Language-integrated querying for efficient payment matching

MSc thesis defence by Simon Christiansen


This thesis presents a .Net library for language integrated querying. It is based on multiset discrimination and symbolic data structures, which support efficient join querying on two tables with any kind of data. The library is evaluated and compared to other joining techniques, such as the classical sort-merge and the language integrated Microsoft LINQ. These comparisons shows asymptotically better performance than LINQ under certain circumstances. It is not the inention to build a Microsoft LINQ-provider. However, this topic is cursorily discussed.

The library is evaluated in the context of a tool for performing efficient payment transaction matching. This is a process that exposes the difference between the bank balance shown in a bank statement, as supplied by the bank, and the corresponding amount shown in the organisation’s own accounting records at a particular point in time, by matching the figures in each table according to a set of rules. We have developed an efficient and scalable implementation of an interactive, automatic, scalable (quasi-linear time for fixed number of rules) payment transaction matching tool, based on expressive user specifications of Boolean matching rules.

Advisor and examinator: Fritz Henglein (DIKU)

External consultant: Nishandan Ganesalingam (Ernst & Young)

External evaluator: Mads Rosendahl (RUC)

All are welcome.

The defense will be held in Danish.