Abstract:
The academic performance of the university students is measured with the
help of the Grade Point Average (GPA). Further, the marks of a subject of a
degree program are accumulated to identify the final academic performance
of the degree. Unfortunatel y, the achievements of special power skills such
as knowledge levels, evaluation skills and application development skills
are not highlighted in the current GPA. Therefore, this research was mainly
designed focusing on measuring students’ power skills aut omatically which
reflect specific performances of the students. Cognitive levels of the
Blooms taxonomy are identified as the categories of power skills.
Knowledge, comprehension, application, analysis, synthesis and evaluation
are the main cognitive skill levels of the Blooms taxonomy. Typically,
summative and formative assessments are held to cover the Intended
Learning Outcome (ILO) of the subject. The questions of the final exam
papers of the Computer Science stream of the Wayamba University were
used a s the dataset. First, a preliminary research was conducted to
categorize exam questions automatically. Natural Language Processing
(NLP) techniques such as tagging, spell correction, lemmatization, parse
tree generation and semantic similarity analysis tec hniques were used to
derive the features for summative assessments classification. Based on the
extracted features, rule set was identified to categorize the questions
automatically. Once the questions were categorized, the portion of the
marks allocated f or each Blooms taxonomy performance level was
identified. Based on the assigned marks for each category, students’
achievements for each category were calculated separately to measure the
power skills levels of students. This identification would immensely be
helpful to academics and universities to develop the best graduates with
high power skills.
