Address information extraction method, device, computer equipment and storage medium

Abstract

The present application relates to an address information extraction method, device, computer equipment and storage medium. The method obtains an express waybill to be identified; extracts the original address information of the express waybill to be identified; marks the hierarchical address information in the original address information to obtain the marked address information; and obtains the standard address information corresponding to the original address information based on the marked address information and a preset key address set. The present application extracts the original address in the express waybill, and then determines the standard address information corresponding to the original address information based on the hierarchical address information marking and standard address recognition. This can effectively improve the address information recognition efficiency during the express delivery process, avoid the situation where the address information is unclear due to typos by the customer, and the delivery process cannot proceed normally, affecting the express delivery efficiency.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, computer device, and storage medium for extracting address information. Background Art

[0002] With the rapid development of computer and logistics technologies, express delivery has become an integral part of people's lives. Express delivery, also known as courier or express transportation, refers to a new mode of transportation in which logistics companies (including freight forwarders) use their own independent networks or collaborative partnerships (i.e., networking) to quickly and securely deliver documents or parcels entrusted by users from sender to recipient's door-to-door (hand-to-hand). Express delivery typically requires the shipper to complete a waybill, which contains information such as the sender, consigned goods, and destination address.

[0003] However, since the address information on waybills is typically the original text filled in by the customer, this text exhibits significant diversity and handwriting variations due to differences in filling habits and address granularity. Currently, the address information on waybill labels is the customer's input. Irregularities in this information significantly impact operator efficiency, and thus, delivery timelines. Summary of the Invention

[0004] Based on this, it is necessary to provide an address information extraction method, device, computer equipment and storage medium that can effectively improve the efficiency of address information recognition in order to address the above technical problems.

[0005] A method for extracting address information, the method comprising:

[0006] Get the express waybill to be identified;

[0007] Extracting the original address information of the express waybill to be identified;

[0008] Marking the hierarchical address information in the original address information to obtain marked address information;

[0009] According to the marked address information and a preset key address set, standard address information corresponding to the original address information is obtained.

[0010] In one embodiment, obtaining the standard address information in the original address information according to the marked address and the preset key address set includes:

[0011] Obtaining the similarity between the key address in the marked address information and the standard key address in the preset key address set;

[0012] identifying, based on the similarity, a standard key address corresponding to the annotated address information;

[0013] According to the marked address information and the standard key address, standard address information corresponding to the original address information is obtained.

[0014] In one embodiment, obtaining the similarity between the marked address information and a standard key address in a preset key address set includes:

[0015] identifying, according to the annotations in the annotated address information, an interest surface address and an annotated key address in the annotated address information;

[0016] Searching for a preset key address set corresponding to the interest surface address;

[0017] The similarity between the marked key address and the standard key address in the preset key address set is obtained by using the fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

[0018] In one embodiment, before obtaining the similarity between the marked address information and a standard key address in a preset key address set, the method further includes:

[0019] Obtain historical express waybill data with hierarchical address information;

[0020] Identifying, based on the hierarchical address information annotation, an address of interest in the historical express waybill data, and an address combination of key addresses corresponding to the address of interest;

[0021] The address combinations with the same interest surface address are clustered to construct a preset key address set corresponding to each interest surface address.

[0022] In one embodiment, before clustering the address combinations with the same interest plane addresses and constructing the preset key address sets corresponding to the interest plane addresses, the method further includes:

[0023] Determine the frequency of occurrence of each address combination in the historical express waybill data and the similarity between each address combination;

[0024] Cleaning the address combinations according to the occurrence frequency and the similarity;

[0025] Clustering the address combinations with the same interest surface addresses to construct a preset key address set corresponding to each interest surface address includes:

[0026] The address combinations with the same cleaned interest surface addresses are clustered to construct a preset key address set corresponding to each interest surface address.

[0027] In one embodiment, after obtaining the standard address information corresponding to the original address information based on the marked address information and the preset key address set, the method further includes:

[0028] The standard key address in the standard address information is marked.

[0029] In one embodiment, the marking of the hierarchical address information in the original address information to obtain the marked address information includes:

[0030] The hierarchical address information in the original address information is labeled by a preset LSTM (Long Short-Term Memory, long short-term memory artificial neural network) information labeling model to obtain the labeled address. The preset LSTM information labeling model is obtained by training historical express waybill data with hierarchical address information labeled.

