Order information processing method and device, electronic equipment and computer storage medium

Abstract

The invention discloses an order information processing method and device, electronic equipment and a storage medium, in the order information processing method, multiple pieces of candidate address information are recalled in a high-confidence positioning address information library, and target candidate address information is screened from the multiple pieces of candidate address information based on candidate text semantic similarity, so that the target candidate address information is obtained; the target candidate address information is in the high-confidence positioning address information base, and the high-confidence positioning address information and the verified high-confidence positioning point information corresponding to the high-confidence positioning address information are stored in the high-confidence positioning address information base, so that the target candidate address information is more matched with the target delivery address information. According to the method and the device, the target candidate address information is acquired, the high-confidence positioning point information corresponding to the target candidate address information, namely the target positioning point information, is highly matched with the target delivery address information, so that a delivery path planned based on the target positioning point information can be conveniently and quickly delivered by a delivery rider, and the delivery accuracy and the delivery efficiency of an order are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet, in particular to an order information processing method, an order information processing device, an electronic device and a computer storage medium. BACKGROUND

[0002] With more and more online orders, the delivery of online orders is also attracting more and more attention. When a user places an order, a delivery address (i.e. a delivery address) will be filled in to facilitate path planning for the order based on the delivery address to deliver the order.

[0003] Specifically, when planning the delivery path of the order, the delivery positioning point (positioning information) corresponding to the delivery address is used for planning. However, the current way of confirming the delivery positioning point corresponding to the delivery address has some problems, so that the confirmed delivery positioning point is not accurate, that is, the delivery positioning point does not match the delivery address: for example, when the user places an order, a positioning point is selected as the delivery positioning point while filling in the delivery address, for example, after the user places an order, a positioning point is generated as the delivery positioning point based on the filled delivery address, for example, the location point where the user places an order is taken as the delivery positioning point; no matter which way the delivery positioning point is determined, there may be a situation that the delivery positioning point does not match the delivery address filled by the user, which further leads to the fact that the delivery rider cannot deliver the order; therefore, in the field of order delivery, how to provide a delivery positioning point that matches the delivery address filled by the user becomes a technical problem to be solved. SUMMARY

[0004] The present application provides an order information processing method to provide a delivery positioning point that matches the delivery address filled by the user, improve the delivery accuracy of the order, and improve the delivery efficiency of the order. The present application also provides another order information processing method, an order information processing device, an electronic device and a computer storage medium.

[0005] In a first aspect, the application provides an order information processing method applied to a server, which comprises: in response to receiving target delivery address information of a to-be-processed order and initial arrival positioning point information preset for the target delivery address information, recalling a plurality of candidate address information related to the text of the target delivery address information from a high-confidence positioning address information library based on the target delivery address information and the initial arrival positioning point information; the high-confidence positioning address information library stores high-confidence positioning address information and high-confidence positioning point information corresponding to the high-confidence positioning address information, and the high-confidence positioning point information is verified positioning point information; determining candidate text semantic similarity between the target delivery address information and each candidate address information according to the target delivery address information and the plurality of candidate address information; screening target candidate address information matching the target delivery address information from the plurality of candidate address information according to the candidate text semantic similarity, and taking high-confidence positioning point information corresponding to the target candidate address information as target positioning point information of the target delivery address information; and providing the target positioning point information to a delivery terminal to enable the delivery terminal to determine a delivery path for delivering the to-be-processed order by using the target positioning point.

[0006] In a second aspect, the application provides an order information processing method applied to a delivery terminal, which comprises: obtaining target positioning point information corresponding to target delivery address information of a to-be-processed order provided by a server; the target positioning point information is high-confidence positioning point information corresponding to target candidate address information, the target candidate address information is address information matching the target delivery address information screened from a plurality of candidate address information based on candidate text semantic similarity between the target delivery address information and each candidate address information; the plurality of candidate address information is candidate address information related to the text of the target delivery address information recalled from a high-confidence positioning address information library based on the target delivery address information and initial arrival positioning point information; the high-confidence positioning address information library stores high-confidence positioning address information and high-confidence positioning point information corresponding to the high-confidence positioning address information, and the high-confidence positioning point information is verified positioning point information; and the initial arrival positioning point information is arrival positioning point information preset for the target delivery address information; and determining a delivery path for delivering the to-be-processed order by using the target positioning point.

[0007] In a third aspect, the application provides an order information processing device applied to a server, which comprises: a recall unit configured to, in response to receiving target delivery address information of an order to be processed and initial arrival positioning point information preset for the target delivery address information, recall, based on the target delivery address information and the initial arrival positioning point information, a plurality of candidate address information related to the text of the target delivery address information from a high-confidence positioning address information library; the high-confidence positioning address information library stores high-confidence positioning address information and high-confidence positioning point information corresponding to the high-confidence positioning address information, and the high-confidence positioning point information is verified positioning point information; a semantic similarity determination unit configured to determine candidate text semantic similarity between the target delivery address information and each candidate address information according to the target delivery address information and the plurality of candidate address information; a screening unit configured to screen, according to the candidate text semantic similarity, target candidate address information matching the target delivery address information from the plurality of candidate address information, and take the high-confidence positioning point information corresponding to the target candidate address information as target positioning point information of the target delivery address information; and a target positioning point information providing unit configured to provide the target positioning point information to a delivery terminal, so that the delivery terminal determines a delivery path of the order to be processed by using the target positioning point.

[0008] In a fourth aspect, the application provides an order information processing device applied to a delivery terminal, which comprises: a target positioning point information obtaining unit configured to obtain target positioning point information corresponding to target delivery address information of an order to be processed provided by a server; the target positioning point information is high-confidence positioning point information corresponding to target candidate address information, the target candidate address information is address information matching the target delivery address information screened from a plurality of candidate address information based on candidate text semantic similarity between the target delivery address information and each candidate address information; the plurality of candidate address information is candidate address information related to the text of the target delivery address information recalled from a high-confidence positioning address information library based on the target delivery address information and initial arrival positioning point information; the high-confidence positioning address information library stores high-confidence positioning address information and high-confidence positioning point information corresponding to the high-confidence positioning address information, and the high-confidence positioning point information is verified positioning point information; and the initial arrival positioning point information is arrival positioning point information preset for the target delivery address information; and a delivery path determination unit configured to determine a delivery path of the order to be processed by using the target positioning point.

[0009] In a fifth aspect, the application provides an electronic device comprising a processor and a memory for storing a computer program, wherein the electronic device is powered on and runs the computer program through the processor to execute the order information processing method.

