Periphraser

Periphraser is a tool for storing and presenting knowledge base of conventionalized periphrastic nominal expressions (i.e. phrases headed by a noun) together with their textually attested realizations. For instance, the database entry for the phrase Robert Lewandowski in the referred below demo for Polish will include the phrase the Polish international while pediatrics will be featured as medical care for children. It allows contacting with database using REST API as well as exporting it to XML or CSV format. For Polish language, it also provides some more complex mechanisms like: automatic semantic and syntactic normalization, errors autodetection, and simple interface for commenting and marking possibli wrong entries or ones needing improvement.

Principal developer: Bartłomiej Nitoń
Periphraser license: BSD 2-Clause License
License for Polish Periphraser data (only free version of export): CC BY-SA 4.0 License

Download

If you want to set up your own instance of the service, you may download source code from the web service git repository.

Exports including (free) Periphraser data for Polish can be downloaded from here. If you are interested in full export (authenticated) of Polish periphrastic data please contact Bartłomiej Nitoń.

Online demo

The Web demo of the service containing data for Polish: http://periphraser.nlp.ipipan.waw.pl/.

Publications

2017

Bartłomiej Nitoń and Maciej Ogrodniczuk. Multi-pass sieve coreference resolution system for Polish. In Jorge Gracia, Francis Bond, John P. McCrae, Paul Buitelaar, Christian Chiarcos, and Sebastian Hellmann, editors, Proceedings of the 1st Conference on Language, Data and Knowledge (LDK 2017), number 10318 in Lecture Notes in Artificial Intelligence, pages 222–236. Springer International Publishing, Berlin, 2017.