Jump to...
Enriching your product data with PowerImprove
Purpose: In this article, you will learn how to use PowerImprove to automatically enhance and enrich your product data. You will discover how PowerImprove’s AI-driven tools resolve data quality issues by completing missing product attributes, detecting inconsistencies, and extracting valuable information from various sources.
What is PowerImprove?
PowerImprove is part of the PowerSuite solution and serves as your all-in-one platform for data enrichment. This module offers powerful AI-driven tools that help you improve your data quality by identifying missing product properties, detecting anomalies, and automatically resolving other data quality issues.
With PowerImprove, you can transform incomplete or inconsistent product data into high-quality, structured information. The module uses advanced scraping techniques and feature extraction to collect valuable data from various sources, such as PDFs, images, product descriptions, and websites. Enriching your product data with PowerImprove not only improves data quality but also increases the operational efficiency of your entire organization.
Use PowerImprove when you want to:
- Add missing product attributes to your product catalog.
- Reduce manual data entry and avoid errors.
- Improve your data quality to drive better business results.
- Save time on cleaning and validating product data.
How to use PowerImprove
The steps below guide you through preparing your dataset, uploading files, reviewing AI suggestions, and exporting the enriched output. This helps you get the most out of your product data with less manual work.
Prepare your dataset
Start by preparing your data in Excel or CSV format. You can include different types of input:
- Product descriptions.
- Columns with features such as materials and colors.
- Links to PDFs.
- Links to images.
- Links to webpages.
Upload your files
- Go to the Uploads tab and upload your file using the blue upload button.
- Give the upload a recognizable name, specify the file type, and select the correct file.
- Select the unique product ID (such as GTIN or article number), define which column contains the product category, and set the language for your dataset.
- Optionally, add supplier and manufacturer information.
- Define the content of each column: Columns with text for extraction (product descriptions), Columns with PDF URLs for document content scraping, Columns with image URLs for feature extraction from visuals and Columns with image URLs for feature extraction from visuals.
- Click Upload to start the process. The system will begin processing your data. For large datasets, this may take some time, especially if many features need to be extracted
Make sure to always specify the product category so the algorithm knows which features to look for
Note: The feature ‘text to search for web scraping’ must be activated by our team. If you are interested, please contact us via the chatbot or by email.
For example, enter the EAN so we can complete missing attributes based on search results for that EAN.
Product review
At the top of each product page, you will see an overview of your source material and the taxonomy class. The taxonomy class tells the AI which product features to extract. Icons show where the information comes from.
Note: You can change the taxonomy class directly if a product is classified incorrectly. The correct features will then be extracted immediately.
- Review each suggestion and approve or reject it.
- Use the Approve all and Reject all buttons for faster processing.
- Settings: Enable the show suppressed features toggle to review additional extracted features.
- Use filters to work more efficiently based on your preferences.
- Optionally, add rules for specific terms the AI does not recognize. These rules help consistently identify abbreviations and technical terms.
Your feedback helps the algorithm learn your preferences and improve future suggestions. A rejected suggestion, for example, will not be shown again for similar products.
Make sure to always specify the product category so the algorithRead more about efficiently reviewing your product data in our guide ‘Using filters effectively in Product and Bulk review’.
Use Bulk Review
Switch to the Bulk Review tab for more efficient processing. Here you will see all features grouped (for example, all colors or lengths together) and can approve suggestions in bulk.
- Go to the Bulk Review tab.
- Set the desired filters.
- Approve or reject suggestions.
Run a Bulk Extraction (optional)
If you are satisfied with the quality of the suggestions and have reviewed entire groups, you can start a Bulk Extraction.
This process recalculates all data and learns from your feedback. As a result, the confidence scores of approved features will automatically increase in the future. At the same time, the rejected suggestions become suppressed features next time. This helps improve the quality of future results overall and makes upcoming review processes even more efficient.
- Go to the Uploads tab.
- Click on Restart extraction.
When starting a bulk extraction, the system will fetch or recalculate all data. This may take longer depending on the number of products.
Export enriched data
- Go to the Exports tab.
- Choose between Excel or CSV format.
- Select the export orientation: per product (PRODUCT) or per feature (FEATURE).
- Apply filters for uploads, taxonomy classes, approved results, and decide whether you want headers displayed in code or name.
- Choose whether to include existing features in the export.
