The time of creation of a document is always assumed to be part of the available meta-data in any given dataset. In some cases however, such as collections of old digitized documents, this data is unavailable or ambiguous. This work aims to attach a date of creation to any document, by considering the temporal behavior of its words. The proposed method combines the lexical similarity between the query documents and the bursty-ness of its terms to tackle this problem.
You can read more: