Data pushing method based on artificial intelligence

Abstract

The invention discloses a data pushing method based on artificial intelligence, and relates to the technical field of data analysis, and the method comprises a data collection module, a feature extraction module, an analysis module, a data screening module and a data pushing module. Uploading the data into a feature extraction module for feature extraction; the feature extraction module extracts feature factors of the user from the acquired data; the analysis module analyzes the consumption preference of the user based on the extracted feature factors to obtain the consumption preference of the current user; the data screening module is used for screening massive data in the database and screening out commodities related to the consumption preference of the current user; according to the method, the user can find new products and brands, the shopping view is widened, and accurate recommendation can guide the user to find commodities which the user is interested in more quickly, so that the purchase possibility is increased, and the sales conversion rate is directly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a data push method based on artificial intelligence. Background Art

[0002] E-commerce big data refers to the massive data sets generated in the field of e-commerce, including user behavior data, transaction data, product information, and merchant data. These data are large in scale, diverse in type, and generated at a fast speed. They are of great significance to the development of e-commerce companies. With the rapid development of information technology, the amount of data has exploded. Traditional data push methods are often based on fixed rules and are difficult to accurately meet the personalized needs of users. At the same time, faced with massive amounts of data, how to efficiently screen out valuable information and push it to users has become an urgent problem to be solved. In this regard, we propose a data push method based on artificial intelligence. Summary of the invention

[0003] In order to solve the above technical problems, a data push method based on artificial intelligence is provided, and this technical solution solves the above problems.

[0004] To achieve the above purpose, the technical solution adopted by the present invention is: a data push method based on artificial intelligence, including: a data acquisition module, a feature extraction module, an analysis module, a data screening module and a data push module, and the push steps are:

[0005] The data collection module collects user consumption history data and uploads the data to the feature extraction module for feature extraction;

[0006] The feature extraction module extracts the user's characteristic factors from the acquired data;

[0007] The analysis module analyzes the user's consumption preferences based on the extracted characteristic factors to obtain the current user's consumption preferences;

[0008] The data screening module screens the massive amount of data in the database to select products related to the current user's consumption preferences;

[0009] The data push module intelligently pushes the filtered consumption preferences to the user client for the user to view.

[0010] Preferably, the data acquisition module obtains the user's historical purchase data from the e-commerce platform, and performs data verification and data cleaning on the obtained data. Data verification determines whether the data format is correct and the values ​​are reasonable. Data cleaning includes cleaning duplicate data, removing outliers and erroneous data, and after finding erroneous and abnormal data, verifying and deleting the data, and uploading the processed data to the feature extraction module.

[0011] Preferably, the feature extraction module extracts information based on the user's purchase history and browsing and search behaviors, statistically classifies the products in the user's purchase history records, determines the products that the user purchases frequently, calculates the frequency of the user's purchases on each product, and analyzes the user's purchase trends for products in different time periods; counts the user's browsing history on the e-commerce platform, including the category, brand and price of the product, calculates the user's browsing time for different products, and counts the keywords entered by the user to analyze the frequency and trend of user searches.

[0012] Preferably, the historical commodity set purchased by the user is S=(S1, S2, S3, ... S n ), where S i Represented as a type of commodity, let commodity S i The number of purchases is a times, and the frequently purchased products are determined based on the comparison of the purchase times of different products; for product S i The purchase frequency calculation formula is:

[0013]

[0014] Where F(S i ) is the user product S i The purchase frequency of product S i The number of purchases by the user, T is the number of purchases by the user; the browsing time of the user is based on the time of starting browsing minus the time of ending browsing, and the product S is obtained. i The browsing time of the products is taken into consideration, and the extracted browsing time and purchase frequency of the products are taken as the extracted features and uploaded to the analysis module.

[0015] Preferably, the analysis module constructs a comprehensive consumption preference index B of the user based on the product browsing time and the product purchase frequency, and the calculation formula is:

[0016] B = α * F (S i )+β*G(S i )

[0017] Where B is the consumption preference index of different commodities, α and β are weight coefficients, α+β=1, G(S i ) are different commodities and the browsing time of the current user, and the consumption preference index of the current user for different commodities is calculated based on the formula.

[0018] Preferably, a threshold of user preference is established based on a statistical distribution method, the mean C and standard deviation D of the consumption preference index are calculated, and the threshold is set to C±kD, where k is a coefficient determined based on historical data. C-kD and C+kD are used as the upper and lower limits of the threshold, and the calculated comprehensive consumption preference index B is compared with the upper and lower limits of the threshold. A value below the lower limit indicates that the user has a low degree of preference for the product, and a value above the upper limit indicates that the user has a high degree of preference for the product.

