The aim of the Knowledge based Business Intelligence (KNOBI) project group is to develop an analytical information system in the context of hospital market analysis (HMA). The intended system should make it possible to "actively" support so-called business users during the analysis of integrated data on the basis of explicitly modelled analysis and domain knowledge. Analysis knowledge includes information about data structures, analysis operators, methods and result visualisations. Domain knowledge, on the other hand, describes which analyses are useful for given questions, which business rules exist and how data is to be interpreted in terms of its semantics. Based on this, the system can, for example, automatically identify regions with a loss of market share and suggest further analysis steps, such as analysing referring doctors from these regions.

"Business users" are typically people in a company who are responsible for strategic decisions and need answers to various questions, such as "Where is my catchment area?" or "How big is my market share in the catchment area?", etc., in order to make a decision. To answer such questions, various data sources usually have to be integrated and analysed in a uniform manner. In the KMA context, this can be, for example, the data of a hospital with further data from the structured quality reports and population data of a Federal State from the associated Federal Statistical Office. Data warehouse systems have established themselves in the field of business intelligence as a standardised, integrated, quality-assured data basis as a so-called "single point of truth", which holds the integrated data in multidimensional structures (cubes). Analysis applications can be based on this, allowing the data to be analysed using online analytical processing (OLAP) paradigms and other geographical and statistical methods. As carrying out analyses in such an analytical information system (DWH + analysis application) requires a deeper understanding of the multidimensional data structures and the analysis procedures that can be carried out on them, these are usually so complex that business users are dependent on the support of specially trained analysts with the corresponding analysis and domain knowledge.

In the German hospital market in particular, legal reforms have led to a greater need for hospitals to develop potential in a targeted manner and expand their range of services competitively in order to secure their competitiveness. It is therefore necessary for hospitals to analyse the respective hospital market. The use of analytical information systems is possible in principle, but very few hospitals have dedicated analysis departments with the necessary expertise to analyse the available data. There is therefore a need for so-called "information self-service" within the framework of an analytical information system. Business users should be enabled to satisfy their information needs in an intuitive, fast and efficient way. To this end, they should be provided with the appropriate data for their questions and offered the possible and meaningful analysis options.

However, current analytical information systems lack the explicit analysis and domain knowledge that trained analysts implicitly possess in order to perform the corresponding support tasks for the business user. In addition to integrated structured data, there is no further information on definitions, business objectives, business rules, strategies, specific analysis methods and procedures or on the semantics of the data and how they relate to each other.

In order to empower business users in the sense of "information self-service" for analysis tasks, an architectural concept for a semantic analytical information system is being realised within the project group. The central element will be a "semantic meta-level" that can integrate, record and manage explicitly modelled analysis and domain knowledge in a machine-readable and comprehensible form in an analytical information system and make it available for new functionalities of intelligent analysis support. Based on this, a semantic search can be implemented that enables business users to find and analyse suitable data and processes in the analytical information system based on their questions. In the analyses, they can navigate along semantic relations with a specific meaning and thus follow different analysis chains - sequences of analysis steps. In this context, a suggestion generation (recommender) is implemented, which, based on the predefined analysis and domain knowledge - business rules, analysis strategies, context information, ... - shows meaningful further analysis options and analysis results appropriate to the question. The technical realisation of the analytical information system and the analysis paradigms - multidimensional data model, OLAP operators, MDX queries, ... - are abstracted as far as possible, so that business users without the relevant IT knowledge are enabled to obtain suitable information intuitively, quickly and efficiently.