Jump to...
Using filters effectively in Product and Bulk review
Purpose: With the right filters in PowerImprove, you work much more focused and efficiently on your data enrichment. You’ll learn which filters are available and how to use them to solve data quality problems faster and optimize your workflow.
What are the Product and Bulk Review tabs?
The Product and Bulk Review tabs are among the most important components of PowerImprove and probably where you’ll spend most of your time working. These tabs provide an overview of all product attributes suggested by the AI, including the source and confidence scores for each suggestion.
In Product Review, you’ll find an individual page for each product you’ve uploaded. Each page shows the suggested product attributes, where the AI extracted this information from, and how confident the algorithm is that the suggestion is correct. At the top of each product page, you’ll see an overview of your source material and taxonomy class. The taxonomy class lets the AI know which product attributes it should look for.
The icons show you where information comes from:
- Black text icon: The data comes from a product description you provided
- Red PDF icon: The information was extracted from a PDF you uploaded
- Blue globe icon: The AI sourced this data from the internet through web scraping
- Black image icon: The algorithm derived this information from one of the images
On the Bulk Review page, you’ll find a collection of attributes from multiple articles. This gives you the advantage of working efficiently on different products so you can complete the review process faster and ultimately save a lot of time.
On both Product and Bulk Review, you’ll see important details for each suggestion:
- Confidence score: Shows how certain the AI is about the suggestion
- Source icon: Indicates what type of data the suggestion came from
- Feature name: Specifies which attribute from your data model the suggestion was made for
- Suggested value: The value the AI proposes for this attribute.
Use filters in Product and Bulk Review when you want to:
- Quickly find and edit specific products.
- Quickly find and edit specific products.
- Spend your time efficiently on products that need the most attention.
- Focus on specific sources or features you frequently approve.
The available filters explained
PowerImprove offers an extensive set of filters that allow you to precisely determine which products and features you want to review. By using these filters smartly, you work much more focused and save time when improving your data quality. The filters are available in both Product Review and Bulk Review, with each tab having its own specific filtering capabilities.
Filters in Product Review
Product Review offers various filters to work effectively on your data enrichment and improve data quality:
Product identifier helps you when looking for a specific product. You can search by article number, EAN, or any other unique number within your product assortment.
Uploads gives you the option to filter on specific uploads. This ensures your team works with the newest data in case of overwrites.
Taxonomy classes (also known as product groups) make it easier to work systematically. Using this filter allows you to focus on one specific or multiple groups that need your attention the most.
Suppliers is useful when you want to work per supplier. This filter provides the opportunity to work on one specific or multiple suppliers.
Manufacturers works the same as the supplier filter, but for manufacturers. This filter is particularly useful when you want to filter on private label products.
Confidence features gives you the opportunity to work on features with low, medium, or high risk levels. Using this filter ensures you spend your time wisely and systematically. You can choose to do the easy bulk work or focus on products that need your attention most. You can also set a specific confidence score range using the slider.
Sorting lets you choose between showing approved products last or sorting by taxonomy classes, which makes working per product group easier.
Toggle ‘Only products with extraction results’ can be useful because there may be products for which no suggestions exist. If, for example, no images, texts, PDFs, or web scraping was done, you can turn these off to see what percentage of products do have results. This actually makes reviewing somewhat easier and faster.
Settings in Product Review
In addition to filters, Product Review also has important settings that influence your workflow:
Realtime mode ensures that data such as the confidence score is updated directly after approving or rejecting features. This lets you literally see what you’re achieving with your adjustments.
Show suppressed features displays features that are normally hidden. A feature becomes suppressed when it was previously rejected or there were multiple alternatives for the found value. The feature with the highest confidence score is shown, while the lower one is suppressed.
Show data model displays the data model below the extracted features. This way you always have insight into what data was searched for based on the taxonomy class.
Filters in Bulk Review
For Bulk Review, there are some additional filters available that help you efficiently process large quantities of product attributes:
Source makes it possible to filter on specific sources. The source indicates how the information was extracted (text, PDF, image, or web scraping).
Features is useful when you’ve frequently approved a certain feature in product review. Filter on this feature in bulk review to approve it faster.
Improved values use this when you’ve chosen to overwrite existing predefined values.
Approved lets you filter on features that have already been approved or still need to be approved.
Settings in Bulk Review
In addition to filters, Bulk Review also has settings that can influence your workflow significantly:
Show suppressed features displays features that are normally hidden. A feature becomes suppressed when it was previously rejected or there were multiple alternatives for the found value. The feature with the highest confidence score is shown, while the lower one is suppressed.
Show only mandatory features shows only mandatory features from your data model
Note: By default all features are optional. Please contact the team if you would like to have mandatory features.
By using filters strategically and intentionally, you maintain clarity and avoid unnecessary effort. This keeps your review process manageable, even when handling a large number of clusters.
