Address detection and processing method and device based on large language model and medium

Abstract

The invention relates to the technical field of address processing and the field of financial science and technology, and provides an address detection and processing method and device based on a large language model, and a medium, and the method comprises the steps: obtaining an address information list, and inputting the address information list into the large language model, judging whether the address information contains regional address information matched with a preset address database or not by utilizing a large language model, and if so, extracting detailed address information; if not, the regional address information is corrected through a model according to a preset address database, and then a detailed address is extracted; secondly, determining a target standardized address corresponding to each piece of address information based on the region address and the detailed address, and further generating a list containing all the target standardized addresses; according to the embodiment of the invention, the address information is extracted and normalized by using the large language model, so that the KYC governance level of a financial institution is improved, and potential risks caused by inaccurate address information are reduced.

Description

Technical Field

[0001] The present disclosure relates to the field of address processing technology and the field of financial technology, and in particular, to a method, device and medium for address detection and processing based on a large language model. Background Art

[0002] With the development of science and technology, the standardized processing of detailed address information in the financial technology field is crucial, which is directly related to the risk management and business accuracy of financial institutions. At present, the industry mainly relies on identifying the three-level administrative divisions (province / municipality / autonomous region, city / autonomous prefecture / region / league, district / county / (county-level) city / banner) issued by the National Civil Affairs Bureau to initially realize the structuring of address information. For example, the address of Nanshan District, Shenzhen City, Guangdong Province can be accurately parsed, providing a basis for subsequent business processing.

[0003] However, the constraints of the three-level administrative divisions seem too loose in the field of financial technology. The address range is still too large at the district / county level, which brings many inconveniences to financial institutions in customer identity authentication (Know Your Customer, KYC) governance. KYC is a series of review and verification procedures conducted by financial institutions to ensure the authenticity of customer identities and the legality of transactions. In particular, when residential addresses need to be detailed to the specific household number, while rural addresses only need to be detailed to the building, traditional address standard detection methods are difficult to cope with such diversity. Summary of the invention

[0004] The embodiments of the present disclosure at least provide an address detection and processing method, device and medium based on a large language model. By utilizing a large language model to extract and normalize address information, the KYC governance level of financial institutions is improved and the potential risks caused by inaccurate address information are reduced.

[0005] The present disclosure provides an address detection and processing method based on a large language model, including:

[0006] Acquire an address information list, and input the address information list into a large language model; wherein the address information list includes a plurality of address information;

[0007] For each address information in the address information list, judging based on the large language model whether the address information contains regional address information matching a preset address database, wherein the regional address information includes three levels of administrative divisions;

[0008] In a case where the address information includes regional address information that matches a preset address database, extracting detailed address information from the address information based on the large language model, wherein the detailed address information includes POI address information;

[0009] In the case where the address information does not include regional address information that matches the preset address database, performing regional address information correction on the address information according to the preset address database by using the large language model, and extracting detailed address information from the address information based on the large language model;

[0010] A target normalized address corresponding to each of the address information is determined based on the area address information and the detailed address information, and a normalized address information list is determined based on the target normalized address corresponding to each of the address information; wherein the normalized address information list includes multiple target normalized addresses.

[0011] The present disclosure provides an address detection and processing device based on a large language model, including:

[0012] An information acquisition module, used to acquire an address information list and input the address information list into a large language model; wherein the address information list includes a plurality of address information;

[0013] An information determination module, for determining, for each address information in the address information list, based on a large language model, whether the address information contains regional address information that matches a preset address database, wherein the regional address information includes three levels of administrative divisions;

[0014] an information extraction module, configured to extract detailed address information from the address information based on the large language model when the address information includes regional address information that matches a preset address database, wherein the detailed address information includes POI address information;

[0015] an information correction module, configured to correct the regional address information of the address information according to the preset address database by using the large language model, and extract detailed address information from the address information based on the large language model, if the address information does not contain regional address information matching the preset address database;

[0016] An address determination module is used to determine a target normalized address corresponding to each of the address information based on the regional address information and the detailed address information, and to determine a normalized address information list based on the target normalized address corresponding to each of the address information; wherein the normalized address information list includes multiple target normalized addresses.

[0017] An embodiment of the present disclosure provides a computer device, including: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the address detection and processing method based on a large language model as described in any possible implementation manner described above is performed.

[0018] An embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the address detection and processing method based on a large language model as described in any possible implementation manner described above is implemented.

