An address matching method and apparatus

Abstract

Embodiments of the present application disclose an address matching method and device, which can be applied to the field of geographic information data processing and used for solving the problem of low address matching accuracy in the prior art. The method comprises the following steps: obtaining a to-be-matched address; performing word segmentation on the to-be-matched address according to a pre-stored word library to obtain a first address element; the pre-stored word library comprises a second address element extracted from the to-be-matched address by a named entity recognition model and a basic word segmentation word library; performing address standardization processing on the first address element to obtain a standardized to-be-matched address; and matching a target address of the standardized to-be-matched address from a pre-set database. The method can accurately recognize the first address element of the to-be-matched address, and the accuracy of matching the target address of the to-be-matched address from the pre-set database can be improved after performing address standardization processing on the first address element.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to an address matching method and device. BACKGROUND

[0002] Address information is an essential resource in social development and has important application value in many industries such as medical treatment and public security. Through further conversion of address data, its corresponding spatial data is obtained, and in the field of geographic information data processing, this process is called address matching. For example, a user inputs a to-be-matched address into a terminal device, the terminal device matches the address with an address in a preset database and returns the corresponding latitude and longitude coordinates, so that the address can be located on a map.

[0003] At present, non-standard address data and inaccurate address segmentation make the accuracy of address matching often low. Address element errors, address element ambiguities and address element omissions and other problems cause great obstacles to address matching tasks. For example, the to-be-matched address is "Zhejiang Province Taizhou Huangyan District Touta Town Zhenxing Road 6 Touta Market", in which the address element "Zhejiang Province" is written as "Zhejiang Province", which is an address element error. For another example, the to-be-matched address is "Zhejiang Province Taizhou Touta Town Zhenxing Road near Touta Market", in which the address element "Zhenxing Road near" is not clear in address pointing, which is an address element ambiguity problem. For another example, the to-be-matched address is "Zhejiang Province Huangyan District Touta Town Zhenxing Road 6 Touta Market", in which the to-be-matched address lacks the prefecture-level address element "Taizhou City", which is an address element omission. For another example, the address is "Jiangsu Province Nanjing City Liuhe District Zhuzhen Town Donghou Street 5 Zhuzhen Town Ethnic Primary School", in which the address element "Zhuzhen Town" may be incorrectly segmented as "Zhuzhen", resulting in low address matching accuracy. SUMMARY

[0004] The embodiments of the present application provide an address matching method and device, which optimizes the address segmentation process by pre-storing a word library for the to-be-matched address, and optimizes the address matching process by combining at least one preset strategy, thereby solving the problem of low address matching accuracy in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a data processing method, which comprises: obtaining a to-be-matched address; segmenting the to-be-matched address according to a pre-stored word library to obtain a first address element; the pre-stored word library comprises a second address element extracted from the to-be-matched address by a named entity recognition model and a basic segmentation word library; performing address standardization processing on the first address element to obtain a standardized to-be-matched address; and matching a target address of the standardized to-be-matched address from a preset database.

[0006] In the above scheme, the pre-stored word library constructed by the basic word segmentation word library and the second address element of the to-be-matched address extracted by the named entity recognition model can optimize the word segmentation process of the to-be-matched address, so that the first address element obtained by word segmentation is more accurate. In addition, the address standardization processing of the first address element is beneficial to address matching, so as to improve the accuracy of address matching.

[0007] Optionally, according to the pre-stored word library, before the word segmentation of the to-be-matched address, the method further includes: encoding the to-be-matched address to obtain feature information of the to-be-matched address; and decoding the feature information of the to-be-matched address to obtain the second address element.

[0008] By using the encoding process and the decoding process, the context semantic information in the to-be-matched address can be obtained, so that the second address element of the to-be-matched address extracted is more accurate, thereby improving the accuracy of address matching.

[0009] Optionally, the first address element includes any one or more of the following contents: a provincial element; a prefectural element; a district or county element; a street, town or village element; a village committee or residential committee element; a road or group element; a road number or group number element; a point of interest element; a building number element; a unit number element; a floor number element; and a room number element.

[0010] Optionally, the address standardization processing of the first address element to obtain the standardized to-be-matched address includes: traversing the first address element according to an administrative division element standard, and mapping non-standardized elements in the first address element into standardized elements; and obtaining the standardized to-be-matched address according to the standardized elements, a hierarchical backtracking mapping relationship in the basic word segmentation word library, and a hierarchical tracking mapping relationship.

[0011] In the above scheme, in the address standardization processing of the first address element, by using the hierarchical backtracking mapping and the hierarchical tracking mapping, the problem of low address matching accuracy caused by the missing of the address element of the to-be-matched address can be solved, thereby improving the accuracy of address matching.

[0012] Optionally, the standardized address to be matched includes a first detailed address and a first point of interest; and the matching of the target address from the preset database according to the standardized address to be matched includes: screening at least one second detailed address and at least one second point of interest from the preset database according to an administrative division field of the standardized address to be matched, the second detailed address corresponding to the second point of interest; matching the second detailed address with the first detailed address; if the matching result is not empty, determining a first similarity between the first point of interest and the second point of interest; if the first similarity is greater than a first threshold, taking the second detailed address and the second point of interest as the target address; or, matching the first point of interest with the second point of interest; if the matching result is not empty, determining a second similarity between the second detailed address and the first detailed address; and if the second similarity is greater than a second threshold, taking the second detailed address and the second point of interest as the target address.

[0013] By using the administrative division field to perform a round of screening from the preset database, a large amount of irrelevant data in the preset database can be filtered out, which helps to reduce the calculation amount in the subsequent matching process.

[0014] Optionally, if the first similarity is less than or equal to the first threshold, or if the second similarity is less than or equal to the second threshold, or if the matching result of the first point of interest and the second point of interest is empty, the matching of the target address from the preset database according to the standardized address to be matched further includes: segmenting the first point of interest, determining a keyword of the first point of interest according to the segmented words; matching the keyword with the at least one second point of interest, if the matching result is not empty, determining a third similarity between the second detailed address and the first detailed address; and if the third similarity is greater than a third threshold, taking the second detailed address and the second point of interest as the target address.

