A Preflow Approach to Maximum Flows in Parametric Networks Applied to Assessing the Legal Information

摘要

The present paper proposes a fractal-like approach to the parametric flow problem, derived from the rules and recursion of generative linguistics. In the same manner in which any sentence can be analysed in terms of "theme" of the sentence (that which is spoken about in the sentence) and "rheme" of the sentence (that which is said about the theme in the sentence), in the proposed parametric preflow-push algorithm, a "partitioning push" (a non-cancelling push of flow in the parametric residual network) might leave the node unbalanced for only a subinterval of the range of parameter values. This will lead to separating the problem for the two disjoints subintervals, which allows the algorithm to continue after the same rules, independently, on each of the partitioned subintervals. The algorithm runs as the template-like structure of a dialogue act which reveals a design where information about the items (part-of-speech) is a two sections vector with one segment for each of the used part of speech categories. The article also proposes a possible application of the algorithm in assessing legal information and in semantic evaluation of legislation.