Express parcel delivery method, device and equipment

Abstract

The invention discloses an express parcel delivery method, device and equipment, and the method is applied to the technical field of logistics, and the method comprises the steps: obtaining historical interaction information between a historical receiving user and a historical courier for a historical express parcel; preprocessing the historical interaction information to obtain processed interaction information; identifying key entities in the processed interaction information by adopting a named entity identification algorithm; inputting the key entity into an intention classification model to obtain a historical receiving portrait label of a historical express parcel by a historical receiving user; storing a historical receiving portrait label corresponding to the historical receiving user in a preset database; and in response to a target express parcel delivery request of a target receiving user, searching a target receiving portrait label matched with the target receiving user in a preset database, and delivering a target express parcel to the target receiving user according to the target receiving portrait label. According to the invention, accurate prediction of the receiving habit of the user is realized, and the delivery efficiency of the express parcel is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a delivery method, device and equipment of express package. BACKGROUND

[0002] The related art is mainly to summarize the receiving habits of customers through the long-term delivery experience of express delivery personnel, and to perform accurate delivery according to the receiving habits of customers. For example, if the customer needs door-to-door delivery, the express delivery personnel will deliver the package to the customer according to memory. This method requires the express delivery personnel to have delivered to the customer and to accurately understand the receiving habits of the customer, and when the next time the customer has a package to be delivered, the express delivery personnel will deliver the package by recalling memory. If there is a memory error, the delivery of the package of the customer cannot meet the receiving requirements of the customer. In addition, if the express delivery personnel in the area is replaced, the experience summary of the previous express delivery personnel cannot be inherited, and the receiving habits of the customers in the area still need to be obtained again. SUMMARY

[0003] The present application provides a delivery method, device and equipment of express package, which realizes accurate prediction of user receiving habits and improves the delivery efficiency of express package.

[0004] In one aspect, the present application provides a delivery method of express package, which comprises: obtaining historical interaction information between a historical receiving user and a historical express delivery personnel for a historical express package; the historical interaction information comprises at least one of historical call data and historical chat text; preprocessing the historical interaction information to obtain processed interaction information, and identifying key entities in the processed interaction information by using a named entity recognition algorithm; inputting the key entities into an intent classification model to obtain a historical receiving portrait label of the historical receiving user for the historical express package; the historical receiving portrait comprises at least one of historical receiving time and historical receiving method; the historical receiving portrait represents the receiving habits of the historical receiving user; storing the historical receiving portrait label corresponding to the historical receiving user in a preset database; in response to a target express package delivery request of a target receiving user, searching for a target receiving portrait label matched with the target receiving user in the preset database, and delivering the target express package to the target receiving user according to the target receiving portrait label.

[0005] In an exemplary embodiment, after the historical receiving portrait label corresponding to the historical receiving user is stored in the preset database, the method further comprises: judging whether the living habits and living environment of the historical receiving user have changed; if it is detected that at least one of the life habits and the living environment of the historical receiving user has changed, obtaining an updated receiving requirement of the historical receiving user; updating the historical receiving portrait label of the historical receiving user according to the updated receiving requirement to obtain an updated receiving portrait label; replacing the historical receiving portrait label corresponding to the historical receiving user in the preset database with the updated receiving portrait label to obtain an updated database; Correspondingly, in response to a target express package delivery request of a target receiving user, the method comprises: In response to a target express package delivery request of a target receiving user, the method comprises:

[0006] In an exemplary embodiment, after storing the historical receiving portrait label corresponding to the historical receiving user in the preset database, the method further comprises: obtaining historical delivery data of the historical receiving user in a historical period; extracting historical delivery attribute information and historical delivery demand labels according to the historical delivery data; the historical delivery attribute information comprises at least one of time, weather, location, and holiday; inputting the historical delivery attribute information into a preset model to predict the user delivery habit, and obtaining a historical delivery habit prediction result; the preset model is a recurrent neural network or a gradient boosting tree model; According to the difference between the historical delivery habit prediction result and the historical delivery demand label, the preset model is trained to obtain a user delivery habit prediction model.

[0007] In an exemplary embodiment, the method further comprises: obtaining target delivery attribute information of the target receiving user; inputting the target delivery attribute information into the user delivery habit prediction model to predict the user delivery habit, and obtaining a user delivery habit prediction result; obtaining real-time delivery demand feedback from the target receiving user; According to the difference between the real-time delivery demand and the user delivery habit prediction result, the weight of the target receiving portrait label is adjusted, and the preset database is updated.

[0008] In an exemplary embodiment, the method comprises: Obtaining sample interaction information of a sample user and a sample courier, and preprocessing the sample interaction information to obtain sample processed interaction information; Inputting the sample processed interaction information into a BiLSTM-CRF model constructed based on the named entity recognition algorithm, the BiLSTM-CRF model including an input layer, a BiLSTM layer, and a CRF layer; the sample processed interaction information is labeled with a sample key entity label; Converting the sample processed interaction information into a word vector according to the input layer; extracting context semantic features of the word vector according to the BiLSTM layer; Performing sample key entity prediction on the context semantic features according to the CRF layer to obtain a sample key entity result; Training the BiLSTM-CRF model according to the difference between the sample key entity result and the sample key entity label to obtain a key entity prediction model; Inputting the processed interaction information into the key entity prediction model for key entity prediction processing to obtain the key entity.

[0009] In an example embodiment, the historical recipient users are multiple, and the method of responding to a target recipient user's target express package delivery request, finding a target recipient image label matching the target recipient user in the preset database according to the target recipient image label, and delivering the target express package to the target recipient user, comprises: In response to a target express package delivery request of a target recipient user, obtaining target user attribute information of the target recipient user and historical user attribute information of each historical recipient user in the preset database; According to the target user attribute information and the historical user attribute information of each historical recipient user, determining the preference similarity between the target recipient user and each historical recipient user; The historical recipient image label corresponding to the historical recipient user with the highest preference similarity is determined as the target recipient image label matching the target recipient user; According to the target recipient image label, delivering the target express package to the target recipient user.