[0010] In a sixth aspect, the present application provides a computer storage medium, which stores computer execution instructions, and the computer execution instructions are run by a processor to execute the order information processing method.

[0011] Compared with the prior art, the present application has the following advantages: The order information processing method provided by the present application, after receiving the target delivery address information of the to-be-processed order and the initial arrival positioning point information preset for the target delivery address information, recalls a plurality of candidate address information related to the text of the target delivery address information in the high-confidence positioning address information library, determines the candidate text semantic similarity between the target delivery address information and each candidate address information according to the target delivery address information and the plurality of candidate address information, and screens the target candidate address information matched with the target delivery address information from the plurality of candidate address information according to the candidate text semantic similarity, and takes the high-confidence positioning point information corresponding to the target candidate address information as the target positioning point information of the target delivery address information. By recalling the plurality of candidate address information in the high-confidence positioning address information library and screening the target candidate address information from the plurality of candidate address information based on the determined candidate text semantic similarity between the target delivery address information and each candidate address information, the target candidate address information is more matched with the target delivery address information in the semantic dimension. Since the target candidate address information is in the high-confidence positioning address information library, the high-confidence positioning address information library stores high-confidence positioning address information and high-confidence positioning point information corresponding to the high-confidence positioning address information. The high-confidence positioning point information is verified positioning point information, and the high-confidence positioning point information corresponding to the target candidate address information, i.e., the target positioning point information, is highly adaptive positioning point information for the target delivery address information. That is, the target positioning point information obtained in this way is more accurate, and the delivery path of the to-be-processed order planned based on the target positioning point information can facilitate the delivery rider to quickly deliver the to-be-processed order, thereby improving the delivery accuracy and efficiency of the order. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0013] Figure 1 Flowchart of the order information processing method provided by the first embodiment of the present application.

[0014] Figure 2A structural schematic diagram of a target semantic analysis model provided for a first embodiment of the present application.

[0015] Figure 3 A flowchart of an order information processing method provided for a second embodiment of the present application.

[0016] Figure 4 A schematic diagram of an order information processing apparatus provided for a third embodiment of the present application.

[0017] Figure 5 A schematic diagram of an order information processing apparatus provided for a fourth embodiment of the present application.

[0018] Figure 6 A schematic diagram of an electronic device provided for a fifth embodiment of the present application. DETAILED DESCRIPTION

[0019] In order for those skilled in the art to better understand the technical solutions of the present application, the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. However, the present application can be implemented in many different ways from the description below, and therefore, all other embodiments obtained by those skilled in the art based on the embodiments provided by the present application without creative labor should fall within the scope of protection of the present application.

[0020] It should be noted that the terms "first", "second", "third", etc. in the claims, the specification and the drawings of the present application are used to distinguish similar objects and are not intended to describe a specific order or sequence. The data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described in the present application. In addition, the terms "include", "have" and their variants are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0021] It should be understood that in the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. "And / or" is only a description of the association relationship between the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects. "Including A, B and / or C" means including any 1 or any 2 or 3 of A, B and C.

[0022] It should be understood that in the embodiments of this application, "B corresponding to A", "B corresponding to A", "A corresponds to B" or "B corresponds to A" means that B is associated with A, and B can be determined based on A. Determining B based on A does not mean that B is determined solely based on A; B can also be determined based on A and / or other information.

[0023] This application provides an order information processing method, an order information processing apparatus, an electronic device, and a computer storage medium. The following specific embodiments describe the order information processing method, the order information processing apparatus, the electronic device, and the computer storage medium. To more clearly illustrate the order information processing method provided by the embodiments of this application, the application scenarios of the order information processing method provided by the embodiments of this application are first introduced.

[0024] The order information processing method of this application can be applied to scenarios where a preset delivery location point is corrected for the delivery address of an order to be processed. When a user places an order, the system obtains the delivery address information filled in by the user for the order to be processed, and simultaneously obtains the preset initial delivery location point for the delivery address information. The source of the preset initial delivery location point for the delivery address information can be in various ways: for example, the user can select a location as the delivery location point after or during the process of filling in the delivery address information; another example is that a location point is generated for the delivery address information after obtaining it; yet another example is that the location where the user was when placing the order can be used as the delivery location point, i.e., the location obtained by the user when placing the order using GPS (Global Positioning System). Regardless of the method used to preset the initial delivery location point for the target delivery address information, the initial delivery location point may not match the address of the target delivery address information, which is the delivery address information of the order to be processed. In fact, when the initial delivery location point preset for the target delivery address information does not match the target delivery address information, the initial delivery location point is an abnormal location point.

[0025] To facilitate obtaining a target location point that accurately matches the target delivery address, a high-confidence location address information database is searched for textually similar to the target delivery address. Then, the high-confidence location point corresponding to the high-confidence location address information is used as the target location point. In this application, the high-confidence location address information database stores both high-confidence location address information and corresponding high-confidence location point information. The high-confidence location point information corresponding to the high-confidence location address information is verified location point information. The high-confidence location address information found in the high-confidence location address information database that is textually similar to the target delivery address is used as the target candidate address information. Since the target candidate address information is the address information within the high-confidence location address information, the high-confidence location point information corresponding to the target candidate address information is used as the target location point information for the target delivery address. The target location point is highly compatible with the target delivery address, meaning it accurately reflects the location of the target delivery address, thus enabling delivery riders to quickly deliver pending orders.

[0026] Specifically, after obtaining the target delivery address information and the preset initial delivery location information for the target delivery address information, multiple candidate address information related to the text of the target delivery address information are recalled from a high-confidence location address information database based on the target delivery address information and the initial delivery location information. For example, when the target delivery address is XX Hotel (back door delivery rack) in C Street, B District, City A, the recalled multiple candidate addresses are XX Hotel (back door delivery cabinet) in C Street, B District, City A, unloading area of ​​XX Hotel in C Street, B District, City A, room 2302 of XX Hotel in C Street, B District, City A, and XX Hotel in C Street; the text of the recalled multiple candidate address information is related to the text of the target delivery address information.

[0027] To ensure that the distance between the recalled candidate addresses and the target delivery address is not too far, multiple candidate addresses are selected during the recall process based on the distance between the initial delivery location and the high-confidence location point of the recalled address. Specifically, during the recall of multiple candidate addresses from the high-confidence location address information database, firstly, using the target delivery address information as the search text, related address information is retrieved from the high-confidence location address information database. Then, based on the distance between the high-confidence location point corresponding to the related address information and the initial delivery location point, address information that meets the first preset distance condition is selected as multiple candidate address information. A preset text retrieval algorithm can be used to retrieve related address information from the high-confidence location address information database using the target delivery address information as the search text.