[0019] The address detection and processing method, device and medium based on the large language model provided in the embodiments of the present disclosure include: first, obtaining an address information list and inputting it into the large language model, using the large language model to determine whether each address information in the address information list contains regional address information that matches a preset address database, and if so, extracting the detailed address information; if not, correcting the regional address information according to the preset address database through the model, and then extracting the detailed address; secondly, based on the regional address and the detailed address, determining the target normalized address corresponding to each address information, and then generating a list containing all target normalized addresses.

[0020] In this way, by introducing a large language model, the present disclosure can not only identify and structure the three-level administrative divisions, but also further extract detailed address information, thereby realizing the deep standardization of address information, and can flexibly respond to the address standardization detection needs in different scenarios, thereby improving the KYC governance level of financial institutions and reducing the potential risks caused by inaccurate address information.

[0021] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required to be cited in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0023] Figure 1A schematic diagram of an application environment of an address detection and processing method based on a large language model provided by an embodiment of the present disclosure is shown;

[0024] Figure 2 A flowchart of an address detection and processing method based on a large language model provided by an embodiment of the present disclosure is shown;

[0025] Figure 3 A flowchart of a detailed address information extraction method provided by an embodiment of the present disclosure is shown;

[0026] Figure 4 A flow chart of a method for correcting regional address information provided by an embodiment of the present disclosure is shown;

[0027] Figure 5 A flow chart of a method for determining a target normalized address provided by an embodiment of the present disclosure is shown;

[0028] Figure 6 A schematic diagram of the structure of an address detection and processing device based on a large language model provided by an embodiment of the present disclosure is shown;

[0029] Figure 7 A schematic diagram of the structure of a computer device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0031] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0032] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0033] To facilitate understanding of this embodiment, the execution subject of the address detection and processing method based on a large language model provided by the embodiment of the present disclosure is first introduced in detail. The address detection and processing method based on a large language model provided by the embodiment of the present invention can be applied to Figure 1 In an application environment, the client communicates with the server through a network. The client can be a mobile device, a user terminal, a terminal, a handheld device, a computing device, etc. The server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, big data and artificial intelligence platforms.

[0034] The following is a detailed description of the address detection and processing method based on a large language model provided in the embodiment of the present application in conjunction with the accompanying drawings. Figure 2 As shown, it is a flowchart of an address detection and processing method based on a large language model provided by an embodiment of the present disclosure, and the method includes the following S201-S205:

[0035] S201, obtaining an address information list, and inputting the address information list into a large language model.

[0036] It is understandable that the address information list includes multiple address information, which may come from the customer application form, online registration information or customer service interaction process. These address information are usually provided by customers when applying for financial services, registering accounts or conducting customer service consultations, and may include key information such as the customer's residential address, office address or business address. The big language model is a natural language processing model based on deep learning technology. In this disclosure, the big language model is used to identify and parse the key content of the address information provided by the customer, such as administrative divisions, POI address information, etc.

[0037] Specifically, in order to improve the accuracy and effectiveness of address information and reduce the financial service risks caused by incorrect or invalid addresses, after the address information list is input into the large language model, the following steps (1) to (3) may be included to pre-process the address information:

[0038] (1) for each piece of address information in the address information list, determining whether the address information is non-address text based on the large language model;

[0039] (2) when the address information is non-address text, determining the address information as an abnormal address;

[0040] (3) When the address information is not non-address text, the step of determining based on the large language model whether the address information contains area address information that matches a preset address database is performed.

[0041] It is understandable that non-address text refers to text composed of preset special characters, wherein the preset special characters include null characters, numbers and letters. Specifically, when it is determined that the address information is non-address text, it means that the information may not be correctly identified as valid address information due to input errors, data contamination or malicious input. Therefore, these address information are determined as abnormal addresses, and corresponding processing measures are taken, such as notifying the customer to re-provide the correct address information, or conducting further verification and validation. When it is determined that the address information is not non-address text, step S202 is executed to implement the processing of the address information.

[0042] In this way, the present disclosure can filter out text that obviously does not meet the address format requirements by judging whether the address information is non-address text, thereby reducing the complexity and error rate of subsequent processing and ensuring the accuracy of subsequent financial services.

[0043] S202: for each piece of address information in the address information list, determine based on a large language model whether the address information contains area address information that matches a preset address database.

