This study aims to compare the results obtained, the efficiency and effectiveness of the implementation of the scancode method in the encoding process applied to Adaptive Resonance Theory (ART) which previously used the ASCII method as the default knowledge encoding. The implementation of ART is also carried out to handle Typographical Errors (typo) which is based on the most common method of distinguishing the spelling of two words in spelling, namely Levenshtein distance, which is comparing point by point similarities between two words and also Hamming. The data used as the research object was obtained randomly from several respondents who were documented in real-time so that the respondents could find out what data we would use. This study gives the results that the use of keyboard scancode as knowledge encoding in ART implementation can work as well as using ASCII as well as the clustering results obtained.