“…The most complex models consider not only the relationship between form fields and their semantic labels, but also some advanced features from each form, such as the existence of groups of semantically or logically related fields [24,29,31,45,81,86,98], mandatory fields [13,22,29,59,81], the domain of each field [13,29,44,75,80,98], the measurement unit of a field [24], the relative order between fields [64], the distinction between fields that represent attributes from a given domain and fields that define how to display the search results [81], or an estimator of the probability of the form offering information about a particular domain [1], amongst others. The extraction of these models is usually performed using similar techniques as in the label extraction problem, with the help of other HTML features, such as separator lines, font sizes, or CSS styles [98].…”