Migration from Weaviate to Qdrant: Distance Value Change¶
After the release of Cognigy.AI 4.74.0, all Cognigy.AI installations have been updated to use Qdrant as the vector database for Knowledge AI instead of Weaviate. This migration changes the Knowledge AI calculates the distance value in the topK object of the Search Extract Output Node.
Weaviate and Qdrant are tools that calculate similarity in different ways:
- Weaviate calculates similarity values within the range of 0 to 2, where 0 represents maximum similarity and 2 represents minimum similarity.
- Qdrant calculates similarity values within the range of -1 to 1, where 1 represents maximum similarity and -1 represents minimum similarity.
These differences lead to a new interpretation of the distance value. If you use distance to make decisions in the Flow, consider the new similarity range of -1 to 1.
Warning
The LLM embedding model you use defines how precisely Knowledge AI calculates distance. The different LLM embedding models can provide considerably different distance values. Use the distance value for decision-making in Flows with caution.
Adapting Distance Values: Before and After 4.74.0 Version¶
If your Flows used distance values within the range of 0 to 2 before the release of Cognigy.AI 4.74.0 and you want to keep this approach, change your Flows as follows:
- Replace all
distanceinstances with1 - distance.
If you are using distance for the first time in Cognigy.AI v4.74.0 or later, use the value within the range of -1 to 1.