Disclosure of Invention
The invention provides an international logistics route recommending method, which aims to solve the technical problem of how to recommend a proper logistics route to a user in cross-border e-commerce shopping.
The technical problems of the invention are solved by the following technical scheme:
An international logistics line recommendation method comprises the following steps:
S1: collecting package order information submitted by a user;
S2: screening the package order information in the step S1 to form user behavior data and line characteristic data;
S3: analyzing the user behavior data in the step S2 to form a line selection data portrait; analyzing and excavating the line characteristic data in the step S2 to form historical line big data;
s4: and recommending a logistics line for the user according to the line selection data portrait and the historical line big data in the step S3.
Preferably, in step S3, the route selection data representation is formed by analyzing the user location, user consumption habit, user preference selection of inexpensive or expensive route, and user preference selection of route with long or short delivery time in the user behavior data.
Preferably, in step S3, the historical route big data is formed by analyzing and mining mailing time length, price, clearance time length, mailing limit, mailing country, region feature in the route characteristic data.
Preferably, in step S3, the line characteristic data and the user behavior data are stored in a line characteristic database and a user behavior database, respectively.
Preferably, in step S4, the method further includes configuring a priority of the logistics line, and adjusting a line recommendation sequence by adjusting the priority.
Preferably, in step S4, a collaborative filtering recommendation algorithm is used to recommend a logistic line to the user in combination with the information of the items to be sent.
Preferably, the user behavior data in step S2 is obtained by performing a cluster analysis on the user by using a bayesian classification algorithm.
Further, after cluster analysis is performed on the users, a user behavior model is built for various users.
Preferably, in step S2, the line characteristic data adopts a K-means clustering algorithm to perform cluster analysis on the logistics lines, and a line behavior model is built for each line.
Further, after the line behavior model is established, a machine learning method is adopted to correct and perfect the data of the logistics line.
The invention also provides a device adopting the international logistics line recommendation method, which comprises a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program in the memory to control the device adopting the international logistics line recommendation method to execute any one of the above methods.
Furthermore, the present invention also proposes a computer readable storage medium having stored thereon a computer program, said medium being used in combination with a computer, which program when executed by a processor realizes the steps of any of the methods described above.
Compared with the prior art, the invention has the beneficial effects that: and screening the acquired package order information to obtain line characteristic data of a user behavior data set, analyzing the user behavior data to form a line selection data portrait, analyzing and mining the line characteristic data to form historical line big data, and recommending a proper international logistics line for a user according to the line selection data image and the historical line big data.
Detailed Description
The invention will be further described with reference to the following drawings in conjunction with the preferred embodiments.
An international logistics recommendation system, as shown in fig. 1, comprising:
A package order module: and processing the package order data submitted by the user, and synchronizing the package order data and the user information to the data acquisition module.
And a data preprocessing module: and the historical order data synchronized by the package order processing module is screened to form historical line big data and user behavior big data, and the historical line big data and the user behavior big data are respectively stored in a line characteristic database and a user behavior database.
User data analysis module: analyzing and mining the user history selection line data, analyzing the user history selection line data, and forming a user line selection data image. The user line selection data image is classified into a time-lapse type, an economical type, a remote type, and the like according to the user location, the user consumption habit, the user preference for selecting a line which is inexpensive or expensive, the user preference for selecting a line which is long in the conveyance time or short in the conveyance time, and the like.
Line data analysis module: analyzing and mining the data generated by the package preprocessing module, and establishing historical line big data, including line transportation time length, line closing capability, shipping article limit, line coverage, package shelf life and the like.
Line intelligent recommendation module: according to the line classification and characteristics analyzed by the line data analysis module, selecting a most suitable candidate line candidate list for a user by combining the user data image, and recommending the logistics line for the user by combining the priority of the added line for logistics line management.
The logistics line management module: and (3) managing the international logistics line, configuring basic data such as the cost, the transportation path, the sending timeliness and the like of the international logistics line, and configuring the priority of the logistics line, wherein the line recommendation result is influenced by adjusting the priority of the line.
The data analysis flow, as shown in FIG. 2, is as follows:
The data acquisition module collects historical package data from the package order module, performs primary data processing, eliminates useless fields and forms input data taking packages as dimensions.
The data preprocessing module preprocesses the received data taking the package as the dimension, decomposes the data taking the user as the dimension and the data taking the line as the dimension, and stores the data and the data into a corresponding distributed file system (such as HDFS) respectively.
The user data analysis module performs data analysis processing on the user data, calculates the behavior characteristics of the historical shipping package selection lines of each user, and forms a user behavior model, namely a line selection data portrait.
The line data analysis module performs data analysis processing on the line data, calculates the characteristics of each logistics line according to the historical data, performs cluster analysis on each logistics line, and establishes line model data.
The line recommendation flow is shown in fig. 3, and the detailed flow is as follows:
when the user submits the package sending request, the user information, the basic information (category, weight, volume) of the articles contained in the package, the sending destination and other basic information are submitted to the intelligent line recommending module.
And the line intelligent recommendation module screens out a line list which is most suitable for the user according to the line model data and the user historical behavior model data and the information of the sent article, and returns the line list to the user side for the user to select.
The line intelligent recommending module can be added with priority of the merchant in line configuration as a reference when selecting the line, and the line configured by the merchant is preferentially recommended when a plurality of lines are calculated, and the merchant effectively improves the profit of the merchant by preferentially recommending the logistics line with higher profit rate.
The information of the object to be sent includes: weight, volume, country of delivery, etc. The algorithm also considers the factors added through background configuration in the execution process, such as recommending lines with higher profit margin preferentially for the same type of logistics lines, and improves the profit of the company while improving the convenience of users.
Based on the above-mentioned international logistics line recommendation system, an international logistics line recommendation method is provided, as shown in fig. 4, comprising the following steps:
s1: collecting package order information submitted by a user; the package order information comprises package basic data such as package weight, package volume, mail item class, mail expense, mail country, region, mail line, mail time length, clearance time length and the like;
S2: screening the package order information in the step S1, screening out required fields, and removing fields irrelevant to line analysis to form user behavior data and line characteristic data; the user behavior data comprise basic data such as age, gender, nationality, residence, payment mode and the like; the line characteristic data comprise line basic information such as mailing duration, price, clearance duration, mailing limit, mailing country, region and the like;
S3: analyzing the user behavior data in the step S2 to form a line selection data image, wherein the user behavior data comprises a user place, a user consumption habit, a user preference selection of a cheap or expensive line and a user preference selection of a line with long or short transportation time; through data analysis, the users are found to like to select the lines with long sending period and low price, and the users can be classified as economic users; analyzing and excavating the line characteristic data in the step S2 to form historical line big data; the historical line big data comprise line transportation time, line closing capability, shipping article limit, line coverage range and the like; and in step S3, the line characteristic data and the user behavior data are respectively stored in a line characteristic database and a user behavior database. The line characteristics database and the user behavior database are provided with corresponding distributed file systems (HDFS); the user behavior data adopts a Bayesian classification algorithm to carry out cluster analysis on users, the users are divided into different types and marked with tag features, a line behavior model is established for various lines, namely, a line selection data image is formed, and the method comprises the following steps: time-efficient (time-critical), economical (preference for selecting inexpensive lines), remote (location remote users), etc., each user corresponding to at least one type according to its behavioral characteristics. The line characteristic data adopts a K-means clustering algorithm to perform clustering analysis on the logistics line, the logistics line is divided into different line types, and a line behavior model is built, and the method comprises the following steps: time-efficient (short shipping period), economical (price benefit), remote (coverage area shopping), packet (e.g. shipping only < = 2KG line), large (e.g. shipping only > = 5KG line), etc., each logistics line corresponds to at least one line type. The line selection data image is formed by analyzing the user location, the user consumption habit, the user bias selection of cheap or expensive lines and the user bias selection of long or short transportation time in the user behavior data; the historical line big data is formed by analyzing and mining mailing time length, price, clearance time length, mailing limit, mailing country and region characteristics in the line characteristic data.
It should be noted that the historical route big data includes route characteristic data formed by all packages, and one route selection data image corresponds to one user.
S4: and recommending a logistics line for the user according to the line selection data image and the historical line big data in the step S3 and the article information to be sent. In step S4, the method further includes configuring the priority of the logistics lines, where the configuration of the priority is that the priority is configured according to the weight of each logistics line, the greater the weight is, the higher the priority is, and by adjusting the priority to adjust the recommendation sequence of the lines, the priority control is an aid, and when a plurality of lines are all suitable for a certain user, the lines with high priority are recommended preferentially. In addition, the method also comprises basic data such as the cost, the transportation path, the sending time and the like for configuring the logistics line; in step S4, a collaborative filtering recommendation algorithm is adopted to recommend a logistics route for the user in combination with the item information to be sent. The logistic recommendation line can further comprise a dispatch point for recommending the package to the user, and when the dispatcher does not directly send the package to the addressee, the addressee can conveniently go to the dispatch point to get.
In this embodiment, after the line behavior model is built, a machine learning method is used to correct and perfect the data of the logistics line. Because the clustering algorithm has certain errors, the accuracy is more and more accurate along with the increase of the data volume of the historical line big data, the newly submitted package data of the user is used as input to update the user behavior data and the line characteristic data, and the user classification is re-divided, so that a positive feedback is achieved, and more accurate recommendation is realized.
The embodiment comprises a device adopting the international logistics line recommendation method, which comprises a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program in the memory to control the device adopting the international logistics line recommendation method to execute the method.
The present embodiment also includes a computer-readable storage medium having stored thereon a computer program for use in conjunction with a computer, which when executed by a processor, performs the steps of the method described above.
The international logistics route recommendation method of the present invention is exemplified below, but is not limited to the following specific examples. For the user who submits the package for the first time, a route selection data portrait is not established yet, the international logistics route recommendation is based on historical route big data, and the following is exemplified by a new user A:
201. the user A submits a meat package to the United kingdom, because the user A is a new user and has no line selection data portrait, the user can only recommend logistics lines for the user according to a line behavior model and a collaborative filtering recommendation algorithm, EMS and EUB (easy mail treasure) are recommended for the user, finally the user selects the EUB, and package basic data such as package weight, package volume, mail item category, mail expense, mail country, region, mail line, mail duration, clearance duration and the like are obtained by collecting package order information submitted by the user A.
202. Screening the package order information, screening out required fields, and removing fields irrelevant to line analysis to form user behavior data and line characteristic data; the user A behavior data comprise 48 years old, female sex, china international, land occupied, shanghai, paying mode, net bank, meat as mail items, 2KG of parcel weight and volume of parcel of 100cm 3; the line characteristic data includes mailing time of 20 days, price of 50 yuan, clear time of 1 day, mailing country of uk, region of scotland, and logistics of EUB.
203. Analyzing the behavior data of the user A to form a line selection data image, and obtaining a route which is long in sending period and low in price and liked by the user A through data analysis, so that the user A is classified as an economic user; analyzing and mining the line characteristic data to form historical line big data, and carrying out clustering analysis on the logistics line by adopting a K-means clustering algorithm to divide the logistics line into economic and small package types. It should be noted that the historical route big data includes route characteristic data which is formed by all packages before.
Steps 201-203 are to reestablish the route selection data portraits and route behavior models of user a based on the parcel characteristics of user a. Next, step 204, a new clothing package recommended international logistics line is submitted for user a:
204. According to the line selection data image and the line behavior model in the step 203, the collaborative filtering recommendation algorithm is adopted to recommend the logistics line for the user in combination with the article information of the clothes to be sent, and according to the data calculation analysis user characteristics and the type of the sent article, the logistics line such as AIR special line with lower price than Yi Bao is pushed for the user, and the logistics line recommendation is completed. The accuracy of recommending the international logistics routes according to the route selection data portraits and the route behavior models can be improved along with the continuous increase and updating of the data, when a user A submits a package, big data can be updated, and the corresponding route behavior models and route selection data portraits can be updated, so that the next more accurate recommendation is realized.
The foregoing is a further detailed description of the invention in connection with the preferred embodiments, and it is not intended that the invention be limited to the specific embodiments described. It will be apparent to those skilled in the art that several equivalent substitutions and obvious modifications can be made without departing from the spirit of the invention, and the same should be considered to be within the scope of the invention.