[0044] Specifically, the preset address database is a database containing standard administrative division codes, which usually contains multi-level administrative division information such as province (state), city (county), district (town), etc., and specifically includes standard three-level administrative division information. Among them, the regional address information includes three-level administrative divisions, namely province / municipality / autonomous region or city / autonomous prefecture / region / league or district / county / (county-level) city / banner.

[0045] S203: When the address information includes regional address information that matches a preset address database, extract detailed address information from the address information based on the large language model.

[0046] Exemplarily, when the address information contains regional address information that matches the standard three-level administrative division information (i.e., province / municipality / autonomous region, city / autonomous prefecture / region / league, district / county / (county-level) city / banner), the large language model can be used to extract the detailed address in the address information. Here, the detailed address information usually contains a more specific and detailed address description, such as POI (Point of Interest) address information. POI address information refers to a place with clear geographic coordinates and attribute information, such as street name, house number, building name, etc. This information plays an important role in determining the specific location of the customer and providing accurate services.

[0047] Specifically, in order to obtain more accurate address information, the preset address database in the present disclosure also includes standard four- and five-level administrative division information (streets / townships / townships and communities / villages), and the detailed address information also includes POI address information and combination information of four- and five-level administrative divisions, so as to achieve more accurate positioning of the specific location required by the user. Figure 3 As shown, a detailed address information extraction method provided by the present disclosure specifically includes the following steps S2031 to S2033:

[0048] S2031, extracting POI address information from the address information based on the large language model, and determining whether the address information contains a fourth- or fifth-level administrative division that matches the standard fourth- or fifth-level administrative division information based on the large language model.

[0049] Specifically, the large language model is used to extract the address information to obtain the POI address information in the address information, and then the large language model is used to determine whether the address information contains a fourth- or fifth-level administrative division that matches the preset standard fourth- or fifth-level administrative division information.

[0050] S2032: When the address information includes a fourth- or fifth-level administrative division that matches the standard fourth- or fifth-level administrative division information, determine the detailed address information based on the fourth- or fifth-level administrative division and the POI address information.

[0051] For example, when the address information contains a level 4 or 5 administrative division that matches the standard level 4 or 5 administrative division information, the detailed address information is determined by combining the extracted POI address information and the matching level 4 or 5 administrative division. Such a combination not only contains accurate geographic location information, but also reflects the hierarchical structure and semantic integrity of the address, which helps to improve the practicality and accuracy of the address information.

[0052] S2033: When the address information does not include a fourth- or fifth-level administrative division that matches the standard fourth- or fifth-level administrative division information, determine the detailed address information based on the POI address information.

[0053] Here, in the case where the address information does not contain a fourth- or fifth-level administrative division that matches the standard fourth- or fifth-level administrative division information, the following steps (a) to (c) may be included:

[0054] (a) determining whether there is a fourth or fifth level administrative division corresponding to the POI address in the POI address database based on the large language model;

[0055] (b) when there is no fourth or fifth level administrative division corresponding to the POI address in the POI address database, determining the detailed address information based on the POI address information;

[0056] (c) when there is a fourth or fifth level administrative division corresponding to the POI address in the POI address database, determining the detailed address information based on the fourth or fifth level administrative division and the POI address information.

[0057] It is understandable that the POI address database is a database that centrally stores geographic location information, and centrally stores detailed geographic location information of various points of interest (POI) including latitude and longitude coordinates, address descriptions, etc. The POI address database disclosed in the present invention can introduce richer point of interest information by connecting with third-party data sources, thereby further improving the accuracy and practicality of the extracted address information, and providing more accurate and comprehensive geographic location services.

[0058] Specifically, if there is no level 4 or 5 administrative division corresponding to the input POI address information in the POI address database, the detailed address information can be determined directly based on the POI address information. Here, even in the absence of administrative division information, the POI address can still provide a valuable geographic location reference. At this time, the detailed address information can include the POI address information itself, as well as supplementary information that may be obtained through other means (such as map services, etc.).

[0059] Specifically, if there is a level 4 or 5 administrative division corresponding to the input POI address information in the POI address database, the detailed address information can be determined based on the determined level 4 or 5 administrative division and the POI address information. The present disclosure makes full use of the detailed administrative division information in the POI address database, so that the final detailed address information not only contains the specific location information, but also reflects the hierarchical structure and semantic integrity of the address.

[0060] In the present disclosure, by using the POI address database as a supplementary data source, it is ensured that even when the administrative division information is missing, an effective geographical location reference can be provided based on the POI address information, thereby further improving the accuracy of address information extraction.