[0010] In an example embodiment, the method of delivering the target express package to the target recipient user according to the target recipient image label, comprises: Inputting the target recipient image label, the recipient address position coordinates, and the delivery time window specified by the target recipient user into a mixed integer programming and simulated annealing algorithm combination model; Taking the delivery time window and the load limit as constraint conditions, an initial feasible solution is generated; The initial feasible solution is locally searched and optimized by using a simulated annealing algorithm, the delivery sequence is randomly exchanged and the stopover point sequence is adjusted, the global optimal solution is gradually approached, and the target delivery strategy is output; According to the target delivery strategy, the target express package is delivered to the target recipient.

[0011] Another aspect provides an express package delivery device, the device comprising: A historical interaction information acquisition module is configured to acquire historical interaction information between a historical recipient and a historical express courier for a historical express package; the historical interaction information comprises at least one of historical call data and historical chat text; A key entity identification module is configured to preprocess the historical interaction information to obtain processed interaction information, and identify key entities in the processed interaction information by using a named entity recognition algorithm; A historical portrait label determination module is configured to input the key entities into an intent classification model to obtain a historical receiving portrait label of the historical express package by the historical recipient; the historical receiving portrait comprises at least one of a historical receiving time and a historical receiving method; the historical receiving portrait represents the receiving habit of the historical recipient; A label storage module is configured to store the historical receiving portrait label corresponding to the historical recipient in a preset database; A package delivery module is configured to, in response to a target express package delivery request of a target recipient, search for a target receiving portrait label matching the target recipient in the preset database, and deliver the target express package to the target recipient according to the target receiving portrait label.

[0012] In an exemplary embodiment, the device further comprises: A judgment module is configured to judge whether the living habit and living environment of the historical recipient have changed; An update requirement acquisition module is configured to, if it is detected that at least one of the living habit and living environment of the historical recipient has changed, acquire an updated receiving requirement of the historical recipient; A label update module is configured to update the historical receiving portrait label according to the updated receiving requirement to obtain an updated receiving portrait label; A database update module is configured to replace the historical receiving portrait label corresponding to the historical recipient in the preset database with the updated receiving portrait label to obtain an updated database; Correspondingly, the package delivery module is further configured to, in response to a target express package delivery request of a target recipient, search for a target receiving portrait label matching the target recipient in the updated database.

[0013] In an example embodiment, the device further comprises: a historical data acquisition module configured to acquire historical delivery data of the historical in-box user in a historical period; a historical label extraction module configured to extract historical delivery attribute information and historical delivery demand labels from the historical delivery data; the historical delivery attribute information comprises at least one of time, weather, location, and holiday; a delivery habit prediction module configured to input the historical delivery attribute information into a preset model to predict a historical delivery habit of the user, and obtain a historical delivery habit prediction result; the preset model is a recurrent neural network or a gradient boosting tree model; a model training module configured to train the preset model according to a difference between the historical delivery habit prediction result and the historical delivery demand labels, and obtain a user delivery habit prediction model.

[0014] In an example embodiment, the device further comprises: a target attribute acquisition module configured to acquire target delivery attribute information of the target in-box user; a delivery habit prediction module configured to input the target delivery attribute information into the user delivery habit prediction model to predict a user delivery habit, and obtain a user delivery habit prediction result; a real-time demand acquisition module configured to acquire real-time delivery demand feedback by the target in-box user; a weight updating module configured to adjust a weight of the target in-box portrait label according to a difference between the real-time delivery demand and the user delivery habit prediction result, and update the preset database.

[0015] In an example embodiment, the key entity recognition module comprises: a sample interaction information acquisition unit configured to acquire sample interaction information of a sample user and a sample courier, and pre-process the sample interaction information to obtain sample processed interaction information; an information input unit configured to input the sample processed interaction information into a BiLSTM-CRF model constructed based on the named entity recognition algorithm, the BiLSTM-CRF model comprising an input layer, a BiLSTM layer, and a CRF layer; the sample processed interaction information is labeled with sample key entity labels; a feature extraction unit configured to convert the sample processed interaction information into a word vector according to the input layer; and extract context semantic features of the word vector according to the BiLSTM layer; a sample entity prediction unit configured to perform sample key entity prediction on the context semantic features according to the CRF layer to obtain a sample key entity result; a model training unit configured to train the BiLSTM-CRF model according to a difference between the sample key entity result and the sample key entity label to obtain a key entity prediction model; a key entity prediction unit configured to input the processed interaction information into the key entity prediction model for key entity prediction processing to obtain the key entity.

[0016] In an example embodiment, the historical consignee users are multiple, and the package delivery module comprises: a target attribute acquisition unit configured to acquire target user attribute information of a target consignee user and historical user attribute information of each historical consignee user in the preset database in response to a target express package delivery request of the target consignee user; a preference similarity determination unit configured to determine a preference similarity between the target consignee user and each historical consignee user according to the target user attribute information and the historical user attribute information of each historical consignee user; a label determination unit configured to determine a historical consignee portrait label corresponding to a historical consignee user with the highest preference similarity as a target consignee portrait label matched with the target consignee user; a target package delivery unit configured to deliver the target express package to the target consignee user according to the target consignee portrait label.

[0017] In an example embodiment, the target package delivery unit comprises: a data input subunit configured to input the target consignee portrait label, consignee address position coordinates, and a delivery time window specified by the target consignee user into a mixed integer programming and simulated annealing algorithm combination model; an initial solution generation subunit configured to generate an initial feasible solution by taking the delivery time window and load limit as constraint conditions; a target strategy output subunit configured to perform local search optimization on the initial feasible solution by using a simulated annealing algorithm, gradually approach a global optimal solution by randomly exchanging delivery order and adjusting stopover point order, and output a target delivery strategy; a package delivery subunit configured to deliver the target express package to the target consignee user according to the target delivery strategy.

[0018] The other aspect provides an electronic device, the device includes a processor and a memory, the memory has at least one instruction or at least one program stored, the at least one instruction or the at least one program is loaded and executed by the processor to realize the delivery method of express package as described above.

[0019] The other aspect provides a computer storage medium, the computer storage medium has at least one instruction or at least one program stored, the at least one instruction or the at least one program is loaded and executed by the processor to realize the delivery method of express package as described above.

[0020] The other aspect provides a computer program product or computer program, the computer program product or computer program includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes to realize the delivery method of express package as described above.

[0021] The delivery method, device and equipment of express package provided by the application have the following technical effects: The application obtains historical interaction information between a historical consignee user and a historical express delivery person for a historical express package. The historical interaction information includes at least one of historical call data and historical chat text. The historical interaction information is preprocessed to obtain processed interaction information. Key entities in the processed interaction information are identified by using a named entity recognition algorithm. The key entities are input into an intent classification model to obtain a historical consignee image label of the historical consignee user for the historical express package. The historical consignee image includes at least one of a historical consignee time and a historical consignee manner. The historical consignee image represents a consignee habit of the historical consignee user. The historical consignee image label corresponding to the historical consignee user is stored in a preset database. In response to a target express package delivery request of a target consignee user, a target consignee image label matched with the target consignee user is searched in the preset database, and the target express package is delivered to the target consignee user according to the target consignee image label. The application solves the problems of high artificial dependence, data update lag and low delivery efficiency in the prior art, and realizes accurate prediction of a consignee habit of a user. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions and advantages of the embodiments or prior art in the specification, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0023] Figure 1 is a schematic diagram of a delivery system for express packages provided by an embodiment of the present specification; Figure 2 is a flowchart of a delivery method for express packages provided by an embodiment of the present specification; Figure 3 is a flowchart of a method for identifying key entities in processed interaction information using a named entity recognition algorithm provided by an embodiment of the present specification; Figure 4 is a flowchart of a method for training a user delivery habit prediction model provided by an embodiment of the present specification; Figure 5 is a flowchart of a method for delivering target express packages to target recipients according to target recipient profile tags provided by an embodiment of the present specification; Figure 6 is a structural diagram of a delivery device for express packages provided by an embodiment of the present specification; Figure 7 is a structural diagram of a server provided by an embodiment of the present specification. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present specification will be described clearly and completely below in conjunction with the drawings in the embodiments of the present specification. Obviously, the described embodiments are only part of the embodiments of the present specification, rather than all the embodiments. Based on the embodiments in the present specification, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present specification.

[0025] It can be understood that in the specific embodiments of the present specification, user information and other related data are involved. When the above embodiments of the present specification are applied to specific products or technologies, user permission or consent needs to be obtained, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards in relevant countries and regions.

[0026] It should be noted that the terms "first", "second", and the like in the description and claims of the application and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0027] Please refer to Figure 1 , Figure 1 is a schematic diagram of a delivery system for express packages provided by an embodiment of the present specification, as Figure 1 shown, the delivery system for express packages can at least include a server 01 and a client 02.

[0028] Specifically, in the present specification, the server 01 can include a standalone server, or a distributed server, or a server cluster composed of multiple servers, and can also be a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms. The server 01 can include a network communication unit, a processor, a memory, and the like. Specifically, the server 01 can be used to train an intent classification model and build a preset database.

[0029] Specifically, in the present specification, the client 02 can include a smart phone, a desktop computer, a tablet computer, a notebook computer, a digital assistant, a smart wearable device, a smart speaker, a vehicle terminal, a smart television, and the like. It can also include software running on physical devices, such as web pages provided by some service providers to users, and applications provided by these service providers to users. Specifically, the client 02 can be used to query online a target recipient portrait label matched with the target recipient user, and deliver the target express package to the target recipient user according to the target recipient portrait label.

[0030] The following describes a delivery method for express packages, Figure 2is a flowchart of a delivery method of an express package provided by an embodiment of the present specification. The present specification provides method operation steps as described in the embodiments or flowcharts, but more or fewer operation steps can be included based on conventional or non-inventive labor. The order of steps listed in the embodiments is only one of the many step execution orders, and does not represent the only execution order. In actual system or server product execution, the method order shown in the embodiments or the drawings can be executed in sequence or in parallel (for example, in a parallel processor or multi-thread processing environment). Specifically as shown in Figure 2 The method can include: S201: Obtain historical interaction information between a historical recipient user and a historical courier for a historical express package; the historical interaction information includes at least one of historical call data and historical chat text. In an embodiment of the present specification, the historical express package is a package delivered by a historical courier to a historical recipient user, the historical call data is call data between the historical recipient user and the historical courier, the historical chat text is dialogue text between the historical recipient user and the historical courier in a delivery application or related application, and the historical interaction information can include information such as delivery time, delivery method, and delivery target location of the historical express package determined by the historical recipient user.

[0031] S203: Preprocess the historical interaction information to obtain processed interaction information, and use a named entity recognition algorithm to identify key entities in the processed interaction information.

[0032] In an embodiment of the present specification, historical call data (call voice) is converted into text by intelligent speech recognition technology, and a customer demand corpus is constructed in combination with historical chat text. Preprocessing can include processing such as word segmentation and stop word removal on the text. Key entities can include information such as delivery time, delivery method, and delivery target location, such as "pick-up point", "workday 18:00", and "front desk".

[0033] S205: Input the key entities into an intent classification model to obtain a historical receipt portrait label of the historical express package by the historical recipient user; the historical receipt portrait includes at least one of historical receipt time and historical receipt method; and the historical receipt portrait represents the receipt habits of the historical recipient user.

