CN117094789A - Cross-platform intelligent commodity loading method based on cross-border electronic commerce - Google Patents

Abstract

The invention discloses a cross-platform intelligent commodity loading method based on a cross-border electronic commerce. The invention relates to the field of software, big data and artificial intelligence, and is mainly applied to cross-border electronic commerce. The invention solves the problems of high labor cost, low precision, complexity and low efficiency of commodity shelf-up on a cross-border e-commerce platform. The technical scheme of the invention mainly comprises the steps of automatically analyzing the data rule of each platform, automatically filling corresponding data for different classifications, automatically analyzing sensitive words and key words, automatically selecting proper pictures, automatically putting on the shelf, tracking the putting on the shelf result and the like. The invention has the main purposes of quickly assembling various materials such as commodity pictures, attributes, keywords and the like, accurately distinguishing categories, accurately avoiding violations, being not influenced by factors such as manpower limit, time difference and the like, and having no difference release on a supported platform; the continuous commodity is put on shelf, no rest exists throughout the year, and the store liveness and commodity advantages are fully improved.

Description

Cross-platform intelligent commodity loading method based on cross-border electronic commerce

Technical Field

The method relates to the field of software, big data and artificial intelligence, and is mainly applied to cross-border electronic commerce.

Background

Selling goods on an e-commerce platform, and firstly, putting the goods on the e-commerce platform. E-commerce platforms generally provide two modes of putting on shelf, namely, inputting through a visual interface and importing through a classification template. The cross-border e-commerce company uses the loading mode provided by the platform to load the platform, or loads the platform through third-party software, the third-party software analyzes data fields required by different classifications of the platforms in advance by manpower, then the user inputs data, and the software synchronizes the data input by the user to the e-commerce platform. The main stream of the on-shelf software needs to analyze the existing classification rules of the platform in advance, and when the classification rules are newly added to the platform or changed, the platform needs to analyze again manually; the commodity information is required to be manually input, the data formats of the platforms are different, the data formats of the same platform with different classifications are also huge in difference, the requirement on business knowledge of people is high, the information accuracy is difficult to ensure during manual input, a large amount of time is required to be consumed, and the efficiency is extremely low; the information such as key words, sensitive words, classification and the like all require extremely high business knowledge, and the current software cannot be intelligently processed. Therefore, an intelligent system is needed to automatically analyze the data rules of each platform, accurately divide commodities into different categories of different platforms, automatically fill corresponding data for different categories, automatically analyze sensitive words and keywords, automatically select proper pictures, and finally automatically put on the shelf and track the put-on result.

Disclosure of Invention

The invention solves the problems of high labor cost, low precision, complexity and low efficiency of commodity shelf-up on a cross-border e-commerce platform. The technical scheme of the invention mainly comprises the steps of automatically analyzing the data rule of each platform, automatically filling corresponding data for different classifications, automatically analyzing sensitive words and key words, automatically selecting proper pictures, automatically putting on the shelf, tracking the putting on the shelf result and the like. The method is mainly used for rapidly assembling various materials such as commodity pictures, attributes, keywords and the like, accurately distinguishing categories, accurately avoiding violations, being not influenced by factors such as manpower limit, time difference and the like, and releasing the supported platforms indiscriminately; the continuous commodity is put on shelf, no rest exists throughout the year, and the store liveness and commodity advantages are fully improved.

Drawings

Fig. 1 is a flowchart of a method for implementing a cross-platform intelligent on-shelf commodity based on a cross-border e-commerce, and the figure simply and directly shows an implementation flow of a method for implementing a cross-platform intelligent on-shelf commodity based on a cross-border e-commerce.

In the figure:

letters: representing step number

Block: representative flow steps

Arrow: the representative flow proceeds to diamond: representing judgment content, the judgment is performed on the flow content in the flow chart, and is used for judging one of different operations to the decision.

Detailed Description

The method comprises the following steps:

a1, the original commodity content comprises titles, classifications, keywords, pictures and the like. In cross-border electronic commerce commodities, the characters are English, and the title is not more than 80 characters; the classification has several levels, and is a tree structure; the keywords can describe commodity attributes more accurately in general; the number of required pictures is different except for the size, such as the picture is divided into a main picture, a thumbnail picture, a detail picture and the like. These are some of the core elements of the release commodity and are available for use in the following steps.

