AI Evaluators
Context Recall
Assesses how closely the retrieved context matches the expected output by identifying key statements in the expected output and evaluating whether each is represented in the retrieved context.
Input
- Required Inputs:
expected_output: The expected response that should be generatedcontext: List of context chunks retrieved for the response
Output
Result: Value in the continuous range [0, 1]Reasoning: Detailed explanation of recall assessment
Interpretation
- Higher score (closer to 1): Better recall - most or all key statements in the expected output are represented in the retrieved context
- Lower score (closer to 0): Poor recall - few or no statements in the expected output are represented in the retrieved context
A higher context recall score reflects a retriever’s ability to capture and return the full set of relevant information from the knowledge base, indicating a more comprehensive retrieval process.