[0061] S204: When the address information does not include regional address information that matches a preset address database, the address information is corrected in regional address information according to the preset address database by using the large language model, and detailed address information in the address information is extracted based on the large language model.

[0062] For example, refer to Figure 4 As shown, when the address information does not contain the regional address information matching the standard three-level administrative division information, the following steps S2041 to S2044 may be included:

[0063] S2041: Determine the error type of the address information based on the large language model.

[0064] Here, the error types include missing administrative divisions, duplicate administrative divisions, and administrative division adjustments. Missing administrative divisions means that the address information lacks one or several levels of administrative division information in the province / municipality / autonomous region, city / autonomous prefecture / region / league or district / county / (county-level) city / banner, making the address unable to be matched and verified according to the standard three-level administrative divisions. Duplicated administrative divisions means that the address information contains duplicate administrative division information, which may be due to input errors or redundant information generated during copying and pasting. Duplicate administrative division information may interfere with the address parsing and matching process, resulting in the address information being unable to be correctly understood and used. Administrative division adjustment means that the administrative division information in the address information is inconsistent with the latest administrative division information due to the merger, splitting, renaming or adjustment of the level of administrative divisions. Administrative division adjustment will affect the accuracy and standardization of the address information, so it needs to be updated and corrected in a timely manner.

[0065] S2042, when the error type is missing administrative division, determine the missing administrative division in the address information based on the large language model according to the standard three-level administrative division information, add the missing administrative division to the address information, and extract the regional address information in the address information after adding the missing administrative division based on the large language model according to the standard three-level administrative division information.

[0066] Specifically, when the error type is missing administrative divisions, it is necessary to use the large language model to determine the missing administrative divisions in the address information based on the standard three-level administrative division information. After determining the missing administrative divisions, add them to the original address information to obtain complete address information. Subsequently, the large language model is used again to extract the regional address information in the address information after adding the missing administrative divisions based on the standard three-level administrative division information to ensure the accuracy and standardization of the extraction results. For example: the address information is Unit 1, Building 1, Shangshui, a certain garden in Hequ County, Shanxi Province. Based on the large language model, it can be confirmed that the administrative division information of Xinzhou City is missing in the address information. At this time, "Xinzhou City" needs to be added to the original address information to obtain Unit 1, Building 1, Shangshui, a certain garden in Hequ County, Xinzhou City, Shanxi Province.

[0067] Exemplarily, if the large language model fails to successfully match or identify the missing administrative divisions in the address information based on the standard three-level administrative division information, it means that the address information may not be accurately verified or located. At this time, the address information is determined to be an abnormal address. For example, the address information is "XX Road, No. XX". This address information obviously lacks key administrative division information, such as province, city, district, etc. When the large language model matches or identifies the missing administrative divisions in this address based on the standard three-level administrative division information, due to the lack of specific administrative division information, even if the road name and house number information in the address are accurate, it cannot be corresponded to the specific geographical location, so it is determined to be an abnormal address.

[0068] S2043, when the error type is administrative division duplication, determine the duplicate administrative divisions in the address information based on the large language model according to the standard three-level administrative division information, delete the duplicate administrative divisions from the address information, and extract the area address information in the address information after deleting the duplicate administrative divisions based on the large language model according to the standard three-level administrative division information.

[0069] Specifically, when the error type is administrative division duplication, the large language model is used to determine the duplicate administrative divisions in the address information based on the standard three-level administrative division information. After identifying the duplicate administrative divisions, they are deleted from the address information to avoid redundancy and confusion. Subsequently, the large language model is used again to extract the regional address information in the address information after deleting the duplicate administrative divisions to ensure the accuracy and standardization of the extraction results. For example: the address information is Building 1 of a certain industrial base in a certain Hai Street, Nanshan District, Shenzhen City, Guangdong Province. "Nanshan District, Shenzhen City" appears twice in the address information, which is an administrative division duplication error type. The duplicate item of "Nanshan District, Shenzhen City" needs to be deleted from the original address information to obtain Building 1 of a certain industrial base in a certain Hai Street, Nanshan District, Shenzhen City, Guangdong Province.

[0070] S2044, when the error type is administrative division adjustment, based on the large language model and according to the standard three-level administrative division information, determine the target administrative division corresponding to the administrative division to be adjusted in the address information, and based on the target administrative division, update the administrative division to be adjusted in the address information, and based on the large language model and according to the standard three-level administrative division information, extract the regional address information in the updated address information.