[0031] An address information extraction device, comprising:

[0032] Waybill acquisition module, used to obtain the express waybill to be identified;

[0033] An address extraction module, used to extract the original address information of the express waybill to be identified;

[0034] An address information marking module, configured to mark the hierarchical address information in the original address information to obtain marked address information;

[0035] The standard address identification module is used to obtain the standard address information corresponding to the original address information according to the marked address information and a preset key address set.

[0036] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0037] Get the express waybill to be identified;

[0038] Extracting the original address information of the express waybill to be identified;

[0039] Marking the hierarchical address information in the original address information to obtain marked address information;

[0040] According to the marked address information and a preset key address set, standard address information corresponding to the original address information is obtained.

[0041] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0042] Get the express waybill to be identified;

[0043] Extracting the original address information of the express waybill to be identified;

[0044] Marking the hierarchical address information in the original address information to obtain marked address information;

[0045] According to the marked address information and a preset key address set, standard address information corresponding to the original address information is obtained.

[0046] The above-mentioned address information extraction method, device, computer equipment and storage medium obtain the express waybill to be identified; extract the original address information of the express waybill to be identified; mark the hierarchical address information in the original address information to obtain the marked address information; and obtain the standard address information corresponding to the original address information based on the marked address information and the preset key address set. This application extracts the original address in the express waybill and then determines the standard address information corresponding to the original address information based on the hierarchical address information marking and standard address recognition. This can effectively improve the address information recognition efficiency during the express delivery process, avoid the situation where the address information is unclear due to typos by the customer, which makes the delivery process unable to proceed normally and affects the express delivery efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A diagram showing an application environment of an address information extraction method in one embodiment;

[0048] Figure 2 1 is a flow chart of a method for extracting address information in one embodiment;

[0049] Figure 3 In one embodiment Figure 2 Schematic diagram of the sub-process of step 207;

[0050] Figure 4 In one embodiment Figure 3 Schematic diagram of the sub-process of step 302;

[0051] Figure 5 A flowchart illustrating the steps of constructing a preset key address set corresponding to each interest surface address in one embodiment;

[0052] Figure 6 is a structural block diagram of an address information extraction device in one embodiment;

[0053] Figure 7 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0055] The address information extraction method provided in this application can be applied to Figure 1 In the application environment shown. The terminal 102 communicates with the server 104 via a network. The terminal 102 can send an express waybill to be identified to the server 104 so that the server 104 can extract the address information in the express waybill to be identified. The server 104 obtains the express waybill to be identified; extracts the original address information of the express waybill to be identified; annotates the hierarchical address information in the original address information to obtain the annotated address information; and obtains the standard address information corresponding to the original address information based on the annotated address information and a preset key address set. The terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices, and the server 104 can be implemented as an independent server or a server cluster consisting of multiple servers.

[0056] In one embodiment, Figure 2 As shown, a method for extracting address information is provided, which is applied to Figure 1 Taking the server 104 in the example as an example, the following steps are included:

[0057] Step 201: Obtain the express waybill to be identified.

[0058] Among them, the express waybill to be identified refers to the information extraction object of the address information extraction method of this application. The express waybill to be identified refers to the waybill information filled out by the user, including the delivery outlet, receiving address, type of items to be sent, and recipient information filled in by the user, wherein the receiving address specifically includes AOI (area of ​​interest) area information and POI (Point of Interest) information. Since the information in the express waybill to be identified is completely filled in manually by the sender personally, due to differences in filling habits and detailed addresses, the text information presents obvious diversity and writing differences. Irregular filled-in content greatly affects the operational efficiency of relevant operators, and thus affects the overall delivery timeliness and customer experience. Therefore, the address information can be extracted by the address information extraction method of this application, and converted into standard address information, thereby improving the operational efficiency of the express identification process.

[0059] First, when terminal 102 needs to extract the address information in the waybill, it can perform the address information extraction work through server 104. First, the express waybill to be identified completed by the user needs to be sent from terminal 102 to server 104. The express waybill to be identified can specifically refer to the electronic version of the express waybill filled out by the user or the pickup staff.

[0060] Step 203: extract the original address information of the express waybill to be identified.

