**3.3 Comparative analysis of intelligent user interfaces**

Intelligibility, predictability, usability and adaptation are four characteristics that must be consider while considering User Experience in the design of IUI as shown in **Table 2**.

The Intelligibility of IUI as it relates to how the users understand the interface actions was high when it comes to direct manipulation and low in indirect manipulation. This gives a better understanding of how and when intelligence can substantially improve the design practice (interaction) in the application of cyber security for threat model [47]. On the accuracy of algorithm which explain the predictability, there are many algorithms such as Dynamic Bayesian Network and Naïve Bayes can be used for user's prediction and techniques such as adaptation techniques that provide intelligence for enhancing the predictability of the UI and creating a lasting UX. The performance of these techniques decrease when the accuracy of algorithm by the system becomes low [44]. The usability principles for evaluating Intelligent User Interfaces (rather than direct manipulation systems) ensures that it does not mislead the user's expectation [48]. The provision of intelligence in a UI is highly reliable and cost efficient in direct manipulation than indirect manipulation as the authoring tool enable easy development and maintenance of the intelligent parts of the system [50] especially when it concerns threat modeling due to consideration for various risk metrics.
