Automatic analysis of inflectional morphems in German nouns

Summary

The underlying algorithm can be used for automatic indexing. It is supposed to analyze every inflected German noun automatically by reducing it to its proper standard form and by distinguishing every inflected noun from non-inflected ones. The algorithm starts out with two presuppositions:

• 1. Nouns first of all must be considered to be the key words for automatic indexing as they contain the most relevant information of all parts of speech, and as they have statistically the greatest frequency.

• 2. Since in German we use capital letters for every noun we are able to distinguish homographs of parts of speech to a large extent.

The testwords are to be looked up in an automatic dictionary and after the complete analysis we get either the correct analysis of the testword or the result that the testword is not contained in the dictionary. The algorithm uses a maximum of four pieces of information with every word: inflectional group singular (1–9), inflectional group plural (1–10); information about the umlaut in the plural or/and the change of β to S S both depending on the inflectional system. The algorithm is supposed to give the following results:

• 1. The correct analysis of every German noun including foreign words, loan words and exceptions.

• 2. Automatic recognition of inflected and non-inflected noun homographs.

• 3. The impossibility of wrong analyses.

• 4. The algorithm works with a minimum of dictionary operations.

• 5. The dictionary can be relatively small according to the number of special terms which the specialist considers necessary for his field or for his special task.