School of Science and Technology 科技學院
Computing Programmes 電腦學系
| Programme | Bachelor of Computing with Honours in Internet Technology |
| Supervisor | Dr. Kelvin LEE |
| Areas | E-learning |
| Year of Completion | 2019 |
Recently, many programmers have used mobile devices, such as smartphone and tablet, to code since IDE (Integrated Development Environment) is available in Android devices, and coding on mobile devices is more convenient while programmers working in the outdoors. However, when using mobile IDEs in Android devices to write apps, the speed of typing will be slower than coding on the computer. Moreover, because of the limited screen size in devices, it also easily leads to typos, causing user typing with slow speed.
This project aims to provide a new keyboard interface for programmers who write Android apps and programs in Android devices in order to achieve a coding speed that is similar to using a traditional hardware keyboard.
To achieve the aim, the main objective is to provide a new keyboard interface for Java, C, and Python programmers to work more convenient
The project has defined a number of sub-objectives as follows:
Techniques and technologies used
N-gram Language Model
N-gram Language Model has dealt with the problem of Auto-Prediction. Based on users' input, Auto-Prediction will predict what programming keywords that users are going to write.
T-9 Keyboard
In order to help Java programmers achieve a coding speed as fast as the coding speed of programmers using hardware keyboards, the project has applied T-9 Keyboard which is an approach to design each key combing with three characters.
KeyboardView
Associated with T-9 Keyboard, KeyboardView helps design a desktop-like QWERTY keyboard, in which each key consists of three to four characters.
Figure 1: The data flow between the components
Text Predictive System will match user inputs to the corresponding keywords.
The user input is actually a series of numbers. Then, pre-processing will convert each of the numbers to the corresponding letters. Before starting the search, a dictionary including common programming keywords is required and is loaded into the system for searching the possible results.
Figure 2: JavaKey’s keyboard layout
Figure 3: Pre-set programming methods
Figure 4: A user searching for programming keywords
The system will return the possible results while users typing on the keyboard. Developers are also able to check out the methods list to see if there are suitable methods. In addition, users are able to manage the programming keywords in the dictionary. Permissions are given to users to add, edit, remove, and retrieve the keywords from the dictionary.
Figure 5: Clearing keyword dictionary
| Keyboard Type | Testing Time |
| Android Traditional Keyboard | Around 251 seconds |
| CodeBoard Keyboard for Coding | Around 186 seconds |
| JavaKey (Our project) | Around 102 seconds |
Table 1: Average of each keyboard typing time
In the test, JavaKey was a lot faster than Android traditional keyboard around three minutes. Besides, by the same token of Keyboard type, CodeBoard Keyboard for Coding was slower than our project. Therefore, among the selected three existing programming keyboards for smartphones, JavaKey was the fastest.
First, newer application should deal with the problem of repeated searching and fixed-length searching engine.
Second, the function of changeable key size should be added since different users might have different finger sizes.
Third, in the future, the project should include the function of uploading the whole dictionary instead of users inputting the programming keywords one by one.
Forth, more common programming languages, such as CSS, HTML, and Kotlin, should be supported in JavaKey.
Finally, in the further version of the application, the keyboard can contain other styles of keyboard interfaces for users to download based on users' preferences.
