Quantifying and neutralising sexually explicit language

Detecting the degree and character of sexually explicit textual content is a feasible and potentially useful enterprise. Such a facility may assist in preventing child exploitation, for example, by automating the detection of the highly sexualised content of on-line grooming, and may be beneficial in SMS for detecting offensive sexting. In addition, an ability to determine the nature and ‘strength’ of sexually explicit content would prove helpful in contexts where students, trainees or other professionals need to be exposed to sexually explicit language in documents. In such settings, we might deploy such detection and quantification toward managing the content, for instance, by means of progressive ‘neutralisation’. In the following, we elaborate upon the relevance of such quantification and describe a variety of steps toward neutralisation