Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
PowerPoint.docx
Скачиваний:
0
Добавлен:
01.07.2025
Размер:
18.33 Mб
Скачать

Publish

A copy activity in the EgressSimilarProductsSqlPipeline and EgressRecommendedProductsSqlPipeline is used to copy the item similarity matrix (product-to-product recommendations) and the personalized recommendation matrix (user-to-product recommendations) to a Microsoft Azure SQL database.

The database server and the productrec database were generated as part of the setup script. (See demo/productrec-accounts.txt file for SQL Server logon details.) After the pipeline is executed successfully, it takes the item similarity matrix and personalized recommendation matrix data in an Azure Blob and copies it to an Azure SQL Server.

The item similarity matrix is copied to the SimilarProducts table in the productrec database.

 

The product recommendation matrix is copied to the PersonalizedRecommendations table in the producrec database.

Consume

The productrec database generated previously is used to power the retailer’s website for displaying product-to-product recommendations and user-to-product recommendations.

As part of deploying setup.ps1, an empty Azure website is created. After you publish the demo retailer website, the product catalog will appear as the default landing page on the website.

 

You can browse through the different artists available in the product catalog. Click an artist’s name to showcase similar artists (product-to-product recommendations) based on community interests and patterns. These product-to-product recommendations (item similarity matrix) were generated as part of the ProductsSimilarityMahoutPipeline in the Analyze step and moved to the Azure SQL database as part of the EgressSimilarProductsSqlPipeline. For example, you can click 1 Giant Leap to view all other similar artists.

 

You can also view personalized product recommendations based on a person’s interests, actions, and past usage patterns. These personalized recommendations were generated as part of the ProductsRecommenderMahoutPipeline in the Analyze step and moved to the Azure SQL database as part of the EgressRecommendedProductsSqlPipeline. For example, you can log on as Gaurav Malhotra to see the top picks (artists) for shopper Gaurav Malhotra.

 

These personalized recommendations allow the online retailer to develop a more engaging customer experience designed to increase sales revenue.

Next steps

This demonstration outlines a personalized product recommendation scenario in which various Azure capabilities (HDInsight, Azure Data Factory, and Mahout) have been used to build an end-to-end personalized product recommendation solution. A demo website has been created to visualize the personalized product recommendations. Refer to the Solution architecture section to see the roles that each service plays in the solution.

The same data-driven workflow (data integration/processing  batch recommendations using Mahout  moving the results into an Azure SQL database  visualize using a demo website) can be applied to various scenarios. Reach out to the Use Case V-team (datausecase@microsoft.com), and we can work with you to shape the demo based on your specific needs.

Соседние файлы в предмете [НЕСОРТИРОВАННОЕ]