الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 148 |  |  |
| | | from | 148 | | |
Abstract
This thesis is interested in exploring the field of automatic assessment of free-constructed responses. It takes interest in investigating the nature of automatic scoring systems of free-constructed responses while shedding some light on their advantages and their points of weakness.
The study aims at enhancing the Arabic natural language processing resources. Therefore; it proposes a system that is capable of automatic assessment of free-constructed responses targeting responses that are short in form and that are written in Arabic language.
In order to automatically assess free-constructed responses of the study data, the system developed employed a hybrid approach that uses full natural language processing techniques to produce semantic networks in the Universal networking language format. The output of this process is a number of semantic networks for all the study data that includes the participants’ answer and the referent model answer, which are then compared against each other for the purpose of matching and hence providing the accurate score. The assessment technique employed enhanced both the correlation and human-computer agreement values. The proposed model scored a Pearson correlation equals r= 0.821 and human-computer agreement = 79.4%.