[0015] By obtaining the keyword of the first point of interest and using the keyword for screening, part of the data can be further effectively filtered, and the calculation amount in the subsequent matching process can be further reduced.

[0016] Optionally, if the third similarity is less than or equal to a third threshold value, or if the keyword and the at least one second matching result of the interest point are empty, the matching of the standardized address to be matched with the target address from the preset database further comprises: determining a fourth similarity of the first detailed address and the at least one second detailed address respectively, and determining a fifth similarity of the first interest point and the at least one second interest point respectively, and calculating a sixth similarity by weighting the fourth similarity and the fifth similarity. According to the maximum similarity in the sixth similarity, the corresponding second detailed address and the second interest point are determined, and the second detailed address and the second interest point are taken as the target address.

[0017] Based on the above scheme, by using at least one preset strategy combined with the hierarchical address matching method, the similarity calculation amount in the address matching process can be effectively reduced.

[0018] In a second aspect, an address matching device is provided, which comprises: an acquisition unit configured to acquire an address to be matched; a processing unit configured to perform word segmentation on the address to be matched to obtain a first address element, and to perform address standardization processing on the first address element to obtain a standardized address to be matched; and to match a target address of the standardized address to be matched from a preset database.

[0019] Optionally, the processing unit is specifically configured to encode the address to be matched to obtain feature information of the address to be matched, and to decode the feature information of the address to be matched to obtain the second address element.

[0020] Optionally, the first address element comprises any one or more of the following: a provincial element; a prefectural element; a district or county element; a street, town or township element; a village committee or residents' committee element; a road or group element; a road number or group number element; an interest point element; a building number element; a unit number element; a floor number element; and a room number element.

[0021] Optionally, the processing unit is specifically configured to traverse the first address element according to an administrative division element standard, to map non-standardized elements in the first address element into standardized elements, and to obtain the standardized address to be matched according to the standardized elements, a hierarchical backtracking mapping relationship in the basic word segmentation dictionary, and a hierarchical tracking mapping relationship.

[0022] Optionally, the standardized address to be matched comprises a first detailed address and a first point of interest; the processing unit is specifically configured to: filter at least one second detailed address and at least one second point of interest from the preset database according to an administrative division field of the standardized address to be matched, the second detailed address corresponding to the second point of interest; match the second detailed address with the first detailed address; if the matching result is not empty, determine a first similarity of the first point of interest and the second point of interest; if the first similarity is greater than a first threshold, take the second detailed address and the second point of interest as the target address; or, match the first point of interest with the second point of interest; if the matching result is not empty, determine a second similarity of the second detailed address and the first detailed address; if the second similarity is greater than a second threshold, take the second detailed address and the second point of interest as the target address.

[0023] Optionally, if the first similarity is less than or equal to the first threshold, or if the second similarity is less than or equal to the second threshold, or if the matching result of the first point of interest and the second point of interest is empty, the processing unit is specifically configured to: segment the first point of interest, determine a keyword of the first point of interest according to the segmented word; match the keyword with the at least one second point of interest, if the matching result is not empty, determine a third similarity of the second detailed address and the first detailed address; if the third similarity is greater than a third threshold, take the second detailed address and the second point of interest as the target address.

[0024] Optionally, if the third similarity is less than or equal to the third threshold, or if the matching result of the keyword and the at least one second point of interest is empty, the processing unit is specifically configured to: respectively determine a fourth similarity of the first detailed address and the at least one second detailed address, and respectively determine a fifth similarity of the first point of interest and the at least one second point of interest, and perform weighted calculation on the fourth similarity and the fifth similarity to obtain a sixth similarity. According to the maximum similarity in the sixth similarity, the corresponding second detailed address and the second point of interest are determined, and the second detailed address and the second point of interest are taken as the target address.

[0025] In a third aspect, an embodiment of the present application provides a computing device, comprising at least one processor and at least one memory, wherein the memory stores a computer program, and when the program is executed by the processor, the processor executes the address matching method in any of the first aspect.

[0026] In a fourth aspect, the embodiments of the present application further provide a computer readable storage medium, which stores a program, and when the program is run on a computer, the computer is caused to implement the address matching method according to any of the first aspect. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0028] Figure 1 A method flow diagram for constructing a basic segmentation word library provided by the embodiments of the present application is shown in the figure.

[0029] Figure 2 A method flow diagram of the address matching method provided by the embodiments of the present application is shown in the figure.

[0030] Figure 3 A method flow diagram for constructing an extended word library provided by the embodiments of the present application is shown in the figure.

[0031] Figure 4 A method flow diagram of the hierarchical address matching method combined with at least one preset strategy provided by the embodiments of the present application is shown in the figure.

[0032] Figure 5 A structure diagram of the address matching device provided by the embodiments of the present application is shown in the figure.

[0033] Figure 6 A structure diagram of the computing device provided by the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0034] In order to make the objects, technical solutions and advantages of the present application more clear, the present application will be further described in detail below with reference to the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort are within the scope of protection of the present application.

[0035] In the following, some terms in the embodiments of the present application will be explained. It should be noted that these explanations are for the convenience of understanding by those skilled in the art, and do not limit the scope of protection required by the present application.

[0036] 1) administrative division

[0037] Administrative division is a region divided by a country for the convenience of administration. Therefore, administrative division is also called administrative region. In the embodiment of the present application, the administrative division is divided into five levels, including provincial administrative region, prefectural administrative region, county-level administrative region, township-level administrative region and village-level administrative region.

