Abstract: A method and apparatus are disclosed for automatically finding and classifying any group of trademarks belonging to an initial series and having similarities with groups of trademarks of another series. The classification is arranged in increasing or decreasing order according to the risks of conflict involved in these similarities. The analysis of the similarities provides an indication of the likelihood of confusion between trademarks which are proposed to be registered and those already registered. The apparatus includes a memory for storing the first and second groupings of trademarks. The groupings are then analyzed in pairs by a comparing and counting device which determines for each pair the numbers of the shortest and longest groupings, the longest sequence of common signs to both groupings, the number of signs common to both groupings other than those in the sequence and the number of signs of the longest groupings not found in the other grouping.