Article Preview
Top1. Introduction
The future of a society largely depends on the success of its young generation. To achieve this objective, it is essential to make foundations robust to the core – and excellent education and knowledge engineering might well be considered the key-factors in this context. But like many other human endeavours, judicially choosing the correct procedure and standards is often a tedious, time-consuming and involved process, further hindered by personal opinions and favouritisms. Since AI has already set wings to the power of the machine by providing assistance in all tasks hitherto handled by humans alone, it seems appropriate to create a comprehensive Student Evaluation model, which, given some internal interaction, can take human-like decisions and automatically come up with intermediate suggestions and final categorization.
Such an innovative educational model, allowing constant interactive updates from students, achieves early detection of outlier performance. Removal of the outliers from such a system facilitates the students to be classified into various groups based on their capability. Adequate measures can thereafter be arranged for each group; exceptional performers remain a class apart, being the select few who can help the weaker section. Those needing this assistance form a class of their own. There exists an interim class consisting of students who can maintain their progress on their own.
Outlier detection is a technique to identify the presence of unusual patterns within a system, which do not conform to the general expected behavior (Singh & Upadhyaya, 2012). In educational domain, outlier performers refer to the group of students who perform below or above a statistically determined permissible range. Outperformers or positive outliers are those receiving marks above the upper limit. Their innovative responses help to enhance existing knowledge bases, opening up new research domains. On the other hand, a few performers exist who fall below the lowest standard – they are the poor performers or negative outliers who cannot be treated at par with any other groups within the student community. They usually need Special Care.
Test based examination is the most common and widely used technique to assess a student’s knowledge level. However, conventional tests do not provide scope for assessing variable degree of understanding and confidence. Improved methodologies are obviously needed to measure these factors by preserving the detailed profile report of each student with graded questions at every quantifiable level of a specific subject. This research work proposes an assessment model to mitigate the above needs.
The general motivation behind the currently proposed model arises from the urge to automatically detect knowledge levels - crucial for perfecting a student’s learning curve. This may require a detailed profiling report, which includes area-wise expertise in a subject, asking for level-specific adaptive assessment techniques at each stage. Literature surveys on existing learning models reveal that most of these lacks in extracting psychological factors such as levels of patience, confidence and perseverance of the participants. But these traits are essential for perfecting a learner’s knowledge base. The proposed model enhances its academic assessment capability by capturing these other character revealing features as well. Thus the proposed system has an edge over existing traditional models in judging both technical and psychological acumen of a student.