[0061] The original address information refers to the recipient address filled in manually by the sender. Due to differences in filling habits and address details, the text in the original address information exhibits significant diversity and handwriting differences. The extraction process specifically involves identifying the address area on the express waybill to be identified and then obtaining the content in the address area on the electronic waybill. This content is the original address information.

[0062] Step 205: Mark the hierarchical address information in the original address information to obtain marked address information.

[0063] Hierarchical address information refers to an address containing information at different levels, such as province, city, district (county), street, and community. In one embodiment, the hierarchical address specifically includes: prov (province) > city (city) > district (district) > town (town) > comm (comm) > road (street) > roadno (street number) > poi (point of interest) > (subpoi (sub-point of interest) >) houseno (building number) > cellno (cell number) > floorno (floor) > roomno (room number). A POI (area of ​​point) specifically refers to an address of interest or key address, i.e., key information such as a building or market corresponding to the recipient address. The address information completed by the sender can be broken down to correspond to each segment of the aforementioned hierarchical address. For example, if the original address information of the destination address of a parcel is No. J, Floor G, Unit H, Floor I, F Building, District C, Street D, Community E, Province A, after splitting and labeling the address information, you can label Province A through prov, City B through city, and Building F through POI, etc. By labeling, you can more effectively extract a valid standard address from the original address information.

[0064] Specifically, by labeling the original address information, it is possible to determine which level of the address hierarchy each piece of information in the original address information belongs to. This allows for more accurate address information extraction in the future. In a specific embodiment, the hierarchical address information extraction can be accomplished using a trained neural network model.

[0065] Step 207: Acquire standard address information corresponding to the original address information based on the marked address information and the preset key address set.

[0066] The preset key address set refers to a standard key address set built into server 104, which can serve as a basis for delivery personnel to determine the actual address during delivery. Standard address information is derived from the marked address information using the preset key address set. This eliminates the problem of address recognition errors caused by differences in sender's filling habits and detailed address details, thereby improving recognition efficiency.

[0067] Specifically, when the sender fills in the shipping address, part of the address is based on the specific information selected in the electronic waybill, such as the province, city, county, and street information. This part of the information is generally selected by the sender from multiple different information at the same level, so there will be no unclear issues. This part of information is called AOI (area of ​​interest) regional information. The further refined POI information is filled in by the sender independently. Due to the difference in the sender's filling habits and the refined address, it may cause recognition difficulties. In this application, the address information is marked, and then based on the server's built-in preset key address set, the key address corresponding to the room no address information filled in by the user is determined by comparison, thereby converting the original address information in the express waybill to be identified into standard address information, which facilitates the waybill recognition efficiency of the relevant delivery personnel during the delivery process.

[0068] The above-mentioned address information extraction method obtains the express waybill to be identified; extracts the original address information of the express waybill to be identified; annotates the hierarchical address information in the original address information to obtain the annotated address information; and obtains the standard address information corresponding to the original address information based on the annotated address information and a preset key address set. This application extracts the original address in the express waybill and then determines the standard address information corresponding to the original address information based on the hierarchical address information annotation and standard address identification. This can effectively improve the address information recognition efficiency during the express delivery process, avoid the situation where the address information is unclear due to typos by the customer, which makes the delivery process unable to proceed normally and affects the express delivery efficiency.

[0069] In one embodiment, Figure 3 As shown, step 207 includes:

[0070] Step 302: Obtain the similarity between the key address in the marked address information and the standard key address in the preset key address set.

[0071] Step 304: Identify the standard key address corresponding to the marked address information based on the similarity.

[0072] Step 306: Obtain standard address information corresponding to the original address information according to the marked address information and the standard key address.

[0073] The key address is the aforementioned POI address. A preset key address set corresponds one-to-one with the aforementioned AOI area information, and a preset key address set includes multiple different key addresses. A standard key address refers to address information pre-stored in server 104 that corresponds to the POI address input by the user. A preset key address set includes multiple different key addresses.

[0074] The process of annotating the hierarchical address information in the original address information includes extracting the POI address and adding the annotation. Therefore, when identifying the annotated address information, the key address in the annotated address information can be directly obtained. Simultaneously, based on the AOI information in the annotated address information, a preset key address set corresponding to the annotated address information can be searched. The similarity between the key address in the annotated address information and each standard key address in the preset key address set is then determined. The server 104 can then convert the POI address entered by the user into the corresponding standard key address by comparing the similarities, thereby converting the original address information into standard address information. In this embodiment, the similarity can effectively identify the standard address information corresponding to the original address information, thereby improving the recognition accuracy during the recognition process.