[0038] Among them, the provincial administrative region is a first-level administrative region, or a national first-level administrative region, which refers to the administrative division directly under the central government. The provincial administrative region includes provinces, municipalities directly under the central government, autonomous regions and special administrative regions. The prefectural administrative region is a second-level administrative region, which is under the jurisdiction of the first-level administrative region. The prefectural administrative region includes prefecture-level cities, districts, autonomous prefectures and leagues. The county-level administrative region is a third-level administrative region, which has the same administrative status as a county, including urban districts, county-level cities, counties, autonomous counties, banners, autonomous banners, special zones and forest zones. The township-level administrative region is a fourth-level administrative region, which has the same administrative status as a township, including streets, towns, townships, ethnic townships, su woods, ethnic su woods and county-level districts. The village-level administrative region is a fifth-level administrative region, including administrative villages and residential committees, which is the smallest basic self-governing unit in China's administrative system, and its management organization is the villagers committee. In order to avoid confusion with the geographical concept of village, village and natural village, the administrative area of the villagers committee with local political meaning is often called administrative village. Generally, an administrative village manages several natural villages or an administrative village manages one natural village. However, when a natural village is large in size, it will be divided into several administrative villages for the convenience of management.

[0039] 2) Administrative division element

[0040] The administrative division element includes the name of the administrative division element and the level to which the administrative division element belongs. The administrative division elements obtained from the above five levels of administrative division are also divided into five levels. Among them, the first-level administrative division element is the provincial administrative division element, the second-level administrative division element is the prefectural administrative division element, the third-level administrative division element is the district-level administrative division element or the county-level administrative division element, the fourth-level administrative division element is the street-level administrative division element or the town-level administrative division element or the township-level administrative division element, and the fifth-level administrative division element is the village-level administrative division element or the residential committee-level administrative division element.

[0041] For example, the administrative division element is "Hangzhou City", which belongs to the prefectural level. For another example, the administrative division element is "Shaanxi Province", which belongs to the provincial level.

[0042] 3) Address element

[0043] The address element refers to all elements in the extracted address, including one or several of administrative division elements (first-level administrative division element, second-level administrative division element, third-level administrative division element, fourth-level administrative division element, fifth-level administrative division element), road group element or number element, house number element or group number element, point of interest element, building number element, unit number element, floor number element and room number element.

[0044] When one administrative village manages several natural villages, the natural villages are grouped, and the group to which the natural village belongs is referred to as a group element, and the group number of the group to which the natural village belongs is referred to as a group number element. The point of interest (POI) refers to any meaningful point on a map that has no geographical meaning. In a geographic information system, a point of interest can represent a building, a store, a scenic spot, a house or a bus stop, etc. For example, “Wanda Plaza” is a point of interest. For another example, “Pudong Bank” is also a point of interest.

[0045] Before performing address matching on the to-be-matched address, the to-be-matched address needs to be segmented. The to-be-matched address can be accurately segmented by using a pre-stored word library. The pre-stored word library includes a basic segmentation word library and an extended word library including the second address element.

[0046] As shown in FIG. 1, a method flow diagram for constructing a basic segmentation word library is provided in an embodiment of the present application. The method includes the following steps: Figure 1

[0047] Step 101: Obtain an administrative division element according to national administrative division data.

[0048] In an embodiment of the present application, the national administrative division data is obtained through the official website of the National Bureau of Statistics of the People's Republic of China, and the administrative division element in the national administrative division data is added to the basic segmentation word library.

[0049] Step 102: Remove the main feature word to obtain the abbreviation of the name of the administrative division element.

[0050] In an embodiment of the present application, the abbreviation of the corresponding administrative division element is obtained by removing the main feature word of the name of the administrative division element. The main feature word includes but is not limited to “province”, “city”, “district”, “county”, “township”, “town”, “street”, “village” and “residential committee”.

[0051] Taking the administrative division element “Zhejiang Province” as an example, the abbreviation of the corresponding administrative division element “Zhejiang” is obtained by removing the main feature word “province”.

[0052] Step 103: Establish an administrative division element mapping table to complete the construction of the basic segmentation word library.

[0053] ​In the embodiments of the present application, the administrative division element mapping table includes an alias standardization mapping table, a hierarchical tracking mapping table, and a hierarchical backtracking mapping table. Please refer to Table 1. Table 1 shows the settings of the administrative division element mapping table and examples thereof.

[0054] Table 1 Administrative division element mapping table

[0055]

[0056]

[0057] Specifically, the alias standardization mapping table (alias_to_standard) records the mapping relationship between the name of an administrative division element and the alias of the administrative division element corresponding thereto. The alias of an administrative division element includes, but is not limited to, the abbreviation, the alternative name, and the abbreviation of the name of the administrative division element. For example, the abbreviation of the administrative division element "Shanghai" is "Shanghai", the alternative name is "Shen", and the abbreviation is "Shanghai". In a possible implementation manner, the alias standardization mapping table can be established according to the abbreviation, the alternative name, and the abbreviation of the name of the administrative division element obtained through network searching.

[0058] The hierarchical tracking mapping table (name_to_hierarchical) records the mapping relationship between the first four-level administrative division elements and the administrative division hierarchy to which the administrative division elements belong. For example, the administrative division element "Taizhou" is mapped to "prefecture-level". For another example, the administrative division element "Shanghai" is mapped to "provincial-level".

[0059] Based on the hierarchical tracking mapping table, a hierarchical backtracking mapping table is established. This hierarchical backtracking mapping table contains three sub-tables: a second-level administrative division element mapping sub-table (city_to_province), a third-level administrative division element mapping sub-table (district_to_city), and a fourth-level administrative division element mapping sub-table (town_to_district). The second-level administrative division element mapping sub-table (city_to_province) records the mapping relationship between second-level administrative division elements and their corresponding first-level administrative division elements. For example, the second-level administrative division element "Taizhou City" will be mapped to its corresponding first-level administrative division element "Zhejiang Province". The third-level administrative division element mapping sub-table (district_to_city) records the mapping relationship between third-level administrative division elements and their corresponding second-level administrative division elements. For example, the third-level administrative division element "Huangyan District" will be mapped to its corresponding second-level administrative division element "Taizhou City". The fourth-level administrative division element mapping sub-table (town_to_district) records the mapping relationship between fourth-level administrative division elements and their corresponding third-level administrative division elements. For example, the fourth-level administrative division element "Tou Tuo Town" will be mapped to its corresponding third-level administrative division element "Huangyan District". It is worth noting that when duplicate names exist, administrative division elements with the same name will not be added to the hierarchical backtracking mapping table. For example, "Chengguan Town" exists in many cities, such as "Chengguan Town, Wuqing District, Tianjin," "Chengguan Town, Wuhe County, Bengbu City, Anhui Province," and "Chengguan Town, Wudu District, Longnan City, Gansu Province." Therefore, it is impossible to determine the parent administrative division element of the administrative division element "Chengguan Town." Consequently, the administrative division element "Chengguan Town" will not be added to the hierarchical backtracking mapping table.

