Slang Word Detection in the Context of User Profiling in the Social Media Platformes

On social media platforms, people express their views in various ways, sometimes using formal words and sometimes in informal form. The informal conversations may contain jargon words i.e. slang words which might be improper for all kinds of audience. In this work, an attempt has been made to identify slang words in a text, posted on the social media platform. The overall experiment has been carried out in 4 modules. In module-1, the suspicious words from the posted text have been identified using minimum edit distance. This experiment has been implemented by the help of a dataset of printable slang words collected from Kaggle. In module-2, the topic of discussion has been analyzed by using a knowledge based approach. This part of the experiment has been carried out by creating a newspaper archive from the online repository of the "Times of India" newspaper. But, this knowledge based strategy could not produce an appreciable accuracy at its baseline. Therefore, as a modification of this technique, in module-3, the context analysis task has been implemented combining knowledge based approach and machine learning based approach. In module-4, the module-1 and module-3 are merged to tag a suspicious word with its level of abusiveness.