[0075] like Figure 4 As shown, in one embodiment, step 302 includes:

[0076] Step 401: Identify the interest surface address and the marked key address in the marked address information according to the mark in the marked address information.

[0077] Step 403: Search for a preset key address set corresponding to the interest surface address.

[0078] Step 405 : Obtain the similarity between the marked key address and the standard key address in the preset key address set by using the fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

[0079] The AOI address, referred to above, corresponds to the prov (province) > city (city) > district (district) > town (town) > comm (community) > road (street) > roadno (street number) hierarchy in the address hierarchy. Using the AOI address, the standard address corresponding to the original address information can be located within a specific area. Each AOI address corresponds to a preset key address set. The labeled key address corresponds to the POI address in the address hierarchy. Each AOI address corresponds to a preset key address set.

[0080] Specifically, the interest surface address and the annotated key address in the annotated address information can be identified based on the annotations in the annotated address information. Then, based on the interest surface address, the preset key address set corresponding to the current interest surface address is determined from the full amount of key addresses, and then the similarity between the annotated key address in the current original address information and each key address in the preset key address set is calculated by the fasttext text similarity algorithm based on calculating the Word Embedding cosine distance, so that the standard key address corresponding to the annotated address information can be identified based on the similarity. In this embodiment, by identifying the interest surface address and the annotated key address in the annotated address information, and then searching for the preset key address set corresponding to the interest surface address, only comparing the preset key address set with the annotated key address, the amount of calculation in the comparison process can be effectively reduced, thereby improving the extraction efficiency in the address information extraction process.

[0081] like Figure 5 As shown, in one embodiment, before step 302, the method further includes:

[0082] Step 502: Obtain historical express waybill data with hierarchical address information.

[0083] Step 504 : Identify the address of interest in the historical express waybill data and the address combination of the key address corresponding to the address of interest based on the hierarchical address information annotation.

[0084] Step 506: Cluster the address combinations with the same interest surface address to construct a preset key address set corresponding to each interest surface address.

[0085] Here, historical express waybill data specifically refers to waybill data extracted from a historical database. An address combination refers to a combination of an interest face address and a key address. In one embodiment, the historical express waybill data can be a portion of historical express waybill data randomly extracted from the FVP (Full-Scale Visualization Platform) Kafka and OMS (Order Management System) Kafka corresponding to the express delivery system. The delivery addresses in this portion of historical express waybill data have been standardized. Server 104 can then construct a set of standardized key addresses corresponding to each interest face address based on this portion of historical express waybill data. First, server 104 obtains historical express waybill data annotated with hierarchical address information and then identifies interest face addresses in the historical express waybill data based on the hierarchical address information annotations. The interest face address represents a region, and this information is the same for all express waybill data, with no differences. Therefore, based on the hierarchical address information annotations in the historical express waybill data, this historical express waybill data can be efficiently clustered. The historical express waybill data with the same interest face address are classified into one category, and an address combination of the interest face address and the key address is constructed and clustered, wherein clustering specifically refers to treating similar key addresses under the same interest face address as the same address. The similarity between each key address in the multiple address sets corresponding to an interest face address is calculated, and then clustering is performed based on the similarity to determine the standard key addresses in the multiple address sets, and all the standard key addresses corresponding to an interest face address are the preset key address set corresponding to the interest face address. In this embodiment, the preset key address set corresponding to each interest face address is constructed using historical express waybill data, which can effectively ensure the validity of the standard key addresses in the constructed preset key address set.

[0086] In one embodiment, prior to step 506, the following further steps are included: Step 601: determining the frequency of occurrence of each address combination in historical express delivery waybill data and the similarity between each address combination; and cleaning each address combination based on the frequency of occurrence and similarity. Step 506 includes clustering the cleaned address combinations with the same interest face address to construct a preset key address set corresponding to each interest face address.