[0028] The pre-set text retrieval algorithm is actually used to facilitate the retrieval of associated address information that is text-related to the target delivery address in the high-confidence location address information database. For example, the associated addresses are XX Hotel (back door delivery cabinet) in C Street, B District, A City, unloading area of ​​XX Hotel in C Street, B District, A City, room 2302 of XX Hotel in C Street, B District, A City, unloading area of ​​XXX Restaurant in A Street, B District, A City, and XXX Restaurant in A Street. Generally, for a certain target delivery address, the number of candidate addresses is less than or equal to the number of associated addresses.

[0029] The preset text retrieval algorithm is such as the BM25 (Best Matching 25) algorithm. BM25 is a text relevance scoring algorithm for information retrieval. It calculates the relevance score between a document and a query by evaluating the frequency and rarity of query terms in a document and the document length. In this scenario, the target delivery address information is used as the query, and all high-confidence location address information stored in the high-confidence location address information database is used as the document. Then, the BM25 algorithm is used to calculate the relevance score between the target delivery address information and all high-confidence location address information stored in the high-confidence location address information database. High-confidence location address information with a score not less than a preset threshold is used as associated address information.

[0030] Since the associated address information is all from the high-confidence location address information database, a high-confidence location point corresponding to each associated address can be obtained. Subsequently, based on the distance between the high-confidence location point corresponding to each associated address and the initial delivery location point, address information that meets the first preset distance condition is selected from the associated address information as multiple candidate address information.

[0031] Specifically, since the high-confidence location point and initial delivery location point corresponding to each associated address can obtain their geographical location information (such as latitude and longitude information), the distance (such as spherical distance) between the high-confidence location point and initial delivery location point corresponding to each associated address can be obtained based on the geographical location information of the location points; satisfying the first preset distance condition, such as less than or equal to 300 meters; adopting this method ensures that the multiple candidate addresses recalled are associated with the target delivery address in the text dimension, thereby making it more likely that the candidate address and the target delivery address belong to the same address; on the other hand, by using the distance between the location points, some addresses that are far from the target delivery address can be filtered out from the associated addresses (such as associated addresses with a distance of more than 300 meters are filtered out).

[0032] In fact, considering that some of the obtained associated addresses may be related to the target delivery address in text, such as some addresses with the same name, but in reality, these addresses are far away from the target delivery address. Since the recalled address information and the target delivery address information need to be input into the target semantic analysis model later, in order to reduce the computational load of the model and thus improve the computational efficiency of the model, these recalled addresses that are far away from the target delivery address are filtered out. For example, if 10 addresses are recalled, the model needs to predict 10 results. Filtering out the 3 distant and irrelevant addresses can improve the model's computational efficiency by 30%.

[0033] After recalling multiple candidate address information textually related to the target delivery address information, the semantic similarity between the target delivery address information and each candidate address information is calculated as the candidate text semantic similarity. Then, based on the candidate text semantic similarity, target candidate address information matching the target delivery address information is selected from the multiple candidate address information. The high-confidence location point information corresponding to the target candidate address information is used as the target location point information of the target delivery address information. For example, when the calculated semantic similarities between the candidate texts of XX Hotel (back door takeout cabinet) in C Street, B District, A City, XX Hotel (back door takeout cabinet) in C Street, B District, A City, XX Hotel (back door takeout cabinet) in C Street, B District, A City, are 0.95, 0.92, 0.90, and 0.80 respectively, the hotel with the highest semantic similarity ranking, XX Hotel (back door takeout cabinet) in C Street, B District, A City, can be used as the target candidate address. Subsequently, the high-confidence location point information corresponding to this target candidate address will be used as the target location point information.

[0034] To facilitate the calculation of the semantic similarity between the target delivery address information and each candidate address information, the target delivery address information and multiple candidate address information are input into the target semantic analysis model to determine the semantic similarity between the target delivery address information and each candidate address information. The target semantic analysis model is a trained neural network model used to determine the semantic similarity between texts to be processed.

[0035] Specifically, target semantic analysis models such as the BERT pre-trained model (BERT, Bidirectional Encoder Representations from Transformers, a bidirectional encoding representation model with Transformers as the main framework; Transformer is a deep neural network model based on a multi-layer self-attention mechanism) perform unsupervised learning by masking the words of the target delivery address information and each candidate address information, and predicting sentence relationships. Through this target semantic analysis model, the semantic similarity of candidate texts between the target delivery address information and each candidate address information can be determined.

[0036] The order information processing method of this application can correct the initial delivery location point, thereby making the target location point information more compatible with the target delivery address information. This allows the delivery route of the pending order planned based on the target location point information to facilitate the delivery rider to deliver the pending order quickly, improving the delivery accuracy and efficiency. It should be noted that the technical solution of this invention can be applied to the transaction and delivery services of instant e-commerce platforms, such as Taobao Flash Sale, Taoxianda, Ele.me takeaway, and retail.

[0037] The above description illustrates one application scenario of the order information processing method of this application. The embodiments of this application do not specifically limit the application scenario of the order information processing method. The above-described application scenario is merely one embodiment of the order information processing method provided in this application. The purpose of providing this application scenario embodiment is to facilitate understanding of the order information processing method provided in this application, and not to limit the order information processing method provided in this application. Other application scenarios of the order information processing method in the embodiments of this application will not be elaborated upon.

[0038] First Embodiment This embodiment provides an order information processing method, in which the server is the executing entity. Please refer to [link / reference needed] for details. Figure 1 This is a flowchart of the order information processing method provided in the first embodiment of this application. For some examples and detailed descriptions of this embodiment, please refer to the above scenario embodiment.

[0039] The order information processing method of this application includes the following steps.

[0040] Step S101: In response to receiving the target delivery address information of the order to be processed and the initial delivery location information preset for the target delivery address information, based on the target delivery address information and the initial delivery location information, recall multiple candidate address information that are text-related to the target delivery address information in the high-confidence location address information database.

[0041] In this embodiment, the execution entity is the server. After receiving the order to be processed, it will simultaneously obtain the target delivery address information of the order to be processed and the initial delivery location information preset for the target delivery address information. For the specific source of the initial delivery location information, please refer to the relevant description in the scenario embodiment. Since the initial location may exist in the case of mismatch between the target delivery address and the order to be processed, it may result in the inability to deliver the order to be processed. Alternatively, during the delivery of the order to be processed, the delivery rider may need to contact the user who placed the order to obtain the delivery location of the order to be processed again, resulting in low delivery efficiency and a poor experience for the delivery rider.