[0071] Specifically, when the error type is administrative division adjustment, the target administrative division corresponding to the administrative division to be adjusted in the address information is determined using the large language model based on the standard three-level administrative division information. After the target administrative division is determined, the administrative division to be adjusted in the address information is updated to reflect the latest administrative division information; then, the large language model is used again to extract the regional address information in the updated address information. For example: the address information is: Building 1, a community in Bin County, Xianyang City, Shaanxi Province. Due to administrative division adjustment, "Bin County" in the standard three-level administrative division information has been changed to "Binzhou City". At this time, the original address information needs to be adjusted to "Building 1, a community in Binzhou City, Xianyang City, Shaanxi Province".

[0072] Exemplarily, when the large language model fails to determine the target administrative division corresponding to the administrative division to be adjusted in the address information based on the standard three-level administrative division information, the address information is determined to be an abnormal address.

[0073] In this disclosure, the accuracy and standardization of address information are improved by using a large language model to judge the error type in address information and correcting errors such as missing administrative divisions, duplications, and adjustments. At the same time, for situations where successful matching or recognition cannot be achieved, the address information is determined to be an abnormal address, aiming to reduce the potential risks caused by inaccurate or incomplete address information, and further ensure the accuracy of subsequent processing.

[0074] Specifically, after the address information is corrected for regional address information according to a preset address database through a large language model, detailed address information in the corrected address information can be extracted based on the large language model. For details, please refer to the above steps S2031 to S2033, which will not be repeated here.

[0075] S205: Determine a target normalized address corresponding to each piece of address information based on the area address information and the detailed address information, and determine a normalized address information list based on the target normalized address corresponding to each piece of address information.

[0076] It is understandable that in order to more accurately achieve the standardization of address information and improve the consistency and availability of data, refer to Figure 5As shown, when determining the target normalized address corresponding to each address information based on the regional address information and the detailed address information, the following steps S2051 to S2053 may be included:

[0077] S2051: Determine a normalized address corresponding to each of the address information based on the regional address information and the detailed address information, and determine whether each of the normalized addresses meets a preset normalization standard based on the large language model.

[0078] Specifically, the regional address information and detailed address information obtained in the previous steps are combined to obtain a standardized address corresponding to each address information, that is, the third-level administrative division + the fourth and fifth-level administrative divisions + POI address or the third-level administrative division + POI address. Then use the large language model to analyze it to determine whether it meets the preset normalization standards. Here, the preset normalization standards are usually formulated according to the actual application scenarios and the characteristics of the address data, aiming to ensure the accuracy and consistency of the address information. For example, in an urban environment, a residential address is generally detailed to a specific household number in a building, so that it is convenient for people to visit, etc.; in relatively remote places such as rural areas, due to the small number of buildings, the address information only needs to be detailed to a certain building.

[0079] S2052: When the normalized address meets a preset normalization standard, determine the normalized address as a target normalized address corresponding to the address information.

[0080] Here, if the normalized address meets the preset normalization standard, that is, the normalized address contains the necessary address elements, and has the correct format and clear semantics, then the normalized address can be considered to meet the normalization requirements and be determined as the target normalized address corresponding to the address information.

[0081] S2053: When the normalized address does not meet the preset normalization standard, the address information corresponding to the normalized address is determined as an abnormal address.

[0082] Here, if the standardized address does not meet the preset standardization standards, that is, the standardized address cannot ensure accurate access by personnel due to the lack of necessary information (such as the lack of street name or house number) or semantic ambiguity (such as the southern section of Chengdong Avenue, XX County, XX City, XX Province), then, since such address information cannot provide accurate navigation guidance, the address information corresponding to the standardized address needs to be judged as an abnormal address.

[0083] Specifically, after judging the normalized address corresponding to each address information, based on the judgment result, the target normalized address corresponding to each address information is integrated into a normalized address information list. Here, the normalized address information list includes multiple target normalized addresses.

[0084] In some other embodiments, the large language model not only outputs a list of normalized address information including multiple target normalized addresses, but also outputs a list containing all abnormal addresses. Here, the list containing all abnormal addresses can record the original information of each abnormal address and the specific reason why the address is determined to be abnormal. In this way, it not only helps the administrator to identify and handle these abnormal addresses, but also provides reference information for subsequent address data correction and optimization.

