In this paper, we present an intelligent architecture, called intelligent virtual environment for language learning, with embedded pedagogical agents for improving listening and speaking skills of non-native English language learners. The proposed architecture integrates virtual environments into the Intelligent Computer-Assisted Language Learning. This architecture supports visual, auditory, and haptic channels of interaction. It allows pedagogical ideas about language skills to be implemented and validated with a minimum design time. Moreover, we design a computational model to evaluate learner's proficiency level, and an automatic adaptation mechanism which adjusts to the learner's learning curve. We have implemented two scenarios based on the proposed architecture to teach learners how to communicate in public places such as airports and TV stores. Inputs to this system include learner's speech and hand motion, and outputs include graphical scenes, force feedback, and speech by a few embodied agents. Throughout interactions, agents discover the proficiency level of the learner and customize the level of communication complexity accordingly. The system is tested on 10 subjects. Experimental results show 14% increase in the number of proper replies, 3% decrease in grammatical errors, 16% decrease in pronunciation duration, and 11% increase in learners' proficiency level within three trials.
7 October 2015 in Journal of Mathematical Imaging and Vision
Form or shape is one of the main aspects of architecture design. A gap exists in scientific studies on categorizing different architecture interior forms according to design. This paper presents a methodology for categorizing interior forms of built places. The main innovation of this study was to evaluate the architecture interior forms of real built places as a base for any analysis on form. We proposed a clustering method by selecting 343 images of living rooms from residential places according to their history and interior design style. We labeled all the images in AutoCAD software depending on form features. The labeling results showed that images had 1104 distinct form features, including sloped, vertical and horizontal linear solids, and edges. Regarding the high dimension of data, we used Graphical Clustering Toolkit software for clustering, which involved the use of correlation coefficients and internal similarity among clusters. The clustering analysis grouped all the images into 25 clusters with the highest internal and the lowest external similarities. The descriptive features of each cluster could show its formal characteristics
15 March 2016 in Multimedia Tools and Applications