You have been asked to oversee the design of a new system for identifying websites that are unsuitable to display targeting advertisements for your brand. Since new websites appear every day you are looking to automate the recognition whether a new website is suitable or not. You realize that the designers have not considered building a predictive model based on the content of web pages. Assume that you can associate a precise target variable of interest to each historical web page. Describe how you would suggest that they represent the web pages as examples in order that they can be taken as input to standard supervised predictive modeling techniques.