[0085] The address detection and processing method, device and medium based on the large language model provided in the embodiments of the present disclosure, by introducing the large language model, can not only identify and structure the three-level administrative divisions, but also can further extract detailed address information, thereby realizing the deep standardization of address information, and can flexibly respond to the address standardization detection needs in different scenarios, thereby improving the KYC governance level of financial institutions and reducing the potential risks caused by inaccurate address information.

[0086] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0087] Based on the same inventive concept, the embodiment of the present disclosure also provides an address detection and processing device based on a large language model corresponding to the address detection and processing method based on a large language model. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned address detection and processing method based on a large language model in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0088] Reference Figure 6 FIG. 1 is a schematic diagram of an address detection and processing device 600 based on a large language model provided in an embodiment of the present disclosure, wherein the device comprises:

[0089] The information acquisition module 601 is used to acquire an address information list and input the address information list into the large language model; wherein the address information list includes a plurality of address information;

[0090] An information determination module 602 is used to determine, for each address information in the address information list, whether the address information contains regional address information matching a preset address database based on a large language model, wherein the regional address information includes three levels of administrative divisions;

[0091] An information extraction module 603 is used to extract detailed address information from the address information based on the large language model when the address information includes regional address information that matches a preset address database, wherein the detailed address information includes POI address information;

[0092] An information correction module 604 is used to correct the regional address information of the address information according to the preset address database by using the large language model, and extract detailed address information from the address information based on the large language model when the address information does not contain regional address information matching the preset address database;

[0093] The address determination module 605 is used to determine the target normalized address corresponding to each of the address information based on the regional address information and the detailed address information, and to determine a normalized address information list based on the target normalized address corresponding to each of the address information; wherein the normalized address information list includes multiple target normalized addresses.

[0094] In some possible embodiments, the information acquisition module 601 is further used to:

[0095] For each address information in the address information list, determining whether the address information is non-address text based on the large language model; wherein the non-address text is text composed of preset special characters, wherein the preset special characters include a blank character, a number, and a letter;

[0096] In the case where the address information is non-address text, determining the address information as an abnormal address;

[0097] In a case where the address information is not non-address text, the step of determining based on the large language model whether the address information includes regional address information that matches a preset address database is performed.

[0098] In some possible embodiments, the preset address database includes standard three-level administrative division information; the information correction module 604 is specifically used to:

[0099] Determining the error type of the address information based on the large language model; wherein the error type includes missing administrative divisions, repeated administrative divisions, and administrative division adjustments;

[0100] In the case where the error type is missing administrative divisions, determining the missing administrative divisions in the address information based on the large language model according to the standard three-level administrative division information, adding the missing administrative divisions to the address information, and extracting the regional address information in the address information after adding the missing administrative divisions based on the large language model according to the standard three-level administrative division information;

[0101] In the case where the error type is administrative division duplication, determining the duplication administrative division in the address information based on the large language model according to the standard three-level administrative division information, and deleting the duplication administrative division from the address information, and extracting the regional address information from the address information after deleting the duplication administrative division based on the large language model according to the standard three-level administrative division information;

[0102] In the case where the error type is administrative division adjustment, the target administrative division corresponding to the administrative division to be adjusted in the address information is determined based on the large language model according to the standard three-level administrative division information, and the administrative division to be adjusted in the address information is updated based on the target administrative division, and the regional address information in the updated address information is extracted based on the large language model according to the standard three-level administrative division information.

[0103] In some possible embodiments, the information correction module 604 is further configured to:

[0104] If the large language model fails to determine the missing administrative division in the address information according to the standard three-level administrative division information, determining the address information as an abnormal address;

[0105] The information correction module 604 is also used for:

[0106] When the large language model fails to determine the target administrative division corresponding to the administrative division to be adjusted in the address information according to the standard three-level administrative division information, the address information is determined to be an abnormal address.

[0107] In some possible embodiments, the preset address database further includes standard fourth- and fifth-level administrative division information, and the detailed address information further includes combined information of POI address information and fourth- and fifth-level administrative divisions; the information extraction module 603 is specifically used to:

[0108] Extracting POI address information from the address information based on the large language model, and determining whether the address information contains a level 4 or 5 administrative division that matches the standard level 4 or 5 administrative division information based on the large language model;

[0109] In a case where the address information includes a level 4 or 5 administrative division that matches the standard level 4 or 5 administrative division information, determining the detailed address information based on the level 4 or 5 administrative division and the POI address information;

[0110] In a case where the address information does not include a fourth- or fifth-level administrative division that matches the standard fourth- or fifth-level administrative division information, the detailed address information is determined based on the POI address information.

