Reporting for Enterprise Systems


Time: Thursday, December 11, 11:00-12:45.

One of the essential features of Enterprise Systems is the ability to provide businesses with reports on, how the business is running and to assist businesses in providing the authorities with the reports required by law.
Over the last couple of decades there has been an increase in the amount of information registered by Enterprise Systems and thus reports typically depend on a huge amount of data. However, businesses still need the reports in a timely manner, preferably in real-time (or near-real-time) to help monitor the business and to satisfy the legal requirements.

The Hypothesis is: Using a strongly normalizing declarative programming language for expressing report functions, where Automatic Incrementalization is used to optimize these, can in many cases yield real-time (or near real-time) computation of report functions.

The progress report consists of four parts addressing practical and theoretical issues of building a preliminary argument for the hypothesis. The parts are:

I. A survey of the existing technologies used for expressing report functions and doing report function computations and a description of why these technologies do not have good support for real-time computation.
II. A proof of concept paper showing how "Automatic incrementalization" of report functions written in the programming language FunSETL, can give an asymptotic improvement, as well as a practical speedup in the computation time of report functions.
III. A generalization of Finite Differencing to list functions, and a description of how this can be used to construct incremental functions, which can lead to real-time computation.
IV. A sketch of how the generalization of finite differencing to list functions, can be used to automatically incrementalize functions in a programming language called FunDIF.

Examiners: Fritz Henglein & Ken Friis Larsen, DIKU
Censor: Arne Glenstrup, ITU.