[0019] Preferably, the data screening module formulates specific screening conditions according to the user's consumption preference characteristics, uses the formulated screening conditions to query the product data in the database, and screens the database through SQL statements to screen out products with consumption preferences related to the current user.

[0020] Preferably, the screening conditions include the product categories that the current user explicitly prefers, products in the same category as products that the user has purchased in the past, and screening out other products in the product category that the user has browsed but not purchased, indicating that the user has potential interest in this type of product but did not purchase it for some reason. Recommendations are made by recording the user's browsing history and searching for similar products in the database.

[0021] Preferably, the data push module obtains the filtered commodity data that meets the user's consumption preferences from the data screening module, the data including the name, description, picture, price, brand and category details of the commodity, formats the obtained commodity data, and displays it on different clients. The data push module sends a push message to the user through the push notification service of the mobile application, and the push notification includes brief commodity information and prompts to guide the user to open the application to view more.

[0022] Preferably, the pushed products are sorted according to the user's consumption preference and the relevance of the products, with products that the user has recently browsed but not purchased being placed at the front, and new products in product categories with high purchase frequency being given higher priority, and the style and form of the pushed content are customized according to different push channels and user groups.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] Users of the present invention can receive product recommendations customized according to their unique consumption preferences and historical behaviors. The system can recommend products that users may not have actively searched for but are in line with their potential interests. This helps users discover new products and brands and broaden their shopping horizons. Accurate recommendations can guide users to find products they are interested in more quickly, thereby increasing the possibility of purchase, which directly improves sales conversion rates. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The present invention pushes the step flow chart. DETAILED DESCRIPTION

[0026] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0027] Reference Figure 1 As shown, a data push method based on artificial intelligence includes: a data acquisition module, a feature extraction module, an analysis module, a data screening module and a data push module, and the push steps are:

[0028] The data collection module collects user consumption history data and uploads the data to the feature extraction module for feature extraction;

[0029] The feature extraction module extracts the user's characteristic factors from the acquired data;

[0030] The analysis module analyzes the user's consumption preferences based on the extracted characteristic factors to obtain the current user's consumption preferences;

[0031] The data screening module screens the massive amount of data in the database to select products related to the current user's consumption preferences;

[0032] The data push module intelligently pushes the filtered consumption preferences to the user client for the user to view.

[0033] Users of this application can receive product recommendations customized according to their unique consumption preferences and historical behaviors, and are no longer overwhelmed by a large amount of irrelevant information, which makes the shopping process more efficient and enjoyable, saving users time and energy in screening products; through in-depth analysis and intelligent screening of user consumption history, the system can recommend some products that users may not have actively searched for but meet their potential interests, which helps users discover new products and brands and broaden their shopping horizons;

[0034] Accurate recommendations can guide users to find products of interest to them more quickly, thereby increasing the likelihood of purchase, which directly improves sales conversion rates and brings more revenue to the company; by providing personalized services, e-commerce companies can enhance user loyalty and stickiness to the platform.

[0035] The data collection module obtains users' historical purchase data based on the e-commerce platform, and performs data verification and data cleaning on the obtained data. Data verification determines whether the data format is correct and the values ​​are reasonable. Data cleaning includes cleaning duplicate data, removing outliers and erroneous data, and verifying and deleting the data after finding erroneous and abnormal data, and uploading the processed data to the feature extraction module.

[0036] Through data verification, this application can ensure that the collected data format is correct and the values ​​are reasonable, which enables subsequent analysis and processing to be based on accurate data and avoids erroneous conclusions due to incorrect data format or unreasonable values, thereby providing high-quality data for subsequent analysis modules, thereby improving the accuracy and effectiveness of the analysis, and obtaining more reliable results whether it is the analysis of user consumption preferences or other data analysis tasks.

[0037] The feature extraction module extracts information based on the user's purchase history, browsing and search behavior, statistically classifies the products in the user's purchase history, determines the products that the user purchases frequently, calculates the frequency of the user's purchases on each product, and analyzes the user's purchase trends for products in different time periods; counts the user's browsing history on the e-commerce platform, including the category, brand and price of the product, calculates the user's browsing time for different products, and counts the keywords entered by the user to analyze the frequency and trend of user searches.

[0038] The historical product set purchased by the user is S = (S1, S2, S3, ... S n ), where S i Represented as a type of commodity, let commodity S i The number of purchases is a times, and the frequently purchased products are determined based on the comparison of the purchase times of different products; for product S i The purchase frequency calculation formula is:

[0039]

[0040] Where F(S i ) is the user product S i The purchase frequency of product S i The number of purchases by the user, T is the number of purchases by the user; the browsing time of the user is based on the time of starting browsing minus the time of ending browsing, and the product S is obtained. i The browsing time of the products is taken into consideration, and the extracted browsing time and purchase frequency of the products are taken as the extracted features and uploaded to the analysis module.

