Fig. 7

Definition of new dictionary entries by identifying optimal keys. As a basic principle, each (re)classification of a given entity during the working process can be stored by updating the corresponding dictionary. In this example the incorrectly spelled city name Tubingen was corrected with Tuebingen (Value, second bold line). To store this substitution the user has the option to select valid keys defining new Key-Value-Pairs in the corresponding dictionary. In case of the city dictionary not only the original fault value is given, but rather a list of keys created by the combinatorial processing of the raw location string (first bold line). A key is valid, when its distinct mapping to the particular key is correct. An optimal key is a valid key which is not substitutable by another valid key. Here, the keys Tubingen (=original fault value), Dept Physiol Cognit Proc and Max Planck Inst Biol Cybernet (both institutions are unique in Germany) are optimal. All other keys are not valid due to their ambiguity (e.g. Max Planck) or not optimal to their substitutability (e.g. D-72076 Tubingen). To simplify the distinction between valid and optimal keys the program checks for key redundancy immediately after (de)selecting a key. In this example the valid key D-72076 Tubingen is inactivated due to its substitutability by the activated key Tubingen. Apparently, the role of the user is to identify optimal keys