Character error rate (CER)

 

 

CER supports Safety by reducing the likelihood of harmful or misleading outputs due to transcription errors, which is especially important in domains like healthcare or legal transcription. It also supports Robustness by providing a measurable indicator of the system's reliability in producing accurate outputs under standard conditions. However, CER does not address the full scope of these objectives, as it does not account for performance under adversarial conditions or broader safety controls.The Character Error Rate (CER) compares, for a given page, the total number of characters (n), including spaces, to the minimum number of insertions (i), substitutions (s) and deletions (d) of characters that are required to obtain the Ground Truth result. The formula to calculate CER is as follows: CER = [ (i + s + d) / n ]*100

Trustworthy AI Relevance

This metric addresses Robustness, Transparency by quantifying relevant system properties. CER connects to Robustness because it quantifies how well an AI system maintains accurate text output despite potential noise, accents, or other adverse input conditions, thus reflecting system resilience and reliability. It supports Transparency by providing a clear, quantifiable metric that can be disclosed to users and stakeholders to communicate the accuracy and limitations of the AI system's outputs, enhancing understanding and openness about system performance..