Social Media Profanity Word Filtering System Development

This work aims to examine a guideline for development of profanity filters for Facebook, and to assess the efficiency of the profanity filter for Facebook using the F-measure, as well as determining the accuracy, recall, and precision. The filter is developed as a Google Chrome extension and will filter profanity words based on dictionary-string comparison and then replace the profanity with designed symbols. The filter interfaces with Facebook through Facebook API to accept data from Facebook users. The technique used in this step was SWATh (Smart Word Analysis for Thai). The data will be then compared with words such as obscene words. If the user types in inappropriate words, there will be a warning shown on the user's screen, and the words will be hidden to ensure that the profanity words will not be shown on social media to reduce conflict and defamation. The filter aims to prevent abuse of Facebook. This study found that the F-measure was 0.95, precision was 1.00 and recall was 0.95.