[0041] The analysis module constructs the user's comprehensive consumption preference index B based on the product browsing time and product purchase frequency. The calculation formula is:

[0042] B = α * F (S i )+β*G(S i )

[0043] Where B is the consumption preference index of different commodities, α and β are weight coefficients, α+β=1, G(S i) are different commodities and the browsing time of the current user, and the consumption preference index of the current user for different commodities is calculated based on the formula.

[0044] Through the analysis of purchase history, this application enables users to obtain recommended products on the platform that are more in line with their actual purchasing habits. Analyzing purchase trends and browsing behaviors in different time periods may reveal changes in interests that the users themselves are not aware of. Features extracted based on users' purchase history and browsing behaviors can provide users with more personalized product recommendations. Understanding users' search frequency and trends can help the platform optimize search functions and product classifications.

[0045] Based on the statistical distribution method, we establish the threshold of user preference, calculate the mean C and standard deviation D of the consumption preference index, set the threshold to C±kD, where k is the coefficient determined based on historical data, and use C-kD and C+kD as the upper and lower limits of the threshold. We compare the calculated comprehensive consumption preference index B with the upper and lower limits of the threshold. Values ​​below the lower limit indicate that the user has a low degree of preference for the product, and values ​​above the upper limit indicate that the user has a high degree of preference for the product.

[0046] The methods for calculating the mean C and the standard deviation D in this application are all existing technologies and will not be elaborated on here. Users can understand their preference for different products through a clear threshold range. The e-commerce platform can provide users with personalized recommendations that are more in line with their preference based on accurate threshold judgment. Users are no longer troubled by a large number of irrelevant or low-preference product recommendations, which improves the satisfaction of the shopping experience. If the user's preference for a certain type of product has always been high, the platform can recommend more related new products or popular styles of this type of product. By analyzing the user's preference thresholds for different products, the platform can better adjust the inventory strategy. For products with low user preference, the inventory can be reduced to avoid backlogs; for products with high user preference, the inventory can be appropriately increased to ensure supply.

[0047] The data screening module formulates specific screening conditions according to the user's consumption preference characteristics, uses the formulated screening conditions to query the product data in the database, and filters in the database through SQL statements to filter out the products with consumption preferences related to the current user.

[0048] Users of this application can quickly obtain product information that is highly matched with their consumption preferences, saving time and energy in searching through massive amounts of products.

[0049] The screening conditions include the product categories that the current user explicitly prefers, products in the same category as products that the user has purchased in the past, and filtering out other products in the product category that the user has browsed but not purchased, indicating that the user has potential interest in this type of product but did not purchase it for some reason. Recommendations are made by recording the user's browsing history and searching for similar products in the database.

[0050] This application can quickly find all the products that match the product categories explicitly specified by the user from the database. For example, when the user explicitly expresses interest in "electronic products", this SQL statement can accurately extract all the products belonging to the "electronic products" category from the huge product database, avoid displaying irrelevant product categories to the user, improve the efficiency of users in finding target products, and provide users with a personalized shopping experience. By meeting the user's demand for specific product categories, the user can feel the platform's understanding and attention to their preferences, thereby enhancing the user's satisfaction and loyalty to the platform. Based on the user's purchase history, the user's potential demand in the same product category can be inferred.

[0051] The data push module obtains the product data that meets the user's consumption preferences from the data screening module. The data includes the product name, description, picture, price, brand and category details. The obtained product data is formatted and displayed on different clients. The data push module sends push messages to users through the push notification service of the mobile application. The push notification includes brief product information and prompts to guide users to open the application to view more.

[0052] The pushed products are sorted according to the user's consumption preference and the relevance of the products. The products that the user has recently browsed but not purchased are placed at the front. New products in product categories with high purchase frequency are given higher priority. The style and form of the pushed content are customized according to different push channels and user groups.

[0053] This application puts the products that the user has recently browsed but not purchased in the front, which can remind the user of the existence of the product again. The user's previous browsing indicates that he has a certain interest in it. Pushing it first can increase the possibility of the user considering buying it again, thereby improving the purchase conversion rate. For example, when you browsed a new mobile phone online but did not buy it immediately for some reason, when you see the push of this product again, it may prompt you to further think about its advantages and disadvantages and finally decide to buy it.

[0054] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention, and these changes and improvements fall within the scope of the present invention claimed.