[0060] The mapping priority of the hierarchical backtracking mapping table is as follows: the mapping of the fourth-level administrative division element mapping sub-table (town_to_district) is higher than the mapping of the third-level administrative division element mapping sub-table (district_to_city), and the mapping of the third-level administrative division element mapping sub-table (district_to_city) is higher than the mapping of the second-level administrative division element mapping sub-table (city_to_province).

[0061] It should be noted that the established administrative division element mapping table is not only used for the construction of the basic word segmentation lexicon, but also for the subsequent standardization processing of the addresses to be matched. For details, please refer to the relevant introduction below, which will not be repeated here.

[0062] like Figure 2 The diagram shown is a flowchart of an address matching method provided in an embodiment of the present invention. The method includes the following steps:

[0063] Step 201: Obtain the address to be matched.

[0064] In the embodiment of the present application, the user inputted address to be matched is acquired. The address to be matched can be in text format or other possible formats. In the following description, the address to be matched is taken as an example in text format.

[0065] In step 202, the address to be matched is segmented according to the pre-stored word library to obtain first address elements.

[0066] In the embodiment of the present application, the pre-stored word library includes a basic segmentation word library and an extended word library including second address elements. The construction process of the basic segmentation word library has been described above and will not be repeated here. The extended word library can be referred to the description below and will not be repeated here. The address to be matched is segmented according to the pre-stored word library to obtain first address elements. Figure 3

[0067] The segmentation refers to the decomposition of long text such as sentences, paragraphs and articles into data structures in word units to facilitate subsequent processing and analysis. After segmentation, the original semantics of the text will be split into more detailed individual words. The structure of the words is simpler than that of the long text, and it is easier for the computer to understand and analyze. Therefore, segmentation is often the first step of natural language processing. For English text, the words in a sentence can be easily divided by spaces, but Chinese has no obvious word division mark, so it needs to be segmented by a special method.

[0068] In a possible implementation, Jieba segmentation model is used for segmentation. Jieba segmentation has a built-in Chinese default dictionary, so it supports Chinese segmentation processing. The Jieba segmentation model is configured with a function interface. Through the function interface, the pre-stored word library is added to the word library of Jieba segmentation. In this way, Jieba segmentation can extract words that are not in its default word library, improving the accuracy of segmentation.

[0069] The following example shows a way to add a pre-stored word library to the word library of the Jieba segmentation model. The jieba.load_userdict(file_name) statement is used to add the pre-stored word library to the word library of the Jieba segmentation model, where the parameter file_name is the file name of the pre-stored word library.

[0070] ​After adding the pre-stored word library to the word library of the Jieba word segmentation model, the to-be-matched address is segmented to obtain the first address element. Adding the pre-stored word library to the Jieba word segmentation is conducive to more accurate segmentation of the to-be-matched address. Taking the to-be-matched address "No. 1, Shilin Road, Zhuzhen Town, Luhe District, Nanjing City, Jiangsu Province" as an example, if the pre-stored word library is not added to the Jieba word segmentation, the Jieba word segmentation will have difficulty in recognizing and processing special words such as "Zhuzhen Town", and therefore, it is likely to segment the to-be-matched address incorrectly as "Jiangsu Province", "Nanjing City", "Luhe District", "Zhuzhen", "Shilin Road", "No. 1", and "Luhe Zhuzhen Ethnic Middle School". However, by adding the pre-stored word library to the Jieba word segmentation, the Jieba word segmentation can extract words that are not in the default word library. As can be seen from the above introduction of the extended word library in the pre-stored word library, the extended word library can correctly recognize special words such as "Zhuzhen Town". Therefore, based on the pre-stored word library, the above to-be-matched address can be successfully segmented as "Jiangsu Province", "Nanjing City", "Luhe District", "Zhuzhen Town", "Shilin Road", "No. 1", and "Luhe Zhuzhen Ethnic Middle School".

[0071] In step 203, the first address element is subjected to address standardization processing to obtain a standardized to-be-matched address.

[0072] In the embodiment of the present application, the address standardization processing includes administrative division element standardization and administrative division element completion.

[0073] The administrative division element standardization refers to traversing the first address element and performing standardization processing thereon. The standardization processing includes alias standardization mapping, hierarchical tracking mapping, and hierarchical backtracking mapping. Specifically, first, the name of a non-standard administrative division element, i.e., the alias of the administrative division element, is mapped to the name of a standard administrative division element through an alias standardization mapping table (alias_to_standard). The establishment of the alias standardization mapping table will be described in detail in the foregoing description, which will not be repeated here. In address matching, the establishment of the alias standardization mapping table can solve the problem of low matching accuracy caused by the to-be-matched address containing a geographical name abbreviation or a geographical name short form. For example, the to-be-matched address is "Zhejiang Province Taizhou City Jiaojiang District Baiyun Mountain South Taizhou City People's Government", wherein "Baiyun Mountain South" is an alias of an administrative division element, and the administrative division level to which it belongs cannot be determined from the two characters "Baiyun Mountain South" alone. Therefore, "Baiyun Mountain South" needs to be mapped to "Baiyun Mountain South Road" according to the alias standardization mapping table, and the standardized to-be-matched address obtained is "Zhejiang Province Taizhou City Jiaojiang District Baiyun Mountain South Road Taizhou City People's Government". Further, the administrative division level to which the first four administrative division elements belong after standardization is obtained through a hierarchical tracking mapping table (name_to_hierarchical), and is recorded.

