Usually usability experts do not have enough time to confront
all of the problematic usability dimensions to improve the overall
web-based information system usability. Therefore, they would
like to know which measures (checklist items) are the most problematic
in terms of the usability performance. The conventional
approaches as summarized in Section 1 merely consider the lowest-
rated checklist items to start improving the usability of a system.
If this were the case in my application, I would have
considered only Table 2 results. Therefore, based on the traditional
usability evaluation and improvement strategies I would have
started the improvement process confronting the usability problems
that can be attributed to A1, N2, and A2, respectively through
the Table 2 results.
However, such an approach ignores the effect of each change on
the overall usability of the WIS in hand. Namely, is it really worth
improving UWIS checklist item A1? It may or may not have much
effect on the overall usability although it is the lowest-rated item
by the end-users. The decision support system that is presented
in Section 2 proposes to calculate the criticality index for each of
the items in Table 2. To achieve this, I first determine the model
which best explains the relationships between the independent
variables (UWIS checklist items) and the dependent variable
(overall usability). Based on a 10-fold cross-validation evaluation,
various machine learning techniques (support vector machines
– SVM, neural networks – NN, and decision trees – DT) and a fundamental
statistical technique (multiple linear regression – MLR)
were employed and compared in terms of performance criteria.
Usually usability experts do not have enough time to confrontall of the problematic usability dimensions to improve the overallweb-based information system usability. Therefore, they wouldlike to know which measures (checklist items) are the most problematicin terms of the usability performance. The conventionalapproaches as summarized in Section 1 merely consider the lowest-rated checklist items to start improving the usability of a system.If this were the case in my application, I would haveconsidered only Table 2 results. Therefore, based on the traditionalusability evaluation and improvement strategies I would havestarted the improvement process confronting the usability problemsthat can be attributed to A1, N2, and A2, respectively throughthe Table 2 results.However, such an approach ignores the effect of each change onthe overall usability of the WIS in hand. Namely, is it really worthimproving UWIS checklist item A1? It may or may not have mucheffect on the overall usability although it is the lowest-rated itemby the end-users. The decision support system that is presentedin Section 2 proposes to calculate the criticality index for each ofthe items in Table 2. To achieve this, I first determine the modelwhich best explains the relationships between the independentvariables (UWIS checklist items) and the dependent variable(overall usability). Based on a 10-fold cross-validation evaluation,various machine learning techniques (support vector machines– SVM, neural networks – NN, and decision trees – DT) and a fundamentalstatistical technique (multiple linear regression – MLR)were employed and compared in terms of performance criteria.
การแปล กรุณารอสักครู่..