[0042] Therefore, after obtaining the target delivery address information of the order to be processed and the initial delivery location information preset for the target delivery address information, multiple candidate address information related to the text of the target delivery address information are recalled from the high-confidence location address information database based on the target delivery address information and the initial delivery location information.

[0043] The high-confidence location address information database stores high-confidence location address information and corresponding high-confidence location point information. The high-confidence location point information is verified location point information. In this embodiment, "high confidence" refers to a high degree of confidence, meaning that the location address information and location point information stored in the high-confidence location address information database have a high degree of compatibility, and the high-confidence location point information can accurately reflect the location of the high-confidence location address information.

[0044] To better understand the construction process of the high-confidence location address information database and how the high-confidence location point information is verified, please refer to the following description.

[0045] Specifically, a high-confidence location address information database is constructed as follows: First, the historical delivery address information and historical delivery location point information corresponding to historical orders are determined; then, based on the historical delivery address information, the historical points of interest to which the historical delivery address information belongs are determined; after determining the historical points of interest to which the historical delivery address information belongs, it is determined whether the distance between the historical delivery location point and the historical point of interest meets the second preset distance condition; if the distance between the historical delivery location point and the historical point of interest meets the second preset distance condition, the historical delivery address information is stored as high-confidence location address information and the historical delivery location point information is stored as high-confidence location point information, thus constructing a high-confidence location address information database.

[0046] In determining historical delivery location points, the process begins by: obtaining the actual delivery location point information for historical orders based on their historical delivery address information; filtering out actual delivery location points that meet the delivery criteria based on the actual delivery location point information; and then aggregating the actual delivery location points that meet the delivery criteria to determine the aggregated delivery location point, which is then used as the historical delivery location point.

[0047] In fact, the reason why the high-confidence location address information database can store high-confidence location address information and corresponding high-confidence location point information is that in the process of constructing the high-confidence location address information database, the historical delivery address information and historical delivery location point information of historical orders that have been delivered are used. Moreover, since the historical delivery address information and historical delivery location points are highly compatible to a large extent, that is, the confidence level of the compatibility between the two is high, the historical delivery address information and historical delivery location points are mapped to construct the high-confidence location address information database. As a result, the address information stored in the high-confidence location address information database is high-confidence location address information, and the high-confidence location point information corresponding to the high-confidence location address information has also been verified.

[0048] Specifically, verifying historical delivery location points (high-confidence location points) actually means: based on historical delivery address information, determining the historical points of interest to which the historical delivery address information belongs; after determining the historical points of interest to which the historical delivery address information belongs, judging whether the distance between the historical delivery location point and the historical points of interest meets a second preset distance condition; if the distance between the historical delivery location point and the historical points of interest meets the second preset distance condition, then the historical delivery address information is stored as high-confidence location address information, and the historical delivery location point information is stored as high-confidence location point information to construct a high-confidence location address information database; in fact, in high-confidence location... The distance between the high-confidence location point (i.e., the historical delivery location point) in the address information database and the historical point of interest to which the historical delivery address information belongs meets the second preset distance condition (meeting the second preset distance condition can specifically be less than the second preset distance), thus making the distance between the historical delivery location point and the location of the historical delivery address very close, and even the historical delivery location point may coincide with the actual location of the historical delivery address. Therefore, the historical delivery location point can be used as the location of the historical delivery address, making the high-confidence location address information in the high-confidence location address information database and the high-confidence location point information corresponding to the high-confidence location address information have a very high degree of fit.

[0049] Since historical points of interest (POIs) can generally reflect the actual location of historical delivery address information, when the distance between the historical delivery location and the historical POI is within a second preset distance (e.g., less than 20 meters), the historical delivery location and historical delivery address information are verified and have a high degree of compatibility. For example, for historical order M, assuming that the historical POI to which its historical delivery address information belongs is POI 1, and the distance between the historical delivery location of historical order M and POI 1 is 10 meters, then the historical delivery address information of historical order M is stored as a high-confidence location address, and the historical delivery location information of historical order M is stored as a high-confidence location point, thereby constructing a high-confidence location address information database.

[0050] In this embodiment, it is also necessary to pre-determine the historical delivery location points of historical orders. The method for determining historical delivery location points is as follows: First, based on the historical delivery address information of historical orders, obtain the actual delivery location point information for delivering historical orders; based on the actual delivery location point information, filter the actual delivery location points that meet the delivery conditions; then, aggregate the actual delivery location points that meet the delivery conditions to determine the aggregated delivery location point, and use the aggregated delivery location point as the historical delivery location point.

[0051] For example, for historical orders 1 to 10, these 10 historical orders have the same or nearly the same historical delivery addresses (same historical delivery addresses mean that these 10 orders were placed by the same user, nearly the same means that these 10 orders were placed by users in the same building). The actual delivery location information for historical orders 1 to 10 can be obtained (actual delivery location information can be represented by geographic location information; the actual delivery location refers to the location where the delivery rider signs in at the delivery terminal). These 10 historical orders have 10 actual delivery locations. These 10 actual delivery locations can be aggregated (clustered) to obtain aggregated delivery locations, which can then be used as the historical delivery locations corresponding to the historical delivery addresses of these 10 historical orders.

[0052] Of course, when there are multiple actual delivery locations corresponding to a certain historical delivery address, it is necessary to aggregate the multiple actual delivery locations to determine the historical delivery location; when there is only one actual delivery location corresponding to a certain historical delivery address, the actual delivery location is directly used as the historical delivery location. When multiple actual delivery locations exist corresponding to a historical delivery address, and it's necessary to aggregate these locations to determine the historical delivery location, to ensure the aggregated location more accurately reflects the historical delivery address, actual delivery locations that don't meet the delivery criteria are filtered out before aggregation. For example, if 7 out of 10 actual delivery locations are concentrated in one area, and the remaining 3 are more dispersed, these 3 locations might not meet the delivery criteria because the delivery rider forgot to confirm delivery on the delivery terminal at the time of delivery, or because the rider confirmed delivery prematurely. The remaining 7 locations meet the delivery criteria. A qualified delivery location can be one that was checked in and delivered within a preset distance from the historical delivery address. The historical delivery address's location is the actual location of the historical delivery address.