[0074] The administrative division element supplementing refers to performing administrative division element level backtracking on the administrative division element in the first address element according to the level backtracking mapping table in the order of mapping priority. For the level backtracking mapping table, refer to the foregoing description, which will not be repeated here. In the address matching, the level backtracking mapping table can solve the problem of low matching accuracy caused by the missing administrative division element of the address to be matched. For example, the address to be matched is "Zhejiang Province Jiaojiang District Baishan Road Taizhou People's Government", which lacks the second-level division element. According to the third-level administrative division element "Jiaojiang District", the level backtracking mapping table can map to obtain the second-level administrative division of the address to be matched, that is, "Taizhou City", and thus the standardized address to be matched is "Zhejiang Province Taizhou City Jiaojiang District Baishan Road Taizhou People's Government".

[0075] In step 203, the standardized address to be matched is obtained, for example, the address can be divided into seven fields. The first field is composed of the first-level administrative division element; the second field is composed of the second-level administrative division element; the third field is composed of the third-level administrative division element; the fourth field is composed of the fourth-level administrative division element; the fifth field is composed of the fifth-level administrative division element; the sixth field is the point of interest field, which is recorded by the point of interest element; and the seventh field is the detailed address field, which is composed of the remaining address elements in the address after removing the administrative division field and the point of interest field. The first five fields are called administrative division fields. It should be noted that the standardized address to be matched can not contain the fifth field.

[0076] Taking the standardized address to be matched "Zhejiang Province Taizhou City Huangyan District Touta Town Zhenxing Road and Hongyu Road Intersection West No. 1 Zhengbin Supermarket" as an example, the address is divided into fields: the administrative division fields "Zhejiang Province", "Taizhou City", "Huangyan District", "Touta Town", the point of interest field "Zhengbin Supermarket", and the detailed address field "Hongyu Road Intersection West No. 1".

[0077] In step 204, the target address of the standardized address to be matched is matched from the preset database.

[0078] In a possible implementation, according to the hierarchical address matching method combined according to at least one preset strategy, the target address of the standardized address to be matched is matched from the preset database.

[0079] Figure 3 A method flow diagram for constructing an extended vocabulary based on a named entity recognition model is provided for an embodiment of the present application. The method includes the following steps:

[0080] In step 301, a named entity recognition model is built.

[0081] Named Entity Recognition (NER) refers to the recognition of entities with specific meanings, mainly including names of people, places, organizations, and proper nouns. At present, the named entity recognition method based on deep learning is widely used in named entity recognition models. For the address to be matched, the named entity recognition technology is applied to obtain the address elements contained therein. The address elements of the above-identified address to be matched are referred to as second address elements. The second address elements constitute an extended word library.

[0082] In a possible implementation manner, an encoder-decoder model is used to build the named entity recognition model. The encoder refers to the conversion of data from one form to another, which can be used for data compression or feature extraction of data. The decoder is used to restore the above-identified information to its original form. Among them, the Nezha (NEZHA) model can be used as the encoder. The Nezha model is a Chinese pre-training language model, which is optimized in many places on the basis of the Bidirectional Encoder Representations from Transformers (BERT) model. The Conditional Random Fields (CRF) model can be used as the decoder. The CRF model is a discriminative probabilistic model, which is often used for labeling or analyzing sequence data, such as natural language text.

[0083] Step 302, the encoder encodes the address to be matched to obtain a feature vector.

[0084] In a possible implementation manner, the Nezha model first obtains an initial embedding vector of the address to be matched. The initial embedding vector includes a WordPiece Embedding vector, a Position Embedding vector, and a Segment Embedding vector. WordPiece Embedding refers to dividing a word into a set of limited common sub-word units, which can strike a balance between the effectiveness of the word and the flexibility of the character. Position Embedding refers to encoding the position information of a word. Position Embedding is a crucial part of introducing word position relationships into the model. Segment Embedding is used to distinguish two sentences. Among them, the Position Embedding vector is obtained by using relative position encoding, and the calculation formula is as follows:

[0085]

[0086]

[0087] wherein a ijis a vector, position i is a value of i in a two-dimensional coordinate (i, j), position j is a value of j in the two-dimensional coordinate (i, j), d z is a hidden layer embedding dimension of each head.

[0088] After obtaining the initial embedding vector, the ZhuZha model learns deep semantic information contained in the input address to be matched and outputs a feature vector. The feature vector refers to that the encoder extracts features of the input address to be matched and then vectorizes the features.

[0089] In step 303, the decoder decodes the feature vector to obtain a second address element and construct an extended vocabulary.

[0090] In one possible implementation, the feature vector is input into a CRF model. The CRF model decodes the feature vector to obtain an address element of the address to be matched, that is, a second address element.

[0091] Taking the address to be matched as "No. 1 Zhuling Road, Shilin, Zhutown, Luhe District, Nanjing City, Jiangsu Province, Luhe Zhutown Ethnic Middle School" as an example, the second address element can be obtained through the named entity recognition model, including: administrative division elements "Jiangsu Province", "Nanjing City", "Luhe District", "Zhutown", road level element "Shilin Road", road number level element "No. 1", and interest point element "Luhe Zhutown Ethnic Middle School".

[0092] Through steps 301 to 303, the extended vocabulary can be obtained.

[0093] Figure 4 A method flowchart of a hierarchical address matching method provided by an embodiment of the present application is provided. For the convenience of description of the scheme, the standardized address to be matched is taken as "No. 1 Zhengbin Supermarket, Xinchang Road and Hongyu Road intersection, Touta Town, Huangyan District, Taizhou City, Zhejiang Province" in the following description. The method includes the following steps:

[0094] In step 401, the administrative division field is screened.

[0095] The addresses consistent with the administrative division field of the standardized address to be matched are screened out in the preset database to obtain screened addresses.