A2, acquiring a platform classification list from cross-border e-commerce platforms such as Amazon, shAN_SNing, abiba International station, eBay and the like through a distributed web crawler technology, combining the platform classification list into a classification tree structure by using a breadth-first search algorithm, and autonomously analyzing the classification tree structure to obtain the fineness of the platform classification. The foundation is laid for the following steps.

B1, searching the category of matching classification by using the existing large text full text retrieval algorithm according to the data obtained in the A2 and the platform to be released, and then obtaining a numerical value describing the matching degree by using the weight ordering algorithm and the C4.5 analysis algorithm, and obtaining ladder data according to the numerical value through bubbling ordering. If the data is from D, the determined error classification needs to be eliminated, and then the classification with the highest matching degree is used.

And B2, knowing the matched classification according to the B1, and downloading the imported templates under the classification to the local through a web crawler technology or an API interface opened by a platform. The Apache POI based on the DOM mode is used for analyzing, and the file is directly loaded into the memory, so that the speed is high. Analyzing and distinguishing necessary filling items, unnecessary filling items and selection items, and loading all options of the selection items into the template to prepare for filling the template.

C1, the steps of assembling commodity attributes are as follows:

(1) And (3) extracting or reorganizing the title and the selling point in the A1 with complete semantics according to a rule method, a decision tree model, a hidden Markov model, a logistic regression and the like. Without keywords, the corresponding words may be extracted from the title, attributes, and selling points of the variant to act as. According to the big data obtained in A2, the points of each character class can be mutually complemented. If the picture type is wrong, the text matching is skipped.

(2) And obtaining a integrable function of the image according to the Laplace operator calculated in the Laplace algorithm, obtaining an index for representing the intensity of gray level change in the image according to Fourier transformation, namely obtaining a transformation coefficient matrix, and taking the variance of the matrix as the judgment of image blurring. Leaving a clear picture. If E1 step is transferred, the picture is switched. If the text error is abnormal, skipping the picture selection.

And C2, automatically filling commodity attributes, titles, keywords and the like matched with the templates in the classified templates through B2 analysis. And the electronic commerce platform is automatically docked, and filled forms which need to be uploaded are transmitted to the platform system without manual operation.

D. And acquiring auditing results issued by the goods on shelf at fixed time through an API provided by the network robot or the electronic commerce platform. And if the verification is not passed, extracting the failure reason. And on the contrary, the release on the shelf is successful.

And E1, judging according to the reasons given by the platform when the auditing result is not passed and the retry times are not up to the upper limit, and classifying the auditing result into text errors, classification errors and picture errors. And C1, continuing to perform C1 on the errors of the characters and the pictures, and continuing to perform B1 on the abnormal classification.

And E2, notifying related operators when the auditing fails and the retry times reach the upper limit, notifying reasons and taking medicine for the symptoms.

And E3, successfully verifying to obtain successful release of the commodity on the E-commerce platform.

Claims (6)

1. The method for cross-platform intelligent shelving of commodities based on cross-border E-commerce is characterized in that existing classification data and rule information in cross-border E-commerce platforms automatically analyze classification data rules of each platform, and an intelligent classification algorithm is used for automatically filling corresponding data for classification of different platforms so as to accurately divide the commodities into the classification of different platforms.

2. A cross-platform intelligent goods loading method based on cross-border electronic commerce is characterized in that intelligent error correction is achieved, the problem of error loading and reporting is solved, error information is actively collected, and loading rules are perfected.

3. A cross-platform intelligent commodity loading method based on cross-border electronic commerce is characterized in that existing sensitive word and violation analysis tools are used, and the sensitive word and key word are automatically analyzed by adding self-analysis cross-border electronic commerce platform rules.

4. A cross-platform intelligent commodity loading method based on a cross-border e-commerce is characterized in that matrix variances of pictures are obtained according to the existing Laplace algorithm and Fourier transformation, and clear pictures are selected from the values of the defined variances above 12.

5. A method for intelligent cross-platform shelving commodity based on cross-border E-commerce is characterized in that the commodity is automatically shelf-mounted to a platform, the cross-border E-commerce platform is docked, and a form which is filled in by a system and needs to be uploaded is transferred to the platform system to finish the shelf-mounting operation.

6. A method for intelligent putting on shelf commodity based on cross-platform of cross-border E-commerce is characterized in that the putting on shelf result is tracked through an API provided by a network robot or an E-commerce platform.