Text Mining on Elementary Forms in Complex Lexical Structures

After showing the advantages of formulating lexical structures with variable elements in terms of symbolic objects (), the Authors propose to introduce the information which determine their building in the analysis of elementary units ( ). It is worth noting that, dealing with symbolic data, the observed textual units disappear by the collapsing procedure. In order to visualize forms, an analysis on elementary data, introducing the external information on the complex structure they belong, has been proposed. This analysis can be usefully performed complementary to the symbolic objects analysis, because it enables to analyze the dependence relations of the forms on the contextual information in which they have been used. In order to enrich the analysis of textual data, it is possible to introduce other external information, related to the fragments where the forms appear. In doing that by a double partial analysis, we represent on low-dimensional spaces the relational structure existing between the two sets of information introduced. Forms and fragments can be represented as supplementary points, in order to study the role they played in those relations. An application dealing with a very large of lexical structures with variable elements, extracted from the Italian newspaper “La Repubblica” during the Nineties, has been performed, in order to show the relation between years and contextual information in the different identified context, and the single forms mainly involved.