The OntoNotes DB Tool provides code for working with the OntoNotes data, including building and accessing the relational database. The linked tables of the database store a text corpus along with the layers of annotation specifying syntactic structure, propositional structure, word senses for nouns and verbs, names, and the coreference between entities mentioned in the text. It also contains metadata such as proposition frames, sense inventories, and where available, speaker information, pointers to the original or translation and lemma details.
Also included are functions that extract particular views of the merged data. Each of the individual input annotation formats, which was used in building the database, can also be extracted as a view. There is also a combined “OntoNotes Normal Form” view that includes all of the layers in a version intended for human review.
In addition to the top-level routines used to build the database and to extract views, the DB Tool provides an API with access to each of the individual tables, allowing users to construct many kinds of database queries more conveniently than would be possible in raw SQL, especially for defining new views.
The OntoNotes DB Tool was built with cross-platform compatibility in mind, and therefore all the core components of the system as well as the external libraries that were used are themselves available for multiple platforms – at least for Linux/Unix and Windows. Given the limited resource availability during development, however, all development was done in the Linux environment and we could not thoroughly test the distribution across other platforms. It would be quite surprising if it does not work out of the box, or with a few minor tweaks in a Windows environment. The majority of the code is written in Python. The version used for development was Python 2.5.1.
Since this is a software in its Beta release and has not been thoroughly tested, it is quite likely that it has bugs in it. Please refer to Appendix A: Reporting Bugs for more details in how to go about reporting bugs that you might find.
See also
The OntoNotes Project The official OntoNotes project webpage, with details about the individual components, latest developments, etc.