Corpus Pattern Analysis (CPA) provides an inventory of syntagmatic patterns for each verb together with links to word's hits in a text corpus. The aim of this thesis is to find and implement an algorithm which process a sentence containing a verb and assign a pattern (or probabilities for each patter) matching the context of the verb in the given sentence. The algorithm will learn rules or probabilities from existing data, it can also use external lexical information like ontologies, or statistical thesaurus. An important part of the work is an evaluation of results.

The work will consist of the Thesis text, working implementation of the algorithm and evaluation of results.