Logs.txt

Release 1 (Done 12/06/2009)
- revised with some errors occurred during conversion of PDFBox and Sentence Boundary Tool

Release 2 (Done 13/06/2009)
- filter strange characters (code from 0 to 25) and remove redundant sentences (containing authors, emails, references and acknowledgements) using heuristics

Release 3 (16/06/2009)
- annotation with topical information

Release 4 (18/06/2009)
- correct previous versions with title (the first line) for each article

Release 5 (11/07/2009)
- preprocessing (tokenization and lowercase)
Errors from PDFBox:
+ some of information extracted from PDF to text conversion were lost, especially errors located at positions near to "Figure, Table" captions
- add citation information for easy identification of later processing (e.g. evaluation)

Release 6 (08/09/2009)
(based on Release 5)
- Extract raw texts from OmniPage and manually correct errors
- Annotation for topic hiararchy

Release 7 (22/09/2009)
(based on Release 6)
- Sentence boundary for Release 6
- Annotation for topic-based evaluation
+ citation for each sentence with claim or references
+ topic for each reference
- (corrected)some unexpected and unknown errors about strange characters (but not affect the results of MEAD toolkit) included in two:
+ p203-wu
+ N09-1042

Release 8 (22/09/2009)
(based on Release 7)
- Tokenization & Lowercase

Release 9 (23/09/2009)
(based on Release 8)
- Stop words removing
- Stemming

Release 10 (01/08/2010) - current version
(based on Release 8)
- Make the dataset available
- Write Readme.doc


Cong Duy Vu Hoang