[0053] The actual delivery locations that meet the delivery criteria are aggregated to determine the aggregated delivery location. This can be done using a pre-defined aggregation algorithm, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise). This algorithm divides data points into core points, boundary points, and noise points by judging the density of sample points. Its core idea is to define a cluster as the largest set of density-connected points, which can automatically identify clusters of arbitrary shapes and process noisy data. For example, the 10 actual delivery locations mentioned above can be used as sample points, and the 3 locations as noise points. Through aggregation, the aggregated delivery location can be determined.

[0054] By following the above method, a large amount of historical delivery address information and historical delivery location information (verified location points) corresponding to historical orders can be stored one by one, thereby constructing a high-confidence location address information database.

[0055] After constructing a high-confidence location address information database, based on the target delivery address information and the initial delivery location point information, multiple candidate address information related to the text of the target delivery address information are retrieved from the high-confidence location address information database. This can refer to: using the target delivery address information as the search text, retrieving related address information from the high-confidence location address information database; and based on the distance between the high-confidence location point and the initial delivery location point corresponding to the related address information, selecting address information that meets a first preset distance condition as multiple candidate address information. For a detailed explanation of how to retrieve multiple candidate address information, please refer to the scenario example above. Meeting the first preset distance condition can specifically mean being less than a first preset distance.

[0056] Step S102: Based on the target delivery address information and multiple candidate address information, determine the semantic similarity of the candidate text between the target delivery address information and each candidate address information.

[0057] Specifically, determining the candidate text semantic similarity between the target delivery address information and each candidate address information based on the target delivery address information and multiple candidate address information can mean: inputting the target delivery address information and multiple candidate address information into a target semantic analysis model to determine the candidate text semantic similarity between the target delivery address information and each candidate address information; the target semantic analysis model is a trained neural network model used to determine the text semantic similarity between the texts to be processed.

[0058] In this embodiment, the target semantic analysis model is a trained neural network model used to determine the textual semantic similarity between the texts to be processed.

[0059] Before performing the step of inputting the target delivery address information and multiple candidate address information into the target semantic analysis model, the method further includes: for each candidate address information, concatenating the target delivery address information and candidate address information to obtain multiple concatenated text sequences; wherein, in the concatenated text, the target delivery address information and candidate address information are separated by a delimiter; inputting the target delivery address information and multiple candidate address information into the target semantic analysis model, and determining the candidate text semantic similarity between the target delivery address information and each candidate address information, including: inputting the multiple concatenated text sequences into the target semantic analysis model respectively, and determining the candidate text semantic similarity between the target delivery address information and each candidate address information respectively.

[0060] For example, if the target delivery address is XX Hotel (back door delivery rack) in C Street, B District, City A, and the multiple candidate addresses recalled are XX Hotel (back door delivery rack) in C Street, B District, City A, unloading area of ​​XX Hotel in C Street, B District, City A, room 2302 of XX Hotel in C Street, B District, City A, and XX Hotel in C Street, then the concatenated text sequences are as follows: Sequence 1: [CLS] XX Hotel (back door delivery rack) ... In [CLS], CLS (Classification Label) refers to the label used in models like BERT to capture global semantics. Its embedding vector, after passing through multiple layers of self-attention mechanisms, integrates contextual information from the input sequence. In classification tasks, the output vector of CLS is directly used as input to the classification layer, mapped to class probabilities through fully connected layers and a Softmax normalization function. SEP in [CLS] represents the separator, allowing subsequent target semantic analysis models to distinguish between the target delivery address information and candidate address information separated by the separator.

[0061] In this embodiment, multiple concatenated text sequences are input into the target semantic analysis model to determine the candidate text semantic similarity between the target delivery address information and each candidate address information. This can be achieved by: for each concatenated text sequence, obtaining a vector representing each concatenated text sequence; performing semantic representation on each concatenated text sequence vector to obtain a first semantic information vector representing the semantics of the target delivery address information and a second semantic information vector representing the semantics of the candidate address information; and determining the candidate text semantic similarity between the target delivery address information and the candidate address information based on the first and second semantic information vectors.

[0062] To better understand the structure of this target semantic analysis model, please refer to [link / reference]. Figure 2 This is a schematic diagram of the target semantic analysis model provided in the first embodiment of this application; wherein the target semantic analysis model 200 is divided into an input layer 201, an encoding layer 202 and an output layer 203.

[0063] Input layer 201 is used to vectorize each concatenated text sequence of the input. Each concatenated text sequence vector can be a 512-dimensional numerical vector.

[0064] Encoding layer 202 employs self-attention, residual connections, layer normalization (Add&Norm), and a feedforward neural network (FNN) to semantically represent the concatenated text sequence vector, obtaining a first semantic information vector representing the semantics of the target delivery address information and a second semantic information vector representing the semantics of the candidate address information. In the encoding layer, semantic representation is actually performed through a semantic representation module. The number of representation iterations for this module can be set according to requirements, such as according to... Figure 2 Set to 12 times.

[0065] Output layer 203 is used to reduce the dimensionality of the first and second semantic information vectors using fully connected layers (such as Linear layers) to obtain the third and fourth semantic information vectors. Both the third and fourth semantic information vectors can be two-dimensional vectors. Then, activation functions (such as Softmax normalization functions) are used to process the third and fourth semantic information vectors to achieve the task objective of binary classification of target delivery address information and candidate address information, obtaining the probability that the target delivery address information and candidate address information belong to the same address. Finally, the probability that the target delivery address information and each candidate address information belong to the same address is output (i.e., the output result). The probability that the target delivery address information and each candidate address information belong to the same address is an example of the candidate text semantic similarity between the target delivery address information and each candidate address information.

[0066] More specifically, for each concatenated text sequence, obtaining each concatenated text sequence vector used for vector representation of each concatenated text sequence can refer to: for each concatenated text sequence, performing character segmentation on the concatenated text sequence to obtain each segmented character corresponding to the concatenated text sequence; for each segmented character, concatenating the character vector used for vector representation of the segmented character and the position vector used for vector representation of the position of the segmented character in the concatenated text sequence to obtain a fused vector of segmented characters; and concatenating the fused vectors of segmented characters according to their positions in the concatenated text sequence to obtain a concatenated text sequence vector. For example, for sequence 1, the segmented characters are [CLS], A, city, B, district, C, street, road, X, X, wine, shop, (, back, door, outside, sell, shelf,), A, city, B, district, C, street, road, X, X, wine, shop, (, back, door, outside, sell, cabinet,). Taking [CLS] as a segmented character, the character vector of [CLS] needs to be concatenated with its position vector (first position) in sequence 1 to obtain the fused vector of [CLS]. In this way, the fused vectors of each segmented character are obtained, and then the fused vectors of each segmented character are concatenated to obtain the concatenated text sequence vector.

