User Manipulation and User Interaction with Input Techniques on Desktop Computer


Student thesis: Master's Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date19 Nov 2020


Although the portability of Desktop PC is much lower than mobile devices such as smartphones or tablets, people still using it for word processing and handling complex tasks. Computer users use their computer with mouse and keyboard as an HCI media to interact with the GUI, but it is inefficient as it is an indirect operation of controlling the mouse cursor and requires a series of operations for every single task. The launch and common adaptation of touchscreen offer a direct touch interface for target-selection which helps to shorten the target-selection time, but it brings out other issues such as the fat finger problem [7, 42, 86] and slow typing speed on a virtual keyboard [51]. Some researchers suggested to use or even create additional hardware [6, 12, 24, 88] to assist specific input tasks, but the cost of implementation is high, and it is a difficult practice to apply the product to everybody. We study the issues that general computer users using keyboard and mouse are facing for daily operations, and investigate the additional design space for keyboard and mouse interactions in which no hardware modification of a PC is required, to explore new techniques to overcome the task completion time and accuracy issues.

In this thesis, we present two contributions to address two existing issues on human-computer interaction (HCI) on a desktop PC:
1. Keyboard Shortcuts are broadly used in typical typing and many computer GUI operations. Modifier-based shortcuts are the most common type of shortcuts being used in different software, which are dependent upon the order of key press and avoid simultaneous key pressing. Although those shortcuts help to shorten the steps and overall task completion time of certain operations, users have difficulties to remember a series of shortcuts. We introduce DownChord and UpChord, a novel keyboard shortcut technique, which utilizes the simultaneous key pressing and releasing for defining new classes of shortcuts. Our technique can be used on ordinary unmodified physical keyboard seamlessly – without interfering major functionality of computer keyboard such as typing and activation of existing modifier-based shortcuts. We conducted controlled experiments to determine the appropriate time thresholds for detecting shortcuts from the input keyboard event queue, which allows our shortcuts to be activated effortlessly with an only average latency of 160 ms. Results of our user study show that DownChord and UpChord are promising shortcut techniques that give a competitive performance as existing modifier-base shortcuts.

2. Target-Selection on Desktop PC is a well-known and long-lasting issue in the HCI area. An indirect operation of controlling the mouse cursor affecting the performance of selecting the target in terms of target-selection time and accuracy. We present Additive Voronoi Cursor (AVC) – a new cursor technique for target-selection by dynamically resizing the area cursor based on the analysis of the two different phases of mouse movement: the ballistic and the correction phases during target-selection. On-screen Targets can be divided into respective areas dynamically based on both target distribution and cursor velocity. We assumed that to select a target, a user will first perform ballistic/fast cursor movement aiming to the target roughly, then correct the cursor position with slower movement towards the desired target. Therefore, after the ballistic movement, the desired target would locate within the local region closed to the cursor. We defined Additive Weighted Voronoi Diagrams with selectable targets by assigning larger weights to the nearby objects right after the ballistic cursor movement. Therefore, the effective areas of the nearby objects are enlarged, and they can be selected more easily and quickly. We had compared our cursor technique with recent developed area-cursor methods. The results showed that our method performed significantly better on certain configurations.