[0096] In the embodiment of the present application, the preset database is composed of national addresses after address standardization processing. The national addresses can be collected from the Internet. The national addresses after address standardization processing contain the latitude and longitude coordinate information corresponding to the addresses. The implementation manner of the address standardization processing of the national addresses can refer to the implementation manner of the address standardization processing of the address to be matched described above, and will not be described here.

[0097] Specifically, the administrative division field of the standardized address to be matched in this example is "Zhejiang Province", "Taizhou City", "Huangyan District", and "Touda Town". The address data in the preset database that has the same administrative division field is filtered to obtain the filtered addresses. The filtered addresses can be one address or multiple addresses, and include all addresses under "Zhejiang Province, Taizhou City, Huangyan District, Touda Town" in the preset database.

[0098] In step 402, address matching is performed according to the first matching strategy, and it is determined whether the matching result is empty. If the matching result is not empty, step 403 is performed; if the matching result is empty, step 404 is performed.

[0099] In the embodiment of the present application, the first matching strategy refers to matching the detailed address field of the standardized address to be matched with the detailed address field of the filtered address. If the matching result is not empty, the matched address is returned and step 403 is performed. The matched address can be one address or multiple addresses, because one detailed address field can correspond to multiple points of interest. If the matching result is empty, step 404 is performed.

[0100] Alternatively, the first matching strategy also refers to matching the point of interest field of the standardized address to be matched with the point of interest field of the filtered address. If the matching result is not empty, the matched address is returned and step 403 is performed. The matched address can be one address or multiple addresses, because one point of interest field can correspond to multiple detailed address fields. If the matching result is empty, step 404 is performed.

[0101] Here, the detailed address field is matched as an example of the first matching strategy. The detailed address field "Xin Xing Road and Hong Yu Road Intersection, 1 West" of the address to be matched is matched with the detailed address field of the filtered address. If the matching result is not empty, the matched address is returned and step 403 is performed. The matched address can be one address or multiple addresses. For example, the detailed address field of one or more addresses in the filtered address can also be "Xin Xing Road and Hong Yu Road Intersection, 1 West", because the building represented by this detailed address field can include many points of interest. If the matching result is empty, step 404 is performed.

[0102] Here, the point of interest field is matched as an example of the first matching strategy. The point of interest field "Zheng Bin Supermarket" of the address to be matched is matched with the point of interest field of the filtered address. If the matching result is not empty, the matched address is returned and step 403 is performed. The matched address can be one address or multiple addresses. For example, there can be multiple "Zheng Bin Supermarkets" in the filtered address, and therefore, the detailed address corresponding to the same point of interest field "Zheng Bin Supermarket" can have multiple addresses. If the matching result is empty, step 404 is performed.

[0103] Step 403, determine whether the maximum value in the first similarity is greater than a set first threshold value or the maximum value in the second similarity is greater than a set second threshold value. If the maximum value in the first similarity is greater than the set first threshold value or the maximum value in the second similarity is greater than the set second threshold value, then execute step 407. If the maximum value in the first similarity is less than or equal to the set first threshold value or the maximum value in the second similarity is less than or equal to the set second threshold value, then execute step 404.

[0104] If step 402 is to match the detailed address field of the standardized address to be matched with the detailed address field of the screening address, after returning the matched address, the similarity between the point of interest field of the standardized address to be matched and the point of interest field of the matched address is calculated, which is called the first similarity. The first similarity can be calculated in various ways, for example, the edit distance can be used to calculate the similarity. The edit distance (ED) is the minimum number of operations required to convert one string into another string. The edit distance directly reflects the difference between the two texts from the literal meaning, that is, the more similar the two texts are, the smaller the edit distance is. The first similarity can include one calculation value or multiple calculation values. Determine whether the maximum value in the first similarity is greater than a set first threshold value τ1. The threshold value τ1 can be set artificially according to the situation. When the maximum value in the first similarity is greater than the set first threshold value τ1, execute step 407. When the maximum value in the first similarity is less than or equal to the set first threshold value τ1, execute step 404.

[0105] If step 402 is to match the point of interest field of the standardized address to be matched with the point of interest field of the screening address, after returning the matched address, the similarity between the detailed address field of the standardized address to be matched and the detailed address field of the matched address is calculated, which is called the second similarity. The first similarity can be calculated in various ways, for example, the edit distance can be used to calculate the similarity. The content about the edit distance can be referred to the above description, which will not be repeated here. The second similarity can include one calculation value or multiple calculation values. Determine whether the maximum value in the second similarity is greater than a set second threshold value τ2. The threshold value τ2 can be set artificially according to the situation. When the maximum value in the second similarity is greater than the set second threshold value τ2, then execute step 407. When the maximum value in the second similarity is less than or equal to the set second threshold value τ2, then execute step 404.

[0106] Step 404, perform the second matching strategy address matching, and determine whether the matching result is empty. If the matching result is not empty, then execute step 405. If the matching result is empty, then execute step 406.

[0107] In the embodiment of the present application, the second matching strategy refers to keyword matching. Specifically, the interest points of the address to be matched are segmented by using the Jieba segmentation tool to obtain a segmented word list, and the Term Frequency-Inverse Document Frequency (TF-IDF) of each word is calculated. The word with the largest TF-IDF value is selected as the keyword. The address search in step 401 contains the keyword obtained above. If the matching result is not empty, step 405 is executed, and if the matching result is empty, step 406 is executed.

[0108] wherein the term frequency-inverse document frequency technique is a commonly used weighting technique for information retrieval and text mining, which can be used to evaluate the importance of a word to a file. The file can refer to a piece of text or a document set or a document in a corpus. In the term frequency-inverse document frequency technique, the importance of a word increases in direct proportion to the number of times it appears in a file, but at the same time, it decreases in inverse proportion to the frequency of its appearance in the corpus. The term frequency (TF) refers to the frequency of a particular word in the file, and the calculation formula is as follows:

[0109]

[0110] The inverse document frequency (IDF) refers to a measure of the general importance of a word, and the frequency is inversely proportional to the degree of commonality of the word. The calculation method is to divide the total number of documents in the corpus by the number of documents containing the word in the corpus, and then take the logarithm of the quotient. The calculation formula is as follows:

[0111]

[0112] If a word is relatively rare but appears multiple times in an article, then this word is likely to reflect the characteristics of the article and is the desired keyword. Multiplying the TF value of a word by the IDF value of the word can obtain the TF-IDF value of the word. The higher the importance of a word to a text or article, the larger its TF-IDF value. Therefore, the word at the top is the keyword of the text or article.

