For people in Indonesia, the regional language is the everyday language used to communicate. One of them is the Java language. In natural language-based research, regional languages are considered difficult languages to develop, given the availability of a limited number of datasets. This study analyzes 2 word stemming methods, namely the Nazief-Adriani method and the Levenshtein Distance method to carry out the Javanese word stemming process. This study wanted to find out the appropriate method with the best accuracy for stemming Javanese words. In addition, this study also considers word weighting to produce better article similarity accuracy. The nazief adriani method produces an average similarity value of 6.8% with an average execution time of 0.0443 second.