Some class-based segmentation models (Sun et al. , 2002; Gao et al. , 2003) well integrate the identification of some regular non-lexicalized units (such as named entities).
Only values are cleared-- all other properties of the cell(such as
formatting, data validation, etc..) are kept.