[0113] For example, the interest point field of the address to be matched "Zheng Bin supermarket" is segmented by using the Jieba segmentation, and a segmented word list is obtained. The segmented word list includes two words "Zheng Bin" and "supermarket". The term frequency-inverse document frequency of each word is calculated, and the word with the maximum TF-IDF value is selected as the keyword. For the interest point field "Zheng Bin supermarket", the keyword is "Zheng Bin". The address containing the keyword "Zheng Bin" is obtained in step 401. If the matching result is not empty, step 405 is performed. For example, the address of the interest point "Zheng Bin supermarket" can be retrieved. If the matching result is empty, step 406 is performed.

[0114] In step 405, it is determined whether the maximum value in the third similarity is greater than a third threshold value. If the maximum value in the third similarity is greater than the third threshold value, step 407 is performed. If the maximum value in the third similarity is less than or equal to the third threshold value, step 406 is performed.

[0115] The similarity between the detailed address field of the address to be matched and the detailed address field of the matched address is calculated, which is referred to as the third similarity. The third similarity can be calculated in various ways, for example, the edit distance can be used to calculate the similarity. The content of the edit distance is described above and will not be repeated here. The third similarity can include one calculation value or multiple calculation values. It is determined whether the maximum value in the third similarity is greater than a third threshold value τ3. The threshold value τ3 can be artificially set according to the situation. When the maximum value in the third similarity is greater than the third threshold value τ3, step 407 is performed. When the maximum value in the third similarity is less than or equal to the third threshold value τ3, step 406 is performed.

[0116] In step 406, the third matching strategy address matching is performed.

[0117] In the embodiment of the present application, the third matching strategy refers to calculating the weighted value of the fourth similarity between the detailed address field of the standard address to be matched and the detailed address field of the filtered address obtained in step 401 and the fifth similarity between the interest point field of the standard address to be matched and the interest point field of the filtered address obtained in step 401, which is referred to as the sixth similarity. The calculation formula of the sixth similarity is:

[0118]

[0119] Wherein, len(a) is the length of the detailed address field; len(p) is the length of the interest point field; sim address is the fourth similarity of the detailed address field; sim poiThe fifth similarity is the similarity of the point of interest field; in this step, the similarity can be calculated by using the edit distance. After the matching in this step, the address corresponding to the maximum value in the sixth similarity is selected as the matching result.

[0120] For example, the detailed address field of the address to be matched, "Xin Xing Road and Hong Yu Road intersection west No. 1", is matched with all the detailed address fields in the screening address to obtain the fourth similarity, and the point of interest field of the address to be matched, "Zheng Bin supermarket", is matched with all the point of interest fields in the screening address to obtain the fifth similarity. The sixth similarity is obtained by weighting the fourth similarity and the fifth similarity, and the address corresponding to the maximum value in the sixth similarity is selected as the matching result.

[0121] In step 407, the matching result is output.

[0122] The matching result is output as the target address of the address to be matched. In this example, the target address of the address to be matched, "Zhejiang Province Taizhou Huangyan District Touta Town Xin Xing Road and Hong Yu Road intersection west No. 1 Zheng Bin supermarket", is output.

[0123] Based on the same inventive concept as described above, the embodiment of the present application also provides an address matching device, which can execute the method in the above-mentioned method embodiment of the present application. The structure of an address matching device provided by the embodiment of the present application can be referred to as Figure 5 The address matching device 500 includes an acquisition unit 501 and a processing unit 502. The acquisition unit 501 is configured to acquire an address to be matched. The processing unit 502 is configured to perform word segmentation on the address to be matched to obtain a first address element, perform address standardization processing on the first address element to obtain a standardized address to be matched, and match a target address of the standardized address to be matched from a preset database.

[0124] The acquisition unit 501 and the processing unit 502 are described in more detail in the method embodiment shown in Figure 2 The above-mentioned acquisition unit 501 and the processing unit 502 are described in more detail in the method embodiment shown in

[0125] Based on the same technical concept, the embodiment of the present application also provides a computing device, as shown in Figure 6 The computing device 600 includes at least one processor 601 and a memory 602 connected to the at least one processor. In the embodiment of the present application, the specific connection medium between the processor 601 and the memory 602 is not limited, Figure 6 for example, the processor 601 and the memory 602 are connected through a bus. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0126] In the embodiments of the present application, the memory 602 stores instructions executable by the at least one processor 601, and the at least one processor 601 can execute the steps included in the address matching method described above by executing the instructions stored in the memory 602.

[0127] The processor 601 is the control center of the computing device, can connect various parts of the computing device through various interfaces and lines, and can realize data processing by running or executing instructions stored in the memory 602 and calling data stored in the memory 602. Optionally, the processor 601 can include one or more processing units, and the processor 601 can integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, an application program and the like, and the modem processor mainly processes an issued instruction. It can be understood that the modem processor can also not be integrated into the processor 601. In some embodiments, the processor 601 and the memory 602 can be implemented on the same chip, and in some embodiments, they can also be implemented on separate chips respectively.

[0128] The processor 601 can be a general-purpose processor, for example, a central processing unit (CPU), a digital signal processor, an application specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in combination with the data processing method embodiment can be directly embodied as hardware processor execution or executed by a combination of hardware and software modules in the processor.

[0129] The memory 602, as a non-volatile computer readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 602 can include at least one type of storage medium, for example, can include flash memory, hard disk, multimedia card, card type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage, magnetic disk, optical disk, etc. The memory 602 is any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but is not limited thereto. The memory 602 in the embodiments of the present application can also be a circuit or any other device capable of realizing a storage function, used to store program instructions and / or data.