[0087] The frequency of occurrence of each address combination is calculated primarily to eliminate combinations with low frequency of occurrence and prevent data from being disrupted by incorrectly expressed key address information. The similarity between address combinations specifically refers to the similarity of key addresses within the same interest face address. This process can be calculated using the FastText text similarity algorithm based on the cosine distance of Word Embedding. After calculating the similarity of key addresses between each address combination, key addresses above the similarity threshold are considered identical, the key address with the highest frequency of occurrence is used as the standard key address, and all other key addresses above the similarity threshold are converted to standard key addresses. In this way, all key addresses in the historical express waybill data are converted to a preset key address set. Address combinations with the same cleaned interest face address can then be clustered to construct a preset key address set corresponding to each interest face address. In this embodiment, cleaning each address combination based on frequency of occurrence and similarity effectively ensures the validity of the clustering results obtained during cluster analysis and eliminates interference from incorrect information.

[0088] In one embodiment, after step 207, the method further includes: marking the standard key address in the standard address information.

[0089] Specifically, after determining the standard key address, the standard key address can be output to the relevant delivery operator. Before outputting the standard key address, server 104 can also mark the standard key address in the standard address information to optimize the presentation of the key address information. Specifically, the marking can be performed by bolding and enlarging the font of the standard key address, or by marking it in a prominent color, such as red. In this embodiment, by marking the standard key address in the standard address information, the recognition speed of the standard key address can be effectively guaranteed during operations such as express delivery, thereby improving operational efficiency.

[0090] In one embodiment, step 205 includes: labeling the hierarchical address information in the original address information through a preset LSTM information labeling model to obtain the labeled address, and the preset LSTM information labeling model is obtained by training historical express waybill data with hierarchical address information labeling.

[0091] The LSTM model is a long short-term memory artificial neural network model, which is a time recurrent neural network specially designed to solve the long-term dependency problem of general RNN (recurrent neural network). All RNNs have a chain form of repeated neural network modules. In this application, the address information in the express waybill is mainly labeled by a pre-built LSTM information labeling model, and the input address is divided into levels. Then, the information of different levels can be combined according to the needs to obtain the interest face address and the key address in the labeled address information. The LSTM model is based on a deep learning algorithm, and its training set is historical express waybill data with hierarchical address information labeling. The initial LSTM model can be trained based on the historical express waybill data, and the training is completed by training methods such as gradient descent method to obtain a preset LSTM information labeling model that can be used for address information labeling. In this embodiment, the labeling of the hierarchical address information in the original address information is completed by the trained preset LSTM information labeling model, which can effectively improve the accuracy of information labeling.

[0092] It should be understood that although Figure 2-5 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 2-5 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The order of execution of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0093] In one embodiment, Figure 6 As shown, an address information extraction device is provided, comprising:

[0094] The waybill obtaining module 601 is used to obtain the express waybill to be identified.

[0095] The address extraction module 603 is used to extract the original address information of the express waybill to be identified.

[0096] The address information marking module 605 is used to mark the hierarchical address information in the original address information to obtain marked address information.

[0097] The standard address identification module 607 is used to obtain the standard address information corresponding to the original address information according to the marked address information and the preset key address set.

[0098] In one embodiment, the standard address identification module 607 is specifically used to: obtain the key address in the marked address information and the similarity with the standard key address in the preset key address set; identify the standard key address corresponding to the marked address information based on the similarity; and obtain the standard address information corresponding to the original address information based on the marked address information and the standard key address.

[0099] In one embodiment, the standard address identification module 607 is also used to: identify the interest surface address and the annotated key address in the annotated address information based on the annotations in the annotated address information; search for a preset key address set corresponding to the interest surface address; and obtain the similarity between the annotated key address and the standard key address in the preset key address set through a fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

[0100] In one of the embodiments, an address set construction module is further included, which is used to: obtain historical express waybill data with hierarchical address information annotations; identify the interest surface addresses in the historical express waybill data and the address combinations of key addresses corresponding to the interest surface addresses based on the hierarchical address information annotations; cluster the address combinations with the same interest surface addresses to construct a preset key address set corresponding to each interest surface address.

[0101] In one embodiment, it also includes a data cleaning module for: determining the frequency of occurrence of each address combination in historical express waybill data and the similarity between each address combination; cleaning each address combination according to the frequency of occurrence and similarity; the address set construction module is specifically used for: clustering the address combinations with the same interest surface addresses after cleaning, and constructing a preset key address set corresponding to each interest surface address.

