix. The general dictionary tree structure is used for searching, the pointer is one-way, and the node can only be queried from top to bottom, and the dictionary tree of the present invention is a bidirectional pointer, and the candidate address can be obtained by backtracking to the first node according to the lower layer node.

The query time is derived from the Debug mode of the Visual Studio software of the same laptop.

方案Program	原始文本数据Raw text data	Access数据库Access database	SQLite数据库SQLite database	字典树结构Dictionary tree structure
数据存储空间Data storage space	60M60M	100M100M	50M50M	10M10M
查询一次用时Query once	--	0.5s-2s0.5s-2s	0.05s0.05s	0.05s0.05s

Embodiment 2 of the present invention is:

The embodiment provides an address error correction terminal comprising one or more processors 1 and a memory 2, the memory 2 storing a program, and being configured to perform the following steps by the one or more processors 1:

S1: Obtain an address to be corrected.

The node in the first dictionary tree represents a province name or a city name; the province name is located on the first layer, and the city name corresponding to the province name is located on the second layer.

Otherwise, set the best address to the correct address.

In summary, the address correction method and terminal provided by the present invention store the national address in stages according to the province, city, county, township, and village streets, and sequentially check the province information, the city and county information, and the city and county information in the address to be corrected. Township village street information, and dynamically retrieve the dictionary tree corresponding to the next-level address according to each verification result, which greatly reduces the memory space occupation in the address error correction process, and has high accuracy. . Further, in the case that the province name of the address to be error-corrected is more serious, the name of the province corresponding to the address to be corrected can be confirmed by the city name, which is advantageous for improving the accuracy of error correction. Further, in general, the provinces, cities, and counties are less likely to be duplicated, and the entire word can be saved as a node. The county level may be a township, a village, or a street. The possibility of repeated occurrence is relatively large, sharing the same prefix. It can effectively reduce redundancy and reduce the space required. Further, by specifying the character of the specific location and selecting the branch information adapted to the character of the specific location as the candidate address, the capacity of the third dictionary tree is reduced, that is, the space required for checking the street address of the township village is reduced. . Further, the first character after the second-level name is generally the first character of the town name, and the fourth character after the second-level name is generally the first character of the village name. Generally, the town and the village after the county name can be screened, which can be effective. Reduce the dictionary tree nodes that need to be generated. Further, the correct rate of selecting the address with the highest similarity to the error-correction address from the one or more candidate addresses as the correct address is improved.

Claims

An address error correction method, comprising:

S1, obtaining an address to be corrected;

S2: Identify, according to the first dictionary tree, a province name corresponding to the to-be-corrected address, to obtain a first-level name; the first dictionary tree is configured to store a province name and a city name;

S3: Obtain a second dictionary tree corresponding to the first-level name; the second dictionary tree is configured to store a city name, a county name, and a district name corresponding to the current province name;

S4. Identify, according to the second dictionary tree, a county name or a zone name corresponding to the to-be-corrected address, and obtain a secondary name;

S5: Obtain a third dictionary tree corresponding to the second-level name; the third dictionary tree is configured to store a township name, a village name, and a street name corresponding to the second-level name;

S6. Acquire one or more candidate addresses corresponding to the to-be-corrected address according to the third dictionary tree, to obtain a candidate address set.
The address error correction method according to claim 1, wherein the S2 is specifically:

When there is no province name adapted to the error correction address in the first dictionary tree, acquiring a city name adapted to the error correction address, obtaining a current city name; acquiring the current city name The corresponding province name gives the first-level name.
The address error correction method according to claim 1, further comprising:

A node in the first dictionary tree represents a province name or a city name;

A node in the second dictionary tree represents a city name, a county name or a zone name;

A node in the third dictionary tree represents one of a township name, a village name, or a street name.
The address error correction method according to claim 1, wherein the S5 is specifically:

Obtaining a dictionary tree corresponding to the second-level name to obtain a third dictionary tree;

Obtaining a character corresponding to the preset order after the second-level name from the to-be-corrected address, to obtain a current character;

And a third dictionary tree to be constructed according to the branch of the third dictionary tree adapted to the current character; the root node of the third dictionary tree is the second-level name.
The address error correction method according to claim 4, further comprising:

The character corresponding to the preset order is a first character after the second-level name and a fourth character after the second-level name.
The address error correction method according to claim 1, wherein after the S6, the method further comprises:

S71. Obtain a candidate address from the candidate address set to obtain a current candidate address.