[0130] Based on the same technical concept, the embodiments of the present application also provide a computer readable storage medium storing a computer program executable by a computing device, which, when running on the computing device, causes the computing device to perform the steps of the address matching method.

[0131] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

[0132] Those skilled in the art will appreciate that embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer usable program code.

[0133] The embodiments of methods, devices (systems), and computer program products of the application can be described in reference to flowchart illustrations and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products. Figure 1 one or more functions specified in the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products. Figure 1 one or more functions specified in the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products.

[0134] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products. Figure 1 one or more functions specified in the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products. Figure 1 one or more functions specified in the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products.

[0135] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products. Figure 1 one or more functions specified in the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products. Figure 1 one or more functions specified in the flowchart and / or block diagrams of the flowchart and / or block diagrams of the methods, devices (systems), and computer program products.

[0136] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the embodiments by those of skill in the art once they have the benefit of the present disclosure. Therefore, the appended claims are intended to encompass within their scope all such variations and modifications as are within the scope of the application. It should be understood that all the terms used herein are descriptive rather than limiting, and that many changes can be made to the preferred embodiments, while still obtaining the intended results.

[0137] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. An address matching method, characterized in that, include: Get the address to be matched; The address to be matched is segmented according to the pre-stored dictionary to obtain the first address element; The step of segmenting the address to be matched according to the pre-stored dictionary includes: adding the pre-stored dictionary to the dictionary of the word segmentation model, and then using the updated word segmentation model to segment the address to be matched; the pre-stored dictionary includes a second address element extracted from the address to be matched through a named entity recognition model, and a basic word segmentation dictionary; The first address element is subjected to address standardization processing to obtain a standardized address to be matched; wherein, the address standardization processing includes alias standardization mapping, hierarchical tracking mapping, and hierarchical backtracking mapping; the alias standardization mapping is used to map the name of a non-standard administrative division element to the name of a standard administrative division element, the hierarchical tracking mapping is used to obtain the administrative division level to which the first four levels of standardized administrative division elements belong, and the hierarchical backtracking mapping is used to complete the administrative division elements; The target address of the standardized address to be matched is matched from the preset database.

2. The method as described in claim 1, characterized in that, Before segmenting the address to be matched according to the pre-stored dictionary, the process also includes: The address to be matched is encoded to obtain the feature information of the address to be matched; The feature information of the address to be matched is decoded to obtain the second address element.

3. The method as described in claim 1, characterized in that, The step of performing address normalization on the first address element to obtain a normalized address to be matched includes: Based on the administrative division element standard, traverse the first address element and map the non-standardized elements in the first address element to standardized elements; Based on the standardized elements, the hierarchical backtracking mapping relationship in the basic word segmentation dictionary, and the hierarchical tracking mapping relationship, the standardized address to be matched is obtained.

4. The method as described in claim 1, characterized in that, The standardized address to be matched includes a first detailed address and a first point of interest; The step of matching the standardized target address from the preset database includes: Based on the standardized administrative division field of the address to be matched, at least one second detailed address and at least one second point of interest are selected from the preset database, wherein the second detailed address corresponds to the second point of interest; Match the second detailed address with the first detailed address; if the matching result is not empty, determine the first similarity between the first point of interest and the second point of interest; if the first similarity is greater than a first threshold, use the second detailed address and the second point of interest as the target address; or... The first point of interest is matched with the second point of interest; if the matching result is not empty, the second similarity between the second detailed address and the first detailed address is determined; if the second similarity is greater than the second threshold, the second detailed address and the second point of interest are used as the target address.

5. The method as described in claim 4, characterized in that, If the first similarity is less than or equal to a first threshold, or if the second similarity is less than or equal to a second threshold, or if the matching result between the first interest point and the second interest point is empty, the step of matching the standardized target address of the address to be matched from the preset database further includes: The first point of interest is segmented into words, and the keywords of the first point of interest are determined based on the segmented words. The keyword is matched with the at least one second point of interest. If the matching result is not empty, the third similarity between the second detailed address and the first detailed address is determined. If the third similarity is greater than the third threshold, the second detailed address and the second point of interest are used as the target address.

6. The method as described in claim 5, characterized in that, If the third similarity is less than or equal to the third threshold, or if the keyword matches the at least one second point of interest (POI) without any matching result, the step of matching the standardized target address of the address to be matched from the preset database further includes: A fourth similarity is determined between the first detailed address and the at least one second detailed address, and a fifth similarity is determined between the first point of interest and the at least one second point of interest. The fourth similarity and the fifth similarity are weighted and calculated to obtain a sixth similarity. Based on the highest similarity in the sixth similarity, the corresponding second detailed address and second point of interest are determined, and the second detailed address and second point of interest are used as the target address.

7. An address matching device, characterized in that, include: The acquisition unit is used to obtain the address to be matched; The processing unit is configured to segment the address to be matched according to a pre-stored lexicon to obtain a first address element; specifically, the processing unit is configured to: add the pre-stored lexicon to the lexicon of the word segmentation model, and then segment the address to be matched using an updated word segmentation model; the pre-stored lexicon includes a second address element extracted from the address to be matched through a named entity recognition model, and a basic word segmentation lexicon; the processing unit is also configured to perform address standardization processing on the first address element to obtain a standardized address to be matched; and to match the target address of the standardized address to be matched from a preset database; wherein, the address standardization processing includes alias standardization mapping, hierarchical tracking mapping, and hierarchical backtracking mapping; the alias standardization mapping is used to map the name of a non-standard administrative division element to the name of a standard administrative division element, the hierarchical tracking mapping is used to obtain the administrative division level to which the first four levels of standardized administrative division elements belong, and the hierarchical backtracking mapping is used to complete the administrative division elements.

8. A computing device, characterized in that, The method includes at least one processor and at least one memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a program that, when run on a computer, causes the computer to perform the method described in any one of claims 1 to 6.

Images