[0102] In one embodiment, a key information marking module is further included, which is used to mark the standard key addresses in the standard address information.

[0103] In one embodiment, the address information labeling module 605 is specifically used to: label the hierarchical address information in the original address information through a preset LSTM information labeling model to obtain the labeled address, and the preset LSTM information labeling model is obtained by training historical express waybill data with hierarchical address information labeling.

[0104] The specific definition of the address information extraction device can be found in the definition of the address information extraction method above and will not be repeated here. The various modules in the above-mentioned address information extraction device can be implemented in whole or in part through software, hardware, or a combination thereof. The above-mentioned modules can be embedded in or independent of the processor of the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0105] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store address information extraction data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it implements an address information extraction method.

[0106] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0107] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0108] Get the express waybill to be identified;

[0109] Extract the original address information of the express waybill to be identified;

[0110] Marking the hierarchical address information in the original address information to obtain the marked address information;

[0111] According to the marked address information and the preset key address set, the standard address information corresponding to the original address information is obtained.

[0112] In one embodiment, when the processor executes the computer program, it also implements the following steps: obtaining the key address in the marked address information and the similarity with the standard key address in the preset key address set; identifying the standard key address corresponding to the marked address information based on the similarity; and obtaining the standard address information corresponding to the original address information based on the marked address information and the standard key address.

[0113] In one embodiment, when the processor executes the computer program, the following steps are also implemented: based on the annotations in the annotated address information, the interest surface address and the annotated key address in the annotated address information are identified; a preset key address set corresponding to the interest surface address is searched; and the similarity between the annotated key address and the standard key address in the preset key address set is obtained through a fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

[0114] In one embodiment, when the processor executes the computer program, it also implements the following steps: obtaining historical express waybill data with hierarchical address information annotations; identifying the interest surface addresses in the historical express waybill data and the address combinations of key addresses corresponding to the interest surface addresses based on the hierarchical address information annotations; clustering the address combinations with the same interest surface addresses, and constructing a preset key address set corresponding to each interest surface address.

[0115] In one embodiment, when the processor executes the computer program, the following steps are further implemented: determining the frequency of occurrence of each address combination in the historical express waybill data and the similarity between each address combination; and cleaning each address combination according to the frequency of occurrence and the similarity.

[0116] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: marking the standard key address in the standard address information.

[0117] In one embodiment, when the processor executes the computer program, the following steps are also implemented: the hierarchical address information in the original address information is labeled through a preset LSTM information labeling model to obtain the labeled address, and the preset LSTM information labeling model is obtained by training historical express waybill data with hierarchical address information labeled.

[0118] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0119] Get the express waybill to be identified;

[0120] Extract the original address information of the express waybill to be identified;

[0121] Marking the hierarchical address information in the original address information to obtain the marked address information;

[0122] According to the marked address information and the preset key address set, the standard address information corresponding to the original address information is obtained.

[0123] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining the key address in the marked address information and the similarity with the standard key address in the preset key address set; identifying the standard key address corresponding to the marked address information based on the similarity; and obtaining the standard address information corresponding to the original address information based on the marked address information and the standard key address.

[0124] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: based on the annotations in the annotated address information, the interest surface address and the annotated key address in the annotated address information are identified; a preset key address set corresponding to the interest surface address is searched; and the similarity between the annotated key address and the standard key address in the preset key address set is obtained through a fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

[0125] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining historical express waybill data with hierarchical address information annotations; identifying the interest surface addresses in the historical express waybill data and the address combinations of key addresses corresponding to the interest surface addresses based on the hierarchical address information annotations; clustering the address combinations with the same interest surface addresses, and constructing a preset key address set corresponding to each interest surface address.

[0126] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: determining the frequency of occurrence of each address combination in the historical express waybill data and the similarity between each address combination; and cleaning each address combination according to the frequency of occurrence and the similarity.

[0127] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: marking the standard key address in the standard address information.

[0128] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: the hierarchical address information in the original address information is labeled through a preset LSTM information labeling model to obtain the labeled address, and the preset LSTM information labeling model is obtained by training historical express waybill data with hierarchical address information labeled.