[0111] In some possible embodiments, the information extraction module 603 is specifically used to:

[0112] Based on the large language model, it is determined whether there is a fourth or fifth level administrative division corresponding to the POI address in the POI address database;

[0113] In the case that there is no fourth or fifth level administrative division corresponding to the POI address in the POI address database, determining the detailed address information based on the POI address information;

[0114] In the case where there are fourth- and fifth-level administrative divisions corresponding to the POI address in the POI address database, the detailed address information is determined based on the fourth- and fifth-level administrative divisions and the POI address information.

[0115] In some possible embodiments, the address determination module 605 is specifically used to:

[0116] Determine a normalized address corresponding to each of the address information based on the regional address information and the detailed address information, and determine whether each of the normalized addresses meets a preset normalization standard based on the large language model;

[0117] In the case where the normalized address meets the preset normalization standard, determining the normalized address as the target normalized address corresponding to the address information;

[0118] In the case that the normalized address does not meet the preset normalization standard, the address information corresponding to the normalized address is determined to be an abnormal address.

[0119] Based on the same technical concept, the embodiment of the present disclosure also provides a computer device. Figure 7 , which is a schematic diagram of the structure of a computer device 700 provided in an embodiment of the present disclosure, including a processor 701, a memory 702, and a bus 703. The memory 702 is used to store execution instructions, including a memory 7021 and an external memory 7022; the memory 7021 is also called an internal memory, which is used to temporarily store the operation data in the processor 701 and the data exchanged with the external memory 7022 such as a hard disk. The processor 701 exchanges data with the external memory 7022 through the memory 7021.

[0120] In the embodiment of the present application, the memory 702 is specifically used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 701. That is, when the computer device 700 is running, the processor 701 communicates with the memory 702 through the bus 703, so that the processor 701 executes the application code stored in the memory 702, and then executes the method described in any of the above embodiments.

[0121] Among them, the memory 702 can be, but is not limited to, random access memory (Random Access Memory, RAM), read only memory (Read Only Memory, ROM), programmable read-only memory (Programmable Read-Only Memory, PROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, EPROM), electrically erasable read-only memory (Electric Erasable Programmable Read-Only Memory, EEPROM), etc.

[0122] Processor 701 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present invention may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0123] It is to be understood that the structure illustrated in the embodiment of the present application does not constitute a specific limitation on the computer device 700. In other embodiments of the present application, the computer device 700 may include more or fewer components than shown in the figure, or combine some components, or split some components, or arrange the components differently. The components shown in the figure may be implemented in hardware, software, or a combination of software and hardware.

[0124] The present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the address detection and processing method based on a large language model described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0125] The embodiments of the present disclosure also provide a computer program product, which carries a program code. The instructions included in the program code can be used to execute the steps of the address detection and processing method based on a large language model described in the above method embodiment. For details, please refer to the above method embodiment, which will not be repeated here.

[0126] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0127] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0128] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0129] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0130] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0131] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for address detection and processing based on a large language model, characterized in that: include: Acquire an address information list, and input the address information list into a large language model; wherein the address information list includes a plurality of address information; For each address information in the address information list, judging based on the large language model whether the address information contains regional address information matching a preset address database, wherein the regional address information includes three levels of administrative divisions; In a case where the address information includes regional address information that matches a preset address database, extracting detailed address information from the address information based on the large language model, wherein the detailed address information includes POI address information; In the case where the address information does not include regional address information that matches the preset address database, performing regional address information correction on the address information according to the preset address database by using the large language model, and extracting detailed address information from the address information based on the large language model; A target normalized address corresponding to each of the address information is determined based on the area address information and the detailed address information, and a normalized address information list is determined based on the target normalized address corresponding to each of the address information; wherein the normalized address information list includes multiple target normalized addresses.

2. The method according to claim 1, characterized in that After inputting the address information list into the large language model, the method further includes: For each address information in the address information list, determining whether the address information is non-address text based on the large language model; wherein the non-address text is text composed of preset special characters, wherein the preset special characters include a blank character, a number, and a letter; In the case where the address information is non-address text, determining the address information as an abnormal address; In a case where the address information is not non-address text, the step of determining based on the large language model whether the address information includes regional address information that matches a preset address database is performed.

