Solutions / Find similar items
Find the closest match.
When the answer isn't one of a fixed set of categories but "which of these existing things is this most like."
The decision
Given a query, which items in your reference data are most similar to it?
Text or images, depending on what you populate the reference data with.
Why it's worth automating
What makes this hard by hand.
Keyword search misses meaning
Two descriptions of the same thing can share almost no words. Similarity on meaning finds what string matching won't.
There's no fixed label to predict
Some questions don't have categories. They have a ranked list of candidates and a relevance ordering.
Your corpus is the value
The useful comparison set is the one you already own — your tickets, your catalog, your incident history.
How it works
Three steps.
-
1
Upload your reference data
The collection you want to search against. No labels required — search functions don't need them.
-
2
Query it
Send an item, get back the nearest matches with similarity scores.
-
3
Use it for retrieval, dedup, or recommendation
Matching a new ticket to a solved one, spotting near-duplicates, or surfacing related items.
Keep reading