KDDML - Knowledge Discovery in Databases Markup Language
Company:
Università di Pisa - Dipartimento di Informatica
Description:
KDDML is a middleware language and system in support of the KDD process heavily based on XML as a representation language for data, models and queries. The language is primarily intended as a middleware language on the basis of which higher abstraction levels can be built, such as vertical applications or more declarative languages. Also, the language tries to be as much as possible independent from lower level implementations of data mining algorithms, with the aim of confining the technicalities at the level of the implementation of the KDDML system. Query tags specify data acquisition, preprocessing, mining and post-processing algorithms taken from possibly distinct suites of tools (Microsoft SQL Server, Oracle, Weka, XQuery). Compositionality of the algorithms is obtained by appropriate wrappers between algorithm internal representation and KDDML representation of data and models.
The KDDML system is implemented in Java and consists of a graphical user interface for editing queries and an interpreter for executing queries. The design of the KDDML system had to take into special account the requirementes of extensibility of the KDDML language, which can be distinguished into data sources, algorithms and models extensibility. The overall architecture of KDDML is structured in layers. Each layer implements a specific functionality and supplies an interface to the layer above.