Claims

1. A data push method based on artificial intelligence, characterized in that: include: Data collection module, feature extraction module, analysis module, data screening module and data push module, the push steps are: The data collection module collects user consumption history data and uploads the data to the feature extraction module for feature extraction; The feature extraction module extracts the user's characteristic factors from the acquired data; The analysis module analyzes the user's consumption preferences based on the extracted characteristic factors to obtain the current user's consumption preferences; The data screening module screens the massive amount of data in the database to select products related to the current user's consumption preferences; The data push module intelligently pushes the filtered consumption preferences to the user client for the user to view.

2. The data push method based on artificial intelligence according to claim 1, characterized in that: The data collection module obtains users' historical purchase data from the e-commerce platform, and performs data verification and data cleaning on the obtained data. Data verification determines whether the data format is correct and the values ​​are reasonable. Data cleaning includes cleaning duplicate data, removing outliers and erroneous data, and verifying and deleting the data after finding erroneous and abnormal data, and uploading the processed data to the feature extraction module.

3. The data push method based on artificial intelligence according to claim 1, characterized in that: The feature extraction module analyzes the user's purchase history, browsing and search behavior, statistically classifies the products in the user's purchase history, determines the products that the user purchases frequently, calculates the frequency of the user's purchase on each product, and analyzes the user's purchase trend for products in different time periods; Statistics are collected on users’ browsing history on e-commerce platforms, including product categories, brands and prices. The browsing time of users on different products is calculated. Statistics are collected on keywords entered by users to analyze the frequency and trends of user searches.

4. The data push method based on artificial intelligence according to claim 3 is characterized in that: The historical product set purchased by the user is S = (S1, S2, S3, ... S n ), where S i Represented as a type of commodity, let commodity S i The number of purchases is a times, and the frequently purchased products are determined based on the comparison of the purchase times of different products; for product S i The purchase frequency calculation formula is: Where F(S i ) is the user product S i The purchase frequency of product S i The number of purchases by the user, T is the number of purchases by the user; the browsing time of the user is based on the time of starting browsing minus the time of ending browsing, and the product S is obtained. i The browsing time of the products is taken into consideration, and the extracted browsing time and purchase frequency of the products are taken as the extracted features and uploaded to the analysis module.

5. The data push method based on artificial intelligence according to claim 1, characterized in that: The analysis module constructs the user's comprehensive consumption preference index B based on the product browsing time and product purchase frequency. The calculation formula is: B=α*F(S i )+β*G(S i ) Where B is the consumption preference index of different commodities, α and β are weight coefficients, α+β=1, G(S i ) are different commodities and the browsing time of the current user, and the consumption preference index of the current user for different commodities is calculated based on the formula.

6. The data push method based on artificial intelligence according to claim 5, characterized in that: Based on the statistical distribution method, we establish the threshold of user preference, calculate the mean C and standard deviation D of the consumption preference index, set the threshold to C±kD, where k is the coefficient determined based on historical data, and use C-kD and C+kD as the upper and lower limits of the threshold. We compare the calculated comprehensive consumption preference index B with the upper and lower limits of the threshold. Values ​​below the lower limit indicate that the user has a low degree of preference for the product, and values ​​above the upper limit indicate that the user has a high degree of preference for the product.

7. The data push method based on artificial intelligence according to claim 1, characterized in that: The data screening module formulates specific screening conditions according to the user's consumption preference characteristics, uses the formulated screening conditions to query the product data in the database, and filters in the database through SQL statements to filter out the products with consumption preferences related to the current user.

8. The artificial intelligence-based data push method according to claim 7, characterized in that: The screening conditions include the product categories that the current user explicitly prefers, products in the same category as products that the user has purchased in the past, and filtering out other products in the product category that the user has browsed but not purchased, indicating that the user has potential interest in this type of product but did not purchase it for some reason. Recommendations are made by recording the user's browsing history and searching for similar products in the database.

9. The data push method based on artificial intelligence according to claim 1, characterized in that: The data push module obtains the product data that meets the user's consumption preferences from the data screening module. The data includes the product name, description, picture, price, brand and category details. The obtained product data is formatted and displayed on different clients. The data push module sends push messages to users through the push notification service of the mobile application. The push notification includes brief product information and prompts to guide users to open the application to view more.

10. The data push method based on artificial intelligence according to claim 1, characterized in that: The pushed products are sorted according to the user's consumption preference and the relevance of the products. The products that the user has recently browsed but not purchased are placed at the front. New products in product categories with high purchase frequency are given higher priority. The style and form of the pushed content are customized according to different push channels and user groups.

Images