S72. Count the number of characters in the same position of the current candidate address and the same location of the to-be-corrected address, and obtain a matching number.

S73. Repeat performing the S71 to the S72 until the candidate address set is traversed;

S74. Obtain a candidate address having the largest matching number in the candidate address set, to obtain an optimal address.

S75. Update the to-be-corrected address according to the optimal address to obtain a correct address.
The address error correction method according to claim 6, wherein the S75 is specifically:

If there are more than two consecutive characters in the optimal address that are not adapted to the address to be corrected, then:

Obtaining, from the optimal address, a character string located before two or more consecutive characters that are not adapted to the error correction address;

Updating the to-be-corrected address according to the string to obtain a correct address;

Otherwise, set the best address to the correct address.
The address error correction method according to claim 1, wherein the S1 is specifically:

The address information in the identity card is identified by an optical character recognition technology to obtain the address to be corrected.
An address correction terminal characterized by comprising one or more processors and a memory, the memory storing a program, and being configured to perform the following steps by the one or more processors:

S1, obtaining an address to be corrected;

S2: Identify, according to the first dictionary tree, a province name corresponding to the to-be-corrected address, to obtain a first-level name; the first dictionary tree is configured to store a province name and a city name;

S3: Obtain a second dictionary tree corresponding to the first-level name; the second dictionary tree is configured to store a city name, a county name, and a district name corresponding to the current province name;

S4. Identify, according to the second dictionary tree, a county name or a zone name corresponding to the to-be-corrected address, and obtain a secondary name;

S5: Obtain a third dictionary tree corresponding to the second-level name; the third dictionary tree is configured to store a township name, a village name, and a street name corresponding to the second-level name;

S6. Acquire one or more candidate addresses corresponding to the to-be-corrected address according to the third dictionary tree, to obtain a candidate address set.
The address error correction terminal according to claim 9, wherein the S2 is specifically:

When there is no province name adapted to the error correction address in the first dictionary tree, acquiring a city name adapted to the error correction address, obtaining a current city name; acquiring the current city name The corresponding province name gives the first-level name.
The address correction terminal according to claim 9, further comprising:

A node in the first dictionary tree represents a province name or a city name;

A node in the second dictionary tree represents a city name, a county name or a zone name;

A node in the third dictionary tree represents one of a township name, a village name, or a street name.
The address correction terminal according to claim 9, wherein the S5 is specifically:

Obtaining a dictionary tree corresponding to the second-level name to obtain a third dictionary tree;

Obtaining a character corresponding to the preset order after the second-level name from the to-be-corrected address, to obtain a current character;

And a third dictionary tree to be constructed according to the branch of the third dictionary tree adapted to the current character; the root node of the third dictionary tree is the second-level name.
The address correction terminal according to claim 12, further comprising:

The character corresponding to the preset order is a first character after the second-level name and a fourth character after the second-level name.
The address correction terminal according to claim 9, wherein after the S6, the method further comprises:

S71. Obtain a candidate address from the candidate address set to obtain a current candidate address.

S72. Count the number of characters in the same position of the current candidate address and the same location of the to-be-corrected address, and obtain a matching number.

S73. Repeat performing the S71 to the S72 until the candidate address set is traversed;

S74. Obtain a candidate address having the largest matching number in the candidate address set, to obtain an optimal address.

S75. Update the to-be-corrected address according to the optimal address to obtain a correct address.
The address correction terminal according to claim 14, wherein the S75 is specifically:

If there are more than two consecutive characters in the optimal address that are not adapted to the address to be corrected, then:

Obtaining, from the optimal address, a character string located before two or more consecutive characters that are not adapted to the error correction address;

Updating the to-be-corrected address according to the string to obtain a correct address;

Otherwise, set the best address to the correct address.
The address error correction terminal according to claim 9, wherein the S1 is specifically:

The address information in the identity card is identified by an optical character recognition technology to obtain the address to be corrected.