[0067] Specifically, determining the semantic similarity of candidate texts between target delivery address information and candidate address information based on the first semantic information vector and the second semantic information vector can mean: performing binary classification on target delivery address information and candidate address information based on the first semantic information vector and the second semantic information vector to obtain the probability that target delivery address information and candidate address information belong to the same address, and using the probability that target delivery address information and candidate address information belong to the same address as the semantic similarity of candidate texts.

[0068] More specifically, classifying the target delivery address information and candidate address information into two categories based on the first semantic information vector and the second semantic information vector to obtain the probability that the target delivery address information and candidate address information belong to the same address can refer to: reducing the dimension of the first semantic information vector to obtain the third semantic information vector; reducing the dimension of the second semantic information vector to obtain the fourth semantic information vector; and classifying the target delivery address information and candidate address information into two categories based on the third semantic information vector and the fourth semantic information vector to obtain the probability that the target delivery address information and candidate address information belong to the same address.

[0069] One method for binary classifying target delivery address information and candidate address information based on the third and fourth semantic information vectors to obtain the probability that the target delivery address information and candidate address information belong to the same address is as follows: The third and fourth semantic information vectors are input into a normalization function to obtain the probability that the target delivery address information and candidate address information belong to the same address. Since each concatenated text sequence is labeled with a [CLS] classification tag during the construction of the concatenated text sequence, for each candidate address information, both the candidate address information and the target delivery address information are placed in the same classification tag. Consequently, the first and second semantic information vectors are also placed in the same classification tag and separated by a delimiter; the third and fourth semantic information vectors are also placed in the same classification tag and separated by a delimiter. Subsequently, the third and fourth semantic information vectors are input into a normalization function to binary classify the target delivery address information and candidate address information to obtain the probability that the target delivery address information and candidate address information belong to the same address. (See reference...) Figure 2 The process of activation function processing.

[0070] Step S103: Based on the semantic similarity of candidate texts, filter the target candidate address information that matches the target delivery address information from multiple candidate address information, and use the high confidence location point information corresponding to the target candidate address information as the target location point information of the target delivery address information.

[0071] Based on the semantic similarity of candidate texts, target candidate address information that matches the target delivery address information is selected from multiple candidate address information. This includes: sorting the candidate text semantic similarity values ​​between the target delivery address information and each candidate address information in descending order; and selecting the candidate address information with the highest similarity value as the target candidate address information. For example, the XX Hotel (back door delivery cabinet) in Street C, District B, City A is selected as the target candidate address.

[0072] Step S104: Provide the target location point information to the delivery terminal so that the delivery terminal can use the target location point to determine the delivery route for the orders to be processed.

[0073] After obtaining the target location information of the order to be processed, the route of the order to be processed can be planned based on the target location information, so that the delivery rider can deliver the order to be processed quickly and accurately.

[0074] This embodiment uses a preset text retrieval algorithm and target semantic analysis model to compare each character of the target delivery address with each character of the candidate address, thereby obtaining the semantic similarity of the candidate text. This enables the search for target location information that matches the target delivery address at the text character level, without requiring the target delivery address and candidate address strings to be completely identical, greatly improving the accuracy of target delivery address location calibration.

[0075] By using the target location information of the target delivery address obtained by the order information processing method of this embodiment for route planning, the delivery rate within 50 meters of the pending orders is significantly improved, while the delivery rate beyond 300 meters is significantly reduced. The delivery rate within 50 meters is the proportion of orders within 50 meters of the delivery rider's check-in point to the total number of orders on that day. The delivery rate beyond 300 meters is the proportion of orders beyond 300 meters of the delivery rider's check-in point to the total number of orders on that day. The calibration delivery point is the target location point.

[0076] The order information processing method provided in this application, after receiving the target delivery address information and the preset initial delivery location information for the target delivery address information of the order to be processed, recalls multiple candidate address information related to the text of the target delivery address information from a high-confidence location address information database. The target delivery address information and the multiple candidate address information are input into a target semantic analysis model to determine the candidate text semantic similarity between the target delivery address information and each candidate address information. Based on the candidate text semantic similarity, target candidate address information matching the target delivery address information is selected from the multiple candidate address information, and the high-confidence location point information corresponding to the target candidate address information is used as the target location point information of the target delivery address information. This process involves recalling multiple candidate address information from the high-confidence location address information database and, based on the determined target delivery address information and each candidate address information... The semantic similarity of candidate texts between selected address information is used to filter out target candidate address information from multiple candidate address information, thereby making the target candidate address information and the target delivery address information more semantically matched. Since the target candidate address information is from a high-confidence location address information database, which stores high-confidence location address information and corresponding high-confidence location point information, the high-confidence location point information is verified location point information. Therefore, the high-confidence location point information corresponding to the target candidate address information, i.e., the target location point information, is highly adapted to the target delivery address information. In other words, the target location point information obtained by this method is more accurate, so that the delivery route of the pending order planned based on the target location point information can facilitate the delivery rider to deliver the pending order quickly, thereby improving the delivery accuracy and efficiency of the order.

[0077] Second Embodiment Corresponding to the first embodiment, the second embodiment of this application provides another order information processing method. The executing entity in this embodiment is a delivery terminal. The parts in the second embodiment that are the same as those in the scenario embodiment and the first embodiment will not be described again; please refer to the relevant parts of the scenario embodiment and the first embodiment for details.

[0078] Please refer to Figure 3 This is a flowchart of the order information processing method provided in the second embodiment of this application.

[0079] The order information processing method of this application includes the following steps.

[0080] Step S301: Obtain the target location information corresponding to the target delivery address information of the order to be processed, provided by the server.

[0081] In this embodiment, the target location point information is high-confidence location point information corresponding to the target candidate address information. The target candidate address information is the address information that matches the target delivery address information selected from multiple candidate address information based on the semantic similarity of candidate text between the target delivery address information and each candidate address information. The multiple candidate address information is the candidate address information that is text-related to the target delivery address information and recalled from the high-confidence location address information database based on the target delivery address information and the initial delivery location point information. The high-confidence location address information database stores high-confidence location address information and high-confidence location point information corresponding to the high-confidence location address information. The high-confidence location point information is the location point information that has been verified. The initial delivery location point information is the delivery location point information preset for the target delivery address information. Step S302: Determine the delivery route for the pending orders using the target location point.

