https://hdl.handle.net/10388/14418

Cynthia, Shamse Tasnim (2023) Improving The Usability of Software Systems Using Group Discussions: A Case Study on Galaxy, USASK Harvest - Theses and Dissertations, https://hdl.handle.net/10388/14418

Usability problems in software systems cause performance degradation, user dissatisfaction and loss in terms of cost. There is a growing need for the software systems to become more accessible, retrievable and usable for the users. The usability test of a software is conducted by getting the opinions directly from the users and its goal is to identify problems, uncover opportunities and learn about target users' preferences. But accessing real users is very difficult for certain software systems. However, there are many popular user forums such as Stack Overflow, Quora, Stack Exchange, Eclipse Community Forum etc. and people from different domains of knowledge use these forums to ask about their problems and post their concerns. So exploring these forums should provide significant knowledge for getting information about a system's usability issues. Previous studies show that investigating these group discussion forums discovered several usability issues that the system was unaware of such as topic categorization, automatic tag prediction, identifying reproducible codes etc. However, there are many Scientific Workflow Management Systems (SWfMSs) such as Galaxy, Taverna, Kepler, iPlant, VizSciFlow etc. and although these SWfMSs are emerging and important for data extensive research, no study has been done earlier to figure out the usability problems of these systems. Therefore, in this thesis, we take Galaxy, a well-known SWfMS, as our use case. We explore the user forum that Galaxy offers where users ask for help from experts and other Galaxy users. We search for the issues users are discussing in the forum and find out several usability problems in different categories. In our first study, we try to group the usability problems to easily identify them and galaxy community can be informed of the existing usability problems of the system. While exploring the posts, we find a significant percentage (up to 28\\%) of them lack tags. If tags are found, they do not reflect the context of the posts properly. This leads to one of the major usability problems for the discussion forums as users will be unable to identify suitable posts without proper tags. Moreover, users will face difficulties to explore the answers in those untagged questions. So in our second study, we try to suggest tags based on the context and proposed a method for automatically suggesting tags. Again in our extensive investigation, we find lots of usability issues but among them, the problem of finding and searching for the appropriate workflows emerges as a great usability problem of the system. Users, especially novice users, ask for workflow design recommendations from the experts but because of the domain-specific nature of SWfMSs, it gets difficult for them to design or implement a workflow according to their new requirements. Any software system's usability is called into question if users face trouble specifying or carrying out certain tasks and are not given the necessary resources. Therefore, to increase the usability of Galaxy, in our third study, we introduce a NLP-based workflow recommendation system where anyone can write their queries using natural language. Our system can recommend the users with the most relevant workflows in return. We develop a tool on the Galaxy platform based on the idea of the proposed method. Lastly, we believe our study findings can guide the Galaxy community to improve and extend the services according to the users' requirements. We are confident that our proposed methods can be applied to any software system to improve the usability of the system by exploring the user forums.

https://harvest.usask.ca/bitstreams/fb1a3f03-518f-4096-a1f3-0a9a1eb6d577/download