UNL-ization of Punjabi with IAN

Abstract

In the field of Natural Language Processing, Universal Networking Language (i.e., UNL) has been an area of immense interest among researchers during last couple of years. Universal Networking Language is an artificial language used for representing information in a natural-language-independent format. This thesis presents UNL-ization of Punjabi sentences with the help of different example sentences, using IAN (Interactive Analyzer) tool. In UNL approach, UNL-ization is a process of converting natural language resource to UNL and NL-ization, is a process of generating a natural language resource out of a UNL graph. IAN is an online tool provided by UNDL Foundation. IAN processes input sentences with the help of TRules and Dictionary entries. The proposed system performs the UNL-ization of all the major Part-of-Speech, numbers and ordinals up to fourteen digits, written in words in Punjabi language, with the help of Analysis Dictionary and TRules. The system is tested on three different corpuses i.e., CORPUS500, UCA1, and UCA2 provided by UNDL Foundation. A detailed description of IAN tool has also been explained. F-Measure of the proposed system has also been calculated. F-Measure is the score used to calculate the grammar’s accuracy. F-Measure of CORPUS500, UCA1, and UCA2 has been calculated and detailed description is given in Chapter 6, i.e., Results and Discussion. F-Measure considers both the precision and recall of the grammar to compute the score. Its value lies between 0 to 1.