Forget about data explanation!
Get to the root cause using query explanation.
Traditional database cleaning and explanation research has focused on techniques such as anomaly/outlier detection, dimensionality reduction, data summarization, predicate synthesis, ETL, and others, as ways to prevent, detect, or resolve (clean) data errors.
However, these data oriented approaches ignore a key vector of data errors -- queries! The majority of user facing systems -- billing, HR, customer service, web application -- generate queries based on user submissions. Can we use the sequence of past queries from the transaction log to both find and repair incorrect errors due to user slip-ups?
QFix is a fundamentally novel approach towards getting to the root cause of errors in your database. Instead of describing data errors by examining data in the database, explains and fixes data errors by analyzing your past queries and suggests ways to repair the incorrect queries that it finds.
- Xiaolan Wang (PhD Student)
- Alexandra Meliou (PI)
- Eugene Wu (PI)