[0034] In the embodiment of the present application, the intent classification model can be used to obtain the historical receipt image label of the historical delivery package of the historical receipt user according to the key entity. The historical receipt image is used to determine the receipt habit of the historical receipt user. The core demand type of the customer is determined by the intent classification model, such as delivery method, time window, and special requirements. Finally, the extracted entity and the classification result are integrated into a multi-dimensional label, such as "delivery method = home delivery" and "time preference = weekday evening", and stored in the customer image database.

[0035] S207: Store the historical receipt image label corresponding to the historical receipt user in the preset database.

[0036] In the embodiment of the present application, a plurality of historical receipt image labels corresponding to a plurality of historical receipt users can be obtained, and a preset database can be constructed.

[0037] S209: In response to a target delivery package delivery request of a target receipt user, a target receipt image label matching the target receipt user is found in the preset database, and the target delivery package is delivered to the target receipt user according to the target receipt image label.

[0038] Wherein, the target receipt image label can include target delivery time, target delivery method, target delivery location and other information of the target delivery package, so as to realize accurate delivery of the target delivery package and meet the user's receipt habit.

[0039] In the embodiment, the historical interaction information between the historical receipt user and the historical delivery man for the historical delivery package is obtained; the historical interaction information includes at least one of historical call data and historical chat text; the historical interaction information is preprocessed to obtain processed interaction information; and a named entity recognition algorithm is used to identify key entities in the processed interaction information; the key entities are input into an intent classification model to obtain a historical receipt image label of the historical receipt user for the historical delivery package; the historical receipt image includes at least one of historical receipt time and historical receipt method; the historical receipt image represents the receipt habit of the historical receipt user; the historical receipt image label corresponding to the historical receipt user is stored in a preset database; in response to a target delivery package delivery request of a target receipt user, a target receipt image label matching the target receipt user is found in the preset database, and the target delivery package is delivered to the target receipt user according to the target receipt image label. The present application solves the problems of high artificial dependence, data update lag and low delivery efficiency in the prior art, and realizes accurate prediction of the user's receipt habit.

[0040] In the embodiment of the present application, when the customer (recipient) calls the courier to tell the courier that the customer's package needs to be delivered to the door or placed in the pick-up point, the courier will mark the customer's receiving portrait in addition to the current single requirement of the customer, that is, through the customer receiving marking function of the courier's APP, fill in the customer's mobile phone number and select the customer's time habits such as door-to-door delivery, and then submit the portrait data.

[0041] After the courier delivers the customer's package, if the customer reflects to the express company that the courier does not deliver according to the customer's receiving requirements, the express company will tell the courier that the customer needs to be delivered according to the requirements of door-to-door delivery or placing in the pick-up point, and the courier will use the customer marking function to mark the delivery portrait of the customer.

[0042] In the embodiment of the present application, as shown in Figure 3 The method comprises the following steps: S2031: obtaining sample interaction information of a sample user and a sample courier, and pre-processing the sample interaction information to obtain sample processed interaction information; S2033: inputting the sample processed interaction information into a BiLSTM-CRF model constructed based on the named entity recognition algorithm, wherein the BiLSTM-CRF model comprises an input layer, a BiLSTM layer and a CRF layer; and the sample processed interaction information is marked with a sample key entity label; S2035: converting the sample processed interaction information into a word vector according to the input layer; and extracting context semantic features of the word vector according to the BiLSTM layer; S2037: performing sample key entity prediction on the context semantic features according to the CRF layer to obtain a sample key entity result; S2039: training the BiLSTM-CRF model according to the difference between the sample key entity result and the sample key entity label to obtain a key entity prediction model; S20311: inputting the processed interaction information into the key entity prediction model for key entity prediction processing to obtain the key entity.

[0043] In the embodiments of the present specification, the named entity recognition adopts a BiLSTM-CRF model, the input layer converts the text into a word vector, the BiLSTM layer captures the context semantic features, and the CRF layer optimizes the label sequence prediction. The model is trained by a labeled express scene corpus, and the recognition accuracy can reach more than 95%; the intent classification model is based on a pre-trained language model BERT, and through fine-tuning of an express demand data set, the text is semantically encoded and then connected to a full connection layer to realize multi-classification, which can distinguish more than ten delivery intents, such as urgent items, refrigeration requirements, designated collection points, etc. Specifically, the target loss data can be determined according to the difference between the sample key entity result and the sample key entity label; then the model parameters of the BiLSTM-CRF model are adjusted according to the target loss data until the training end condition is met, and the model at the end of training is determined as the key entity prediction model; wherein the training end condition can be determined according to at least one of the target loss data and the number of training iterations. For example, the training end condition can be set as the target loss data being less than a preset threshold. Thus, the key entity prediction model can be used to predict the key entity of the processed interaction information quickly and accurately.

[0044] In the embodiments of the present specification, after storing the historical receiving portrait label corresponding to the historical receiving user in the preset database, the method further comprises: determining whether the living habits and living environment of the historical receiving user have changed; if it is detected that at least one of the living habits and living environment of the historical receiving user has changed, obtaining an updated receiving requirement of the historical receiving user; updating the historical receiving portrait label according to the updated receiving requirement to obtain an updated receiving portrait label; replacing the historical receiving portrait label corresponding to the historical receiving user in the preset database with the updated receiving portrait label to obtain an updated database; Correspondingly, in response to a target express package delivery request of a target receiving user, searching for a target receiving portrait label matching the target receiving user in the preset database, comprising: in response to a target express package delivery request of a target receiving user, searching for a target receiving portrait label matching the target receiving user in the updated database.

[0045] In the embodiments of the present application, when the customer's living habits and living environment change, the delivery requirements may change accordingly, and the delivery personnel will re-label the customer's delivery portrait according to the customer's new delivery requirements. After submitting the customer delivery portrait, the data enters the customer delivery portrait database, which is used for subsequent customer package delivery requirement labeling. When the preset database is updated to an updated database, a target delivery portrait label matching the target delivery user can be found in the updated database in response to a target delivery package delivery request of the target delivery user.

