The query builder quickly allows you to generate a query for three common tasks that you will want to use to examine your data and reduces the need for extensive SQL skills.
- Similarity search - EXISTING (within a dataset), find items similar to a specific example item Example use: You have ingested & processed images for a Fashion catalog. Item #13 in your catalog is a picture of a red shirt and you want to find all other images similar to it.
- Similarity search - ADHOC (outside selected dataset), find items similar to an ad-hoc query item that is not in your data set
- Example use: You have ingested & processed images for a Fashion catalog. A customer gives you a URL link to a picture of a shirt they already own. You want to use it to search for similar items in your existing catalog.
- Example use: You have ingested & processed restaurant reviews and want to search for ones semantically similar to the ad-hoc phrase "bars with healthy food".
- View Predicted Enrichment output - easily display the labels that an enrichment created for your dataset, with the option to filter on specific values
- Example use: You have ingested & processed images for a Fashion catalog. You have also created an Enrichment that labels your images as ‘shirt’, ‘pants’, or ‘dress’. Now you want to easily retrieve the IDs of the images with shirts in them.
- Low Confidence Predictions - find predictions which Graft is unsure about. So you can improve your model
Query Builder can be found in the toolbar on the right hand side
Click on Build to start query builder
Start a new Query
We recommend that you create a new empty query ahead of starting query builder by Clicking on the + in the left hand side of the toolbar
Select an Entity
From the pop up window Select the Entity you wish to design a query for, once selected the builder options will become available
Select a template
Now that we have an entity (in this example Fashion_images) Select one of the available options