A knowledge-based model of parliamentary election

The paper presents a new approach to the analysis of the electoral campaign based on the use of the methodology of machine learning. Attributes describing socio-political programmes of parties or individual candidates taking part in election are examined. The attributes are used to construct a knowledge-based model of parliamentary election. Parties or candidates are treated as examples. The sets of examples and their partition into disjoint classes form a starting point in the process of machine learning, which is supposed to lead to the descriptions of the considered classes. The classes are described in the form of the following decision rules: “IF certain conditions are satisfied, THEN a given example is a member of a specific class”. They can be used to form a set of action rules that specify the conditions of a hypothetical transfer to another class, i.e., the winners or the losers. The case study, based on the Polish Parliamentary election of 2007, illustrates the application of the proposed methodology.