[0046] In the embodiments of the present application, after storing the historical delivery portrait label corresponding to the historical delivery user in the preset database, as shown in Figure 4 The method further comprises: S401: Obtain historical delivery data of the historical delivery user in a historical period; S403: Extract historical delivery attribute information and historical delivery demand label according to the historical delivery data; the historical delivery attribute information includes at least one of time, weather, location and holiday; S405: Input the historical delivery attribute information into a preset model for user delivery habit prediction to obtain a historical delivery habit prediction result; the preset model is a recurrent neural network or a gradient boosting tree model; S407: According to the difference between the historical delivery habit prediction result and the historical delivery demand label, train the preset model to obtain a user delivery habit prediction model.

[0047] In the embodiments of the present application, a user delivery habit prediction model can be constructed based on the historical delivery data of the historical delivery user in the historical period, to automatically infer potential demand and dynamically adjust the portrait weight according to the real-time feedback of the customer. Specifically, after the delivery portrait data enters the portrait database of the express company, when a new order is generated, the system will check whether there is a matching delivery customer in the portrait data. If there is, the order will be labeled according to the delivery requirements of the portrait and synchronized to the system that needs to use the delivery portrait.

[0048] Collect customer historical delivery data, including delivery time, location, signing method, change record, etc., construct time series feature matrix, such as "door-to-door" frequency in the past 30 days, time period distribution of collection point, train customer habit prediction model through recurrent neural network or gradient boosting tree model, input contains time, weather, holiday, etc. Contextual features, output potential demand labels, such as "rainy day tends to collection point" and "weekend refuses to deliver", while introducing time difference learning algorithm, when customers give real-time feedback, such as temporary change of delivery requirements, calculate the error value of predicted label and actual demand, dynamically adjust the weight of portrait, realize real-time calibration of portrait. Among them, the preset model can adopt RNN model, the RNN model adopts long short-term memory (LSTM, Long Short-Term Memory) unit, inputs the delivery records of the past 7 days in time order into the network, captures the long-term dependence relationship through the hidden layer, and outputs the label probability through the Softmax function; Temporal-difference learning (TDLearning) maintains the value function of each label, calculates the difference between the immediate reward and the next state value estimation at the moment of customer feedback, updates the value function to adjust the label priority.

[0049] In the embodiments of the present specification, the method further comprises: obtaining target delivery attribute information of the target delivery user; inputting the target delivery attribute information into the user delivery habit prediction model to predict the user delivery habit, and obtaining a user delivery habit prediction result; obtaining real-time delivery demand feedback by the target delivery user; According to the difference between the real-time delivery demand and the user delivery habit prediction result, adjust the weight of the target delivery portrait label, and update the preset database.

[0050] In the embodiments of the present specification, the target delivery attribute information and the historical delivery attribute information are information of the same type, the target delivery attribute information includes delivery information corresponding to at least one of time, weather, location, and holiday; the target delivery attribute information is input into a user delivery habit prediction model to predict the user delivery habit, and a user delivery habit prediction result is obtained; real-time delivery demand feedback by a target receiving user is obtained; and the weight of a target receiving portrait label is adjusted according to the difference between the real-time delivery demand and the user delivery habit prediction result. Specifically, the real-time delivery demand and the user delivery habit prediction result can include the same type of data, the target delivery attribute information can be used to analyze the target receiving portrait label, and in the model prediction process, the user delivery habit prediction result can be determined according to the weight of each target receiving portrait label, so that the weight of each target receiving portrait label can be adjusted according to the difference between the real-time delivery demand and the user delivery habit prediction result; until the difference between the two satisfies a preset condition; for example, the difference between the two is less than a target value to satisfy the preset condition. Thus, the preset database is updated to improve the accuracy of the data in the preset database.

[0051] In the embodiments of the present specification, the historical receiving users are multiple, as shown in Figure 5 As shown, in response to a target express package delivery request of a target receiving user, a target receiving portrait label matched with the target receiving user is searched in the preset database, and the target express package is delivered to the target receiving user according to the target receiving portrait label, comprising: S2091: In response to a target express package delivery request of a target receiving user, target user attribute information of the target receiving user and historical user attribute information of each historical receiving user in the preset database are obtained; S2093: According to the target user attribute information and the historical user attribute information of each historical receiving user, a preference similarity between the target receiving user and each historical receiving user is determined; S2095: A historical receiving portrait label corresponding to a historical receiving user with the highest preference similarity is determined as a target receiving portrait label matched with the target receiving user; S2097: According to the target receiving portrait label, the target express package is delivered to the target receiving user.

[0052] In the embodiments of the present specification, in the order generation stage, the collaborative filtering algorithm automatically matches high-frequency tags for new orders according to customer historical tags, similar customer group behavior, and preference similarity between customers calculated by cosine similarity. The target user attribute information of the target recipient user and the historical user attribute information of each historical recipient user in the preset database can be obtained in response to a target express package delivery request of a target recipient user. Then, the preference similarity between the target recipient user and each historical recipient user is determined according to the target user attribute information and the historical user attribute information of each historical recipient user. The historical recipient portrait tag corresponding to the historical recipient user with the highest preference similarity is determined as the target recipient portrait tag matched with the target recipient user, so that the target recipient portrait tag can be quickly determined, and the receiving habit of the target recipient user can be determined.

[0053] In the embodiments of the present specification, the target express package is delivered to the target recipient user according to the target recipient portrait tag, including: The target recipient portrait tag, the receiving address position coordinates, and the delivery time window specified by the target recipient user are input into a mixed integer programming and simulated annealing algorithm combination model; The delivery time window and the load limit are used as constraint conditions to generate an initial feasible solution; The initial feasible solution is locally searched and optimized by using a simulated annealing algorithm, the delivery sequence is randomly exchanged, the stopover point sequence is adjusted, the global optimal solution is gradually approached, and a target delivery order strategy is output; The target express package is delivered to the target recipient user according to the target delivery order strategy.

