This paper studies the problem of mining for data values with high information gain in relational tables. High information gain can help data analysts and secondary data mining algorithms gain insights into strong statistical dependencies and causality relationship between key metrics. In this paper, we will study the problem of high information gain identification for scenarios involving temporal relations where new records are added continuously to the relations. We show that information gain can be efficiently maintained in an incremental fashion, making it possible to monitor continuously high information gain values.