[0129] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0130] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0131] The above embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for extracting address information, the method comprising: Get the express waybill to be identified; Extracting the original address information of the express waybill to be identified; Marking the hierarchical address information in the original address information to obtain marked address information; Obtain historical express waybill data with hierarchical address information; Identifying, based on the hierarchical address information annotation, an address of interest in the historical express waybill data, and an address combination of a key address corresponding to the address of interest, wherein the address combination consists of an address of interest and a key address combination; Clustering the address combinations with the same interest surface address to construct a preset key address set corresponding to each interest surface address; Obtaining a similarity between a key address in the annotated address information and a standard key address in a preset key address set, wherein the annotated address information includes an interest surface area and a key address, wherein the interest surface area includes a province, city, county, and street, and is selected by the sender of the express waybill to be identified from multiple pieces of information at the same level. The key address is filled in by the sender based on their filling habits and is a point of interest address that is further refined from the interest surface address; identifying, based on the similarity, a standard key address corresponding to the annotated address information; According to the marked address information and the standard key address, standard address information corresponding to the original address information is obtained.

2. The method according to claim 1, characterized in that The obtaining of the similarity between the key address in the marked address information and the standard key address in the preset key address set includes: identifying, according to the annotations in the annotated address information, an interest surface address and an annotated key address in the annotated address information; Searching for a preset key address set corresponding to the interest surface address; The similarity between the marked key address and the standard key address in the preset key address set is obtained by using the fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

3. The method according to claim 1, characterized in that Before clustering the address combinations with the same interest surface addresses and constructing the preset key address sets corresponding to the interest surface addresses, the method further includes: Determine the frequency of occurrence of each address combination in the historical express waybill data and the similarity between each address combination; Cleaning the address combinations according to the occurrence frequency and the similarity; Clustering the address combinations with the same interest surface addresses to construct a preset key address set corresponding to each interest surface address includes: The address combinations with the same cleaned interest surface addresses are clustered to construct a preset key address set corresponding to each interest surface address.

4. The method according to claim 1, wherein After acquiring the standard address information corresponding to the original address information according to the marked address information and the standard key address, the method further includes: The standard key address in the standard address information is marked.

5. The method according to claim 1, wherein The marking of the hierarchical address information in the original address information to obtain the marked address information includes: The hierarchical address information in the original address information is annotated by a preset LSTM information annotation model to obtain the annotated address. The preset LSTM information annotation model is obtained by training historical express waybill data annotated with hierarchical address information.

6. An address information extraction device, characterized in that: The device comprises: Waybill acquisition module, used to obtain the express waybill to be identified; An address extraction module, used to extract the original address information of the express waybill to be identified; An address information marking module, configured to mark the hierarchical address information in the original address information to obtain marked address information; An address set construction module is configured to obtain historical express waybill data annotated with hierarchical address information; identify, based on the hierarchical address information annotated, addresses of interest in the historical express waybill data, and address combinations of key addresses corresponding to the addresses of interest, wherein the address combinations consist of an address of interest and a key address combination; cluster the address combinations with the same address of interest to construct a preset key address set corresponding to each address of interest; A standard address identification module is used to obtain the key address in the marked address information and its similarity with the standard key address in the preset key address set. The marked address information includes an interest area and a key address. The interest area includes provinces, cities, counties and streets. The sender of the express waybill to be identified selects one from multiple information of the same level. The key address is filled in by the sender based on the filling habit. The key address is the point of interest address that is further refined from the interest area address; according to the similarity, the standard key address corresponding to the marked address information is identified; according to the marked address information and the standard key address, the standard address information corresponding to the original address information is obtained.

7. The device according to claim 6, characterized in that The standard address identification module is specifically used to: identify the interest surface address and the annotated key address in the annotated address information based on the annotations in the annotated address information; search for the preset key address set corresponding to the interest surface address; and obtain the similarity between the annotated key address and the standard key address in the preset key address set through the fasttext text similarity algorithm based on calculating the Word Embedding cosine distance.

8. The device according to claim 6, characterized in that It also includes a data cleaning module, which is used to: determine the frequency of occurrence of each address combination in the historical express waybill data and the similarity between each address combination; clean the each address combination according to the frequency of occurrence and the similarity; the address set construction module is specifically used to: cluster the address combinations with the same interest surface address after cleaning, and construct a preset key address set corresponding to each interest surface address.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images