Controlled natural languages (CNLs) are engineered languages that are based on natural language, but have their vocabulary, syntax, and/or semantics restricted. The motivation behind them is to have a language that, on one hand, looks as natural as possible and, on the other hand, is simple and unambiguous. Depending on the purpose of a CNL, it may have a completely formal semantics or can be informal or semi-formal.

The application area of CNLs is quite broad and includes writing technical documentation, usage in news, communication between ships and harbors, mathematical authoring, legal texts, exchanging medical information, crisis management, etc. There are also general-purpose CNLs which are mapped directly to some kind of formal logic, independently from any application domain. Attempto Common English is probably the best known such a CNL. While fragments of English dominate the landscape of CNLs, controlled sublanguages have been developed for many other languages as well, including not only those that have well-developed computational resources, but also some others whose computational resources are scarce.

The goal of the project is the development of Georgian CNL. So far, there no such language created for Georgian. We aim at a general-purpose solution: A CNL which is not bound to any specific domain, with clearly defined logical semantics and the capabilities to be adapted to specific areas. Such adaptation capabilities should make it possible to develop special-purpose Georgian CNLs for various applications (crisis management, medical information, technical documentation, tourism industry) in a modular way. Developing such domain specific languages would increase the understandability and precision of communication in critical areas, as well as the computational processibility of a number of written Georgian applications.

Our contributions will be a new general domain CNL for Georgian, a tool facilitating writing in the new CNL, and also evaluation of the obtained language. Studying the ambiguity of Georgian and ways for reducing it, would also bring a lot of benefits in terms of scientific discoveries and further natural language processing for Georgian. The project will be interesting also for the CNL community in general, to see how the established techniques can be used or adapted to a language that is pretty different from most of the languages for which CNLs have been defined so far (for a list of such languages, see, e.g., Kuhn, T. (2014). A Survey and Classification of Controlled Natural Languages. Comp. Ling., 40(1), 121-170.).

Project members:

• Besik Dundua (Principal Investigator)