Mining Interesting Aggregate Tuples

International audience ; In business intelligence, the concept of data cube offers aggregate views over multiple dimensions of business. Computing the data cube is a challenge because of the exponential number of cuboids. This number is not only an important problem of computing, but also of searching for what is interesting or useful in the data cube. This paper presents the concept of interesting aggregate tuple that can be useful formanagers to decide on their business. This concept is usefull because (i)interesting aggregate tuples are those with important and credible aggregate values, and (ii) the number of interesting aggregate tuples is very small that can be considered by humans. The algorithm for searching forinteresting aggregate tuples is implemented and experienced on the real datasets.