Making data analysis understandable, reproducible and accountable
An investigation into whether natural-language analytics can be more accessible without surrendering calculations, evidence or human judgement.
Read the studyResearch, case studies and reflections connected by this topic.
An investigation into whether natural-language analytics can be more accessible without surrendering calculations, evidence or human judgement.
Read the studyWhy language interpretation should not decide the numerical answer, and how a governed question contract separates intent from execution.
Read the studyWhy an understandable analytical result needs a route back to its calculation, evidence and provenance.
Read the studyHow a finding can become a report without losing its calculation, context or relationship to the original evidence.
Read the studyHow different people can receive useful analytical explanations without role or organisation context altering the underlying result or evidence.
Read the studyWhy an application rollback should not become a customer-data rollback, and what that boundary means for integrity, recovery and product responsibility.
Read the study