[0082] The order information processing method provided in this application involves retrieving multiple candidate address information from a high-confidence location address information database and filtering them based on the semantic similarity of the candidate text between the determined target delivery address information and each candidate address information. This makes the target candidate address information and the target delivery address information more semantically matched. Since the target candidate address information is from a high-confidence location address information database, which stores high-confidence location address information and corresponding high-confidence location point information, the high-confidence location point information is verified location point information. Therefore, the high-confidence location point information corresponding to the target candidate address information, i.e., the target location point information, is highly adapted to the target delivery address information. In other words, the target location point information obtained in this way is more accurate, making the delivery route of the pending order planned based on the target location point information easier for delivery riders to deliver the pending order quickly, thus improving the delivery accuracy and efficiency of the order.

[0083] Third Embodiment Corresponding to the order information processing method provided in the first embodiment of this application, the third embodiment of this application also provides an order information processing apparatus. Since the apparatus embodiment is basically similar to the first embodiment, the description is relatively simple; relevant details can be found in the description of the first embodiment. The apparatus embodiments described below are merely illustrative.

[0084] Please refer to Figure 4 This is a schematic diagram of the order information processing device provided in the third embodiment of this application.

[0085] The order information processing device 400, applied to a server, includes: a recall unit 401, configured to, in response to receiving target delivery address information and preset initial delivery location information for the target delivery address information, recall multiple candidate address information text-related to the target delivery address information from a high-confidence location address information database based on the target delivery address information and the initial delivery location information; the high-confidence location address information database stores high-confidence location address information and high-confidence location point information corresponding to the high-confidence location address information, wherein the high-confidence location point information is verified location point information; and a semantic similarity determination unit 402, configured to determine the semantic similarity based on the target delivery address information and the multiple candidate addresses. The system uses information to determine the semantic similarity between the target delivery address information and each candidate address information; the target semantic analysis model is a trained neural network model used to determine the semantic similarity between texts to be processed; the filtering unit 403 is used to filter target candidate address information that matches the target delivery address information from the multiple candidate address information according to the semantic similarity of the candidate text, and to use the high confidence location point information corresponding to the target candidate address information as the target location point information of the target delivery address information; the target location point information providing unit 404 is used to provide the target location point information to the delivery terminal so that the delivery terminal can use the target location point to determine the delivery route for the order to be processed.

[0086] Fourth embodiment Corresponding to the order information processing method provided in the second embodiment of this application, the fourth embodiment of this application also provides an order information processing apparatus. Since the apparatus embodiment is basically similar to the second embodiment, the description is relatively simple; relevant details can be found in the description of the second embodiment. The apparatus embodiments described below are merely illustrative.

[0087] Please refer to Figure 5 This is a schematic diagram of the order information processing device provided in the fourth embodiment of this application.

[0088] The order information processing device 500 is applied to a delivery terminal. The device includes: a target location point information acquisition unit 501, used to acquire target location point information corresponding to the target delivery address information of the order to be processed, provided by the server; the target location point information is high-confidence location point information corresponding to target candidate address information, and the target candidate address information is address information that matches the target delivery address information by filtering multiple candidate address information based on the semantic similarity of candidate text between the target delivery address information and each candidate address information; the multiple candidate address information is candidate address information related to the text of the target delivery address information recalled from a high-confidence location address information database based on the target delivery address information and initial delivery location point information; the high-confidence location address information database stores high-confidence location address information and high-confidence location point information corresponding to the high-confidence location address information, and the high-confidence location point information is verified location point information; the initial delivery location point information is delivery location point information preset for the target delivery address information; and a delivery route determination unit 502, used to determine the delivery route for the order to be processed using the target location point.

[0089] Fifth Embodiment Corresponding to the methods of the first to second embodiments of this application, the fifth embodiment of this application also provides an electronic device.

[0090] The electronic device includes: a processor; and a memory for storing a computer program. After the electronic device is powered on and runs the computer program through the processor, it executes the methods of the first to second embodiments. Figure 6 As shown, Figure 6 This is a schematic diagram of an electronic device provided according to the fifth embodiment of this application. The electronic device specifically includes: at least one processor 601, at least one communication interface 602, at least one memory 603, and at least one communication bus 604. Optionally, the communication interface 602 can be an interface for a communication module, such as the interface for a GSM module. The processor 601 may be a CPU, an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The memory 603 may include high-speed RAM and may also include non-volatile memory, such as at least one disk storage device. The memory 603 stores a program, and the processor 601 calls the program stored in the memory 603 to execute the methods of the first to second embodiments.

[0091] Sixth Embodiment Corresponding to the methods of the first to second embodiments of this application, the sixth embodiment of this application also provides a computer storage medium storing computer execution instructions, which are executed by a processor to perform the methods of the first to second embodiments of this application.

[0092] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0093] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0094] 1. Computer-readable media includes both permanent and non-permanent, removable and non-removable media, which can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined in this application, computer-readable media does not include non-transitory computer-readable media, such as modulated data signals and carrier waves.

[0095] 2. Those skilled in the art will understand that embodiments of this application can provide methods, systems, or computer program products. Therefore, embodiments of this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, embodiments of this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] 3. This application embodiment may involve the use of user data. In practical applications, user-specific personal data may be used within the scope permitted by applicable laws and regulations of the country in which the application is located (e.g., with the user's explicit consent and effective notification to the user, etc.). Furthermore, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. The collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0097] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

Claims

1. A method for processing order information, characterized in that, Applied to the server side, the method includes: In response to receiving the target delivery address information of the order to be processed and the preset initial delivery location information for the target delivery address information, based on the target delivery address information and the initial delivery location information, multiple candidate address information related to the text of the target delivery address information are recalled in a high-confidence location address information database; the high-confidence location address information database stores high-confidence location address information and high-confidence location point information corresponding to the high-confidence location address information, and the high-confidence location point information is verified location point information; Based on the target delivery address information and the multiple candidate address information, the candidate text semantic similarity between the target delivery address information and each candidate address information is determined respectively; Based on the semantic similarity of the candidate texts, target candidate address information that matches the target delivery address information is filtered from the plurality of candidate address information, and the high confidence location point information corresponding to the target candidate address information is used as the target location point information of the target delivery address information. The target location information is provided to the delivery terminal so that the delivery terminal can use the target location to determine the delivery route for the order to be processed.