3. The method according to claim 2, characterized in that The preset address database includes standard three-level administrative division information; the address information is corrected according to the preset address database by the large language model, and detailed address information in the corrected address information is extracted based on the large language model, including: Determining the error type of the address information based on the large language model; wherein the error type includes missing administrative divisions, repeated administrative divisions, and administrative division adjustments; In the case where the error type is missing administrative divisions, determining the missing administrative divisions in the address information based on the large language model according to the standard three-level administrative division information, adding the missing administrative divisions to the address information, and extracting the regional address information in the address information after adding the missing administrative divisions based on the large language model according to the standard three-level administrative division information; In the case where the error type is administrative division duplication, determining the duplication administrative division in the address information based on the large language model according to the standard three-level administrative division information, and deleting the duplication administrative division from the address information, and extracting the regional address information from the address information after deleting the duplication administrative division based on the large language model according to the standard three-level administrative division information; In the case where the error type is administrative division adjustment, the target administrative division corresponding to the administrative division to be adjusted in the address information is determined based on the large language model according to the standard three-level administrative division information, and the administrative division to be adjusted in the address information is updated based on the target administrative division, and the regional address information in the updated address information is extracted based on the large language model according to the standard three-level administrative division information.

4. The method according to claim 3, characterized in that The determining, according to the standard three-level administrative division information and based on the large language model, the missing administrative divisions in the address information further includes: If the large language model fails to determine the missing administrative division in the address information according to the standard three-level administrative division information, determining the address information as an abnormal address; The step of determining the target administrative division corresponding to the administrative division to be adjusted in the address information based on the large language model and the standard three-level administrative division information further includes: When the large language model fails to determine the target administrative division corresponding to the administrative division to be adjusted in the address information according to the standard three-level administrative division information, the address information is determined to be an abnormal address.

5. The method according to claim 3, characterized in that: The preset address database also includes standard fourth- and fifth-level administrative division information, and the detailed address information also includes POI address information and combination information of fourth- and fifth-level administrative divisions; and the detailed address information extracted from the address information based on the large language model includes: Extracting POI address information from the address information based on the large language model, and determining whether the address information contains a level 4 or 5 administrative division that matches the standard level 4 or 5 administrative division information based on the large language model; In a case where the address information includes a level 4 or 5 administrative division that matches the standard level 4 or 5 administrative division information, determining the detailed address information based on the level 4 or 5 administrative division and the POI address information; In a case where the address information does not include a fourth- or fifth-level administrative division that matches the standard fourth- or fifth-level administrative division information, the detailed address information is determined based on the POI address information.

6. The method according to claim 5, characterized in that The determining the detailed address information based on the POI address information includes: Based on the large language model, it is determined whether there is a fourth or fifth level administrative division corresponding to the POI address in the POI address database; In the case that there is no fourth or fifth level administrative division corresponding to the POI address in the POI address database, determining the detailed address information based on the POI address information; In the case where there are fourth- and fifth-level administrative divisions corresponding to the POI address in the POI address database, the detailed address information is determined based on the fourth- and fifth-level administrative divisions and the POI address information.

7. The method according to claim 1, characterized in that The determining a target normalized address corresponding to each of the address information based on the regional address information and the detailed address information comprises: Determine a normalized address corresponding to each of the address information based on the regional address information and the detailed address information, and determine whether each of the normalized addresses meets a preset normalization standard based on the large language model; In the case where the normalized address meets the preset normalization standard, determining the normalized address as the target normalized address corresponding to the address information; In the case that the normalized address does not meet the preset normalization standard, the address information corresponding to the normalized address is determined to be an abnormal address.

8. An address detection and processing device based on a large language model, characterized in that: include: An information acquisition module, used to acquire an address information list and input the address information list into a large language model; wherein the address information list includes a plurality of address information; An information determination module, for determining, for each address information in the address information list, based on a large language model, whether the address information contains regional address information that matches a preset address database, wherein the regional address information includes three levels of administrative divisions; an information extraction module, configured to extract detailed address information from the address information based on the large language model when the address information includes regional address information that matches a preset address database, wherein the detailed address information includes POI address information; an information correction module, configured to correct the regional address information of the address information according to the preset address database by using the large language model, and extract detailed address information from the address information based on the large language model, if the address information does not contain regional address information matching the preset address database; An address determination module is used to determine a target normalized address corresponding to each of the address information based on the regional address information and the detailed address information, and to determine a normalized address information list based on the target normalized address corresponding to each of the address information; wherein the normalized address information list includes multiple target normalized addresses.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images