[0054] In the embodiment of the present specification, the package label, geographic position coordinates, customer specified time window input mixed integer programming and simulated annealing algorithm combined model: first, a path optimization mathematical model with constraints is constructed by MIP, the objective function is to minimize travel time, the constraint conditions include time window, load limit, and an initial feasible solution is generated; then the solution is locally searched and optimized by using the simulated annealing algorithm, the delivery order and the stop point order are randomly exchanged and adjusted to gradually approach the global optimal solution, and finally the optimal route and sorting priority are output; the collaborative filtering adopts the user-based CF algorithm, constructs the customer-label co-occurrence matrix, calculates the cosine similarity between the target customer and other customers, and takes the high-frequency labels of the top 10 similar customers as the recommended labels; in the path optimization, the MIP model models the problem as a vehicle routing problem with time windows, the MIP model (mixed integer programming model) is a mathematical optimization model combining linear programming and integer programming, its core feature is to limit part of the decision variables to integers, and the other part allows to be continuous real numbers, and it is widely used in logistics, manufacturing, resource allocation and other fields. The initial route is generated by the dynamic programming algorithm (Gurobi) solver, the simulated annealing algorithm controls the randomness of the search with a temperature parameter, and the solution that makes the objective function value decrease is selected and retained in the neighborhood solution, the temperature is reduced by 10% every 100 iterations, and finally converges to an efficient delivery route.

[0055] From the technical solutions provided by the above embodiments of the present specification, the historical interaction information between the historical recipient user and the historical courier for the historical express package is obtained; the historical interaction information includes at least one of historical call data and historical chat text; the historical interaction information is preprocessed to obtain processed interaction information; and a named entity recognition algorithm is used to identify key entities in the processed interaction information; the key entities are input into an intent classification model to obtain historical recipient portrait labels of the historical recipient user for the historical express package; the historical recipient portrait includes at least one of historical recipient time and historical recipient method; the historical recipient portrait represents the habits of the historical recipient user; the historical recipient portrait label corresponding to the historical recipient user is stored in a preset database; in response to a target express package delivery request of a target recipient user, a target recipient portrait label matching the target recipient user is found in the preset database, and the target express package is delivered to the target recipient user according to the target recipient portrait label. The present application solves the problems of high artificial dependence, data update lag and low delivery efficiency in the prior art, and realizes accurate prediction of user receipt habits.

[0056] The embodiment of the present specification also provides a delivery device for express packages, as shown in Figure 6 The device comprises: The historical interaction information acquisition module 610 is used to acquire historical interaction information between historical recipients and historical couriers regarding historical express parcels; the historical interaction information includes at least one of historical call data and historical chat text. The key entity recognition module 620 is used to preprocess the historical interaction information to obtain processed interaction information; and to use a named entity recognition algorithm to identify key entities in the processed interaction information. The historical profile tag determination module 630 is used to input the key entity into the intent classification model to obtain the historical receipt profile tag of the historical recipient for the historical express package; the historical receipt profile includes at least one of historical receipt time and historical receipt method; the historical receipt profile represents the historical recipient's receipt habits. Tag storage module 640 is used to store historical recipient profile tags corresponding to the historical recipients in a preset database; The parcel delivery module 650 is used to respond to a target parcel delivery request from a target recipient, search the preset database for a target recipient profile tag that matches the target recipient, and deliver the target parcel to the target recipient based on the target recipient profile tag.

[0057] In one exemplary embodiment, the apparatus further includes: The judgment module is used to determine whether the living habits and living environment of the historical recipients have changed; The update request acquisition module is used to acquire the update receipt request of the historical recipient if it is detected that at least one of the historical recipient's lifestyle habits or living environment has changed. The tag update module is used to update the historical recipient profile tags according to the updated recipient requirements to obtain updated recipient profile tags; The database update module is used to replace the historical recipient profile tags corresponding to the historical recipients in the preset database with the updated recipient profile tags to obtain the updated database. Correspondingly, the package delivery module is also used to respond to the target recipient's target express package delivery request by searching the updated database for a target recipient profile tag that matches the target recipient.

[0058] In one exemplary embodiment, the apparatus further includes: The historical data acquisition module is used to acquire the historical delivery data of the historical recipient within a historical time period; The historical tag extraction module is used to extract historical delivery attribute information and historical delivery demand tags based on the historical delivery data; the historical delivery attribute information includes at least one of time, weather, location, and holidays; The delivery habit prediction module is used to input the historical delivery attribute information into a preset model to predict user delivery habits and obtain historical delivery habit prediction results; the preset model is a recurrent neural network or a gradient boosting tree model. The model training module is used to train the preset model based on the difference between the historical delivery habit prediction results and the historical delivery demand tags to obtain a user delivery habit prediction model.

[0059] In one exemplary embodiment, the apparatus further includes: The target attribute acquisition module is used to acquire the target delivery attribute information of the target recipient user; The delivery habit prediction module is used to input the target delivery attribute information into the user delivery habit prediction model to predict the user delivery habit and obtain the user delivery habit prediction result. The real-time demand acquisition module is used to acquire the real-time delivery demand fed back by the target recipient user; The weight update module is used to adjust the weight of the target recipient profile tag based on the difference between the real-time delivery demand and the predicted results of the user delivery habits, and to update the preset database.