2. The method according to claim 1, characterized in that, The step of recalling multiple candidate address information text-related to the target delivery address information in a high-confidence location address information database based on the target delivery address information and the initial delivery location information includes: Using the target delivery address information as the search text, retrieve related address information related to the text of the target delivery address information in the high-confidence location address information database; Based on the distance between the high-confidence location point and the initial delivery location point corresponding to the associated address information, address information that meets the first preset distance condition is selected from the associated address information as the multiple candidate address information.

3. The method according to claim 1, characterized in that, The high-confidence location address information database is constructed using the following method: Determine the historical delivery address information and historical delivery location information corresponding to historical orders; Based on the historical delivery address information, determine the historical point of interest to which the historical delivery address information belongs; Determine whether the distance between the historical delivery location and the historical point of interest meets the second preset distance condition; If the distance between the historical delivery location and the historical point of interest meets the second preset distance condition, then the historical delivery address information is stored as high-confidence location address information and the historical delivery location information is stored as high-confidence location point information to construct the high-confidence location address information database.

4. The method according to claim 3, characterized in that, Also includes: Based on the historical delivery address information of the historical orders, obtain the actual delivery location information of the historical orders; Based on the actual delivery location information, select actual delivery location points that meet the delivery conditions from the actual delivery location points. The actual delivery locations that meet the delivery conditions are aggregated to determine the aggregated delivery location, and the aggregated delivery location is used as the historical delivery location.

5. The method according to claim 1, characterized in that, The step of determining the candidate text semantic similarity between the target delivery address information and each candidate address information based on the target delivery address information and the plurality of candidate address information includes: The target delivery address information and the multiple candidate address information are input into the target semantic analysis model to determine the candidate text semantic similarity between the target delivery address information and each candidate address information; the target semantic analysis model is a trained neural network model used to determine the text semantic similarity between the texts to be processed.

6. The method according to claim 5, characterized in that, Before performing the step of inputting the target delivery address information and the plurality of candidate address information into the target semantic analysis model, the method further includes: For each candidate address information, the target delivery address information and the candidate address information are concatenated to obtain multiple concatenated text sequences; wherein, in the concatenated text, the target delivery address information and the candidate address information are separated by a delimiter; The step of inputting the target delivery address information and the plurality of candidate address information into the target semantic analysis model, and determining the candidate text semantic similarity between the target delivery address information and each candidate address information respectively, includes: inputting the plurality of concatenated text sequences into the target semantic analysis model respectively, and determining the candidate text semantic similarity between the target delivery address information and each candidate address information respectively.

7. The method according to claim 6, characterized in that, The step of inputting multiple concatenated text sequences into the target semantic analysis model to determine the semantic similarity of candidate texts between the target delivery address information and each candidate address information includes: For each concatenated text sequence, obtain a vector for each concatenated text sequence to be represented by a vector. For each concatenated text sequence vector, semantic representation is performed on the concatenated text sequence vector to obtain a first semantic information vector for vector representation of the semantics of the target delivery address information and a second semantic information vector for vector representation of the semantics of the candidate address information; Based on the first semantic information vector and the second semantic information vector, the semantic similarity of candidate texts between the target delivery address information and the candidate address information is determined.

8. The method according to claim 7, characterized in that, The step of obtaining each concatenated text sequence vector for vector representation of each concatenated text sequence includes: For each concatenated text sequence, the concatenated text sequence is segmented into characters to obtain each segmented character corresponding to the concatenated text sequence; For each segmented character, the character vector used to represent the segmented character and the position vector used to represent the position of the segmented character in the concatenated text sequence are concatenated to obtain the segmented character fusion vector; According to the position of each segmented character in the concatenated text sequence, the fusion vectors of each segmented character are concatenated to obtain the concatenated text sequence vector.

9. The method according to claim 7, characterized in that, The step of determining the candidate text semantic similarity between the target delivery address information and the candidate address information based on the first semantic information vector and the second semantic information vector includes: Based on the first semantic information vector and the second semantic information vector, the target delivery address information and the candidate address information are binary classified to obtain the probability that the target delivery address information and the candidate address information belong to the same address, and the probability that the target delivery address information and the candidate address information belong to the same address is used as the semantic similarity of the candidate text.

10. The method according to claim 9, characterized in that, The step of performing binary classification on the target delivery address information and the candidate address information based on the first semantic information vector and the second semantic information vector to obtain the probability that the target delivery address information and the candidate address information belong to the same address includes: The dimension of the first semantic information vector is reduced to obtain the third semantic information vector; the dimension of the second semantic information vector is reduced to obtain the fourth semantic information vector. Based on the third semantic information vector and the fourth semantic information vector, the target delivery address information and the candidate address information are binary classified to obtain the probability that the target delivery address information and the candidate address information belong to the same address.

11. The method according to claim 10, characterized in that, The step of performing binary classification on the target delivery address information and the candidate address information based on the third semantic information vector and the fourth semantic information vector to obtain the probability that the target delivery address information and the candidate address information belong to the same address includes: The third semantic information vector and the fourth semantic information vector are respectively input into a normalization function to obtain the probability that the target delivery address information and the candidate address information belong to the same address.

12. The method according to claim 1, characterized in that, The step of filtering target candidate address information that matches the target delivery address information from the plurality of candidate address information based on the semantic similarity of the candidate text includes: Sort the candidate text semantic similarity values ​​between the target delivery address information and each candidate address information in descending order; The candidate address information that ranks first in the similarity value ranking is used as the target candidate address information.

13. A method for processing order information, characterized in that, Applied to a delivery terminal, the method includes: The system obtains target location information corresponding to the target delivery address information of the order to be processed, provided by the server. The target location information is high-confidence location information corresponding to target candidate address information. The target candidate address information is selected from multiple candidate address information based on the semantic similarity of candidate text between the target delivery address information and each candidate address information, and matches the target delivery address information. The multiple candidate address information is text-related candidate address information recalled from a high-confidence location address information database based on the target delivery address information and initial delivery location information. The high-confidence location address information database stores high-confidence location address information and corresponding high-confidence location point information, where the high-confidence location point information is verified location point information. The initial delivery location point information is a preset delivery location point information for the target delivery address information. The delivery route for the pending orders is determined using the target location point.

14. An electronic device, characterized in that, include: processor; And a memory for storing a computer program, wherein after the electronic device is powered on and the computer program is run by the processor, it performs the method according to any one of claims 1-13.

15. A computer storage medium, characterized in that, The computer storage medium stores computer execution instructions, which are executed by a processor to perform the method described in any one of claims 1-13.

Images