| dc.description.abstract | When searching for information, the temporal dimension of the results is an important fac-tor regarding the information quality. Using temporal intent as a condition when searchingfor information is a field that is gaining increasing interest. When we search for informa-tion on search engines, such as Google, we have the option to use time of creation as partof the search criteria. Unfortunately, when searching on the web we have no guaranteethat the timestamps for the results corresponds to the actual date the content was created.Since the timestamps provided on the Internet can not be trusted it would be of great useif there existed a method for timestamping documents without knowing the actual dateof creation. In this thesis, we have presented and implemented some existing approachesto this problem, modified them and added some parameters for tweaking and fine tuningthe results. These approaches are so called content based approaches, and they use sta-tistical analysis on the textual contents of documents in a collection in order to predict adocuments time of origin. In order to evaluate our implementation, we have performedextensive experiments and compared our results with results achieved in earlier research. | |