[0060] In one exemplary embodiment, the key entity identification module includes: The sample interaction information acquisition unit is used to acquire sample interaction information between sample users and sample couriers, and to preprocess the sample interaction information to obtain sample processing interaction information. An information input unit is used to input the sample processing interaction information into a BiLSTM-CRF model constructed based on the named entity recognition algorithm. The BiLSTM-CRF model includes an input layer, a BiLSTM layer, and a CRF layer. The sample processing interaction information is labeled with key entity tags of the sample. The feature extraction unit is used to convert the sample processing interaction information into word vectors based on the input layer; and to extract the contextual semantic features of the word vectors based on the BiLSTM layer. The sample entity prediction unit is used to predict the key entities of the sample based on the context semantic features of the CRF layer, and obtain the key entity results of the sample. The model training unit is used to train the BiLSTM-CRF model based on the difference between the sample key entity results and the sample key entity labels to obtain a key entity prediction model. The key entity prediction unit is used to input the processed interaction information into the key entity prediction model to perform key entity prediction processing and obtain the key entity.

[0061] In one exemplary embodiment, there are multiple historical recipients, and the package delivery module includes: The target attribute acquisition unit is used to acquire the target user attribute information of the target recipient and the historical user attribute information of each historical recipient in the preset database in response to the target recipient's target express package delivery request. The preference similarity determination unit is used to determine the preference similarity between the target user and each historical user based on the target user attribute information and the historical user attribute information of each historical recipient. The tag determination unit is used to determine the historical recipient profile tag corresponding to the historical recipient with the highest preference similarity as the target recipient profile tag that matches the target recipient. The target package delivery unit is used to deliver the target express package to the target recipient based on the target recipient profile tag.

[0062] In one exemplary embodiment, the target package delivery unit includes: The data input subunit is used to input the target recipient profile label, the recipient address location coordinates, and the delivery time window specified by the target recipient user into the hybrid integer programming and simulated annealing algorithm combination model. An initial solution generation sub-unit is used to generate an initial feasible solution using the dispatch time window and load limit as constraints. The target strategy output subunit is used to perform local search optimization on the initial feasible solution using the simulated annealing algorithm. By randomly exchanging the delivery order and adjusting the stop order, it gradually approaches the global optimal solution and outputs the target dispatch strategy. The parcel delivery subunit is used to deliver the target express parcel to the target recipient according to the target order dispatch strategy.

[0063] The apparatus and method embodiments described herein are based on the same inventive concept.

[0064] This specification provides an electronic device including a processor and a memory. The memory stores at least one instruction or at least one program, which is loaded and executed by the processor to implement the express parcel delivery method provided in the above method embodiments.

[0065] Embodiments of the present invention also provide a computer storage medium, which can be disposed in a terminal to store at least one instruction or at least one program related to implementing a method for delivering express parcels in the method embodiments. The at least one instruction or at least one program is loaded and executed by the processor to implement the method for delivering express parcels provided in the above-described method embodiments.

[0066] Embodiments of the present invention also provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the parcel delivery method provided in the above-described method embodiments.

[0067] Optionally, in the embodiments of this specification, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0068] The memory described in the embodiments of this specification can be used to store software programs and modules. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for the functions, etc.; the data storage area may store data created according to the use of the device, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory may also include a memory controller to provide the processor with access to the memory.

[0069] The parcel delivery method embodiments provided in this specification can be executed on a mobile terminal, computer terminal, server, or similar computing device. Taking a server as an example... Figure 7 This is a hardware structure block diagram of a server for a method of delivering express parcels, as provided in the embodiments of this specification. Figure 7As shown, the server 700 can vary significantly due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 710 (CPUs 710 may include, but are not limited to, microprocessors (MCUs) or programmable logic devices (FPGAs), a memory 730 for storing data, and one or more storage media 720 (e.g., one or more mass storage devices) for storing application programs 723 or data 722. The memory 730 and storage media 720 may be temporary or persistent storage. The program stored in the storage media 720 may include one or more modules, each module may include a series of instruction operations on the server. Furthermore, the CPU 710 may be configured to communicate with the storage media 720 and execute the series of instruction operations stored in the storage media 720 on the server 700. Server 700 may also include one or more power supplies 760, one or more wired or wireless network interfaces 750, one or more input / output interfaces 740, and / or one or more operating systems 721, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0070] The input / output interface 740 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of server 700. In one example, the input / output interface 740 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 740 may be a radio frequency (RF) module used for wireless communication with the Internet.

[0071] Those skilled in the art will understand that Figure 7 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, server 700 may also include... Figure 7 The more or fewer components shown, or having the same Figure 7 The different configurations shown.

[0072] As can be seen from the embodiments of the express parcel delivery method, apparatus, electronic device, or storage medium provided by the present invention, the present invention acquires historical interaction information between historical recipients and historical couriers regarding historical express parcels; the historical interaction information includes at least one of historical call data and historical chat text; the historical interaction information is preprocessed to obtain processed interaction information; and a named entity recognition algorithm is used to identify key entities in the processed interaction information; the key entities are input into an intent classification model to obtain historical recipient profile tags for historical express parcels; the historical recipient profile includes at least one of historical receipt time and historical receipt method; the historical recipient profile represents the historical recipient's receipt habits; the historical recipient profile tags corresponding to historical recipients are stored in a preset database; in response to the target recipient's target express parcel delivery request, the target recipient profile tag matching the target recipient is searched in the preset database, and the target express parcel is delivered to the target recipient according to the target recipient profile tag. The present invention solves the problems of high manual dependence, delayed data updates, and low delivery efficiency in the prior art, and achieves accurate prediction of user receipt habits.

[0073] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0074] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0075] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer storage medium, such as a read-only memory, a disk, or an optical disk.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A training method for a user mail receipt profile prediction model, characterized in that, The method includes: Obtain historical interaction information between historical recipients and historical couriers regarding historical parcels; the historical interaction information includes at least one of historical call data and historical chat text. The historical interaction information is preprocessed to obtain processed interaction information; and a named entity recognition algorithm is used to identify key entities in the processed interaction information. The key entities are input into the intent classification model to obtain the historical receipt profile tags of the historical recipients of the historical express parcels; the historical receipt profile includes at least one of historical receipt time and historical receipt method; the historical receipt profile represents the receipt habits of the historical recipients. The historical recipient profile tags corresponding to the historical recipients are stored in a preset database; In response to a target recipient's request for delivery of a target parcel, the system searches the preset database for a target recipient profile tag that matches the target recipient, and then delivers the target parcel to the target recipient based on the target recipient profile tag.

2. The method according to claim 1, characterized in that, After storing the historical recipient profile tags corresponding to the historical recipients in the preset database, the method further includes: Determine whether the historical recipients' lifestyles and living environments have changed; If at least one of the historical recipient's lifestyle habits or living environment is detected to have changed, the updated receipt request of the historical recipient is obtained; The historical recipient profile tags are updated according to the updated recipient requirements to obtain updated recipient profile tags; Replace the historical recipient profile tags corresponding to the historical recipients in the preset database with the updated recipient profile tags to obtain the updated database; Accordingly, the step of responding to the target recipient's target parcel delivery request by searching the preset database for a target recipient profile tag that matches the target recipient includes: In response to a target recipient's request for delivery of a target parcel, the system searches the updated database for a target recipient profile tag that matches the target recipient.

3. The method according to claim 1, characterized in that, After storing the historical recipient profile tags corresponding to the historical recipients in the preset database, the method further includes: Obtain the historical delivery data of the historical recipients within the historical time period; Historical delivery attribute information and historical delivery demand tags are extracted from the historical delivery data; the historical delivery attribute information includes at least one of time, weather, location, and holidays; The historical delivery attribute information is input into a preset model to predict user delivery habits, and the historical delivery habit prediction result is obtained; the preset model is a recurrent neural network or a gradient boosting tree model. Based on the difference between the historical delivery habit prediction results and the historical delivery demand tags, the preset model is trained to obtain the user delivery habit prediction model.

4. The method according to claim 3, characterized in that, The method further includes: Obtain the target delivery attribute information of the target recipient user; The target delivery attribute information is input into the user delivery habit prediction model to predict user delivery habits and obtain the user delivery habit prediction result. Obtain the real-time delivery request from the target recipient; Based on the difference between the real-time delivery demand and the predicted results of user delivery habits, the weight of the target recipient profile tags is adjusted, and the preset database is updated.

5. The method according to claim 1, characterized in that, The process of using a named entity recognition algorithm to identify key entities in the processed interactive information includes: Obtain sample interaction information between sample users and sample couriers, and preprocess the sample interaction information to obtain sample processing interaction information; The sample processing interaction information is input into a BiLSTM-CRF model constructed based on the named entity recognition algorithm. The BiLSTM-CRF model includes an input layer, a BiLSTM layer, and a CRF layer. The sample processing interaction information is labeled with key entity tags of the sample. The input layer converts the sample processing interaction information into word vectors; the BiLSTM layer extracts the contextual semantic features of the word vectors. Based on the CRF layer, the sample key entity prediction is performed on the context semantic features to obtain the sample key entity result; Based on the difference between the sample key entity results and the sample key entity labels, the BiLSTM-CRF model is trained to obtain the key entity prediction model; The processed interactive information is input into the key entity prediction model for key entity prediction processing to obtain the key entity.

6. The method according to claim 1, characterized in that, The historical recipients are multiple. The step of responding to a target recipient's request for delivery of a target parcel, searching the preset database for a target recipient profile tag matching the target recipient, and delivering the target parcel to the target recipient based on the target recipient profile tag, includes: In response to a target parcel delivery request from a target recipient, the system obtains the target user attribute information of the target recipient and the historical user attribute information of each historical recipient in the preset database. Based on the target user attribute information and the historical user attribute information of each historical recipient, determine the preference similarity between the target recipient and each historical recipient; The historical message profile tags corresponding to the historical recipients with the highest preference similarity are determined as the target message profile tags that match the target recipients. Based on the target recipient profile tags, the target express package is delivered to the target recipient.

7. The method according to claim 6, characterized in that, The step of delivering the target parcel to the target recipient based on the target recipient profile tag includes: Input the target recipient profile label, the recipient address location coordinates, and the delivery time window specified by the target recipient user into a hybrid integer programming and simulated annealing algorithm combination model. Using the delivery time window and load limit as constraints, an initial feasible solution is generated; The simulated annealing algorithm is used to perform local search optimization on the initial feasible solution. By randomly exchanging the delivery order and adjusting the stop order, the global optimal solution is gradually approached, and the target dispatch strategy is output. The target parcel is delivered to the target recipient according to the target order dispatch strategy.

8. A delivery device for express parcels, characterized in that, The device includes: The historical interaction information acquisition module is used to acquire historical interaction information between historical recipients and historical couriers regarding historical express parcels; the historical interaction information includes at least one of historical call data and historical chat text. The key entity identification module is used to preprocess the historical interaction information to obtain processed interaction information; and to use a named entity recognition algorithm to identify key entities in the processed interaction information. The historical profile tag determination module is used to input the key entity into the intent classification model to obtain the historical receipt profile tags of the historical recipient for the historical express parcel; the historical receipt profile includes at least one of historical receipt time and historical receipt method; the historical receipt profile represents the historical recipient's receipt habits. The tag storage module is used to store historical recipient profile tags corresponding to the historical recipients in a preset database; The parcel delivery module is used to respond to a target recipient's parcel delivery request, search the preset database for a target recipient profile tag that matches the target recipient, and deliver the target parcel to the target recipient based on the target recipient profile tag.

9. An electronic device, characterized in that, The device includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, the at least one instruction or at least one program being loaded and executed by the processor to implement the method for delivering express parcels as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores at least one instruction or at least one program, which is loaded and executed by a processor to implement the method for delivering express parcels as described in any one of claims 1-7.

Images