Description

This thesis is available as a BACHELOR thesis only!

Code reuse is a common practice when developing machine learning (ML) solution prototypes in Jupyter Notebooks. However, developers must still create search queries manually and search for suitable code themselves. To reduce the effort required to find reusable code, we have developed JupyRecSys, a code cell recommender system integrated as an extension into JupyterLab.

While prior work has evaluated the performance of JupyRecSys, it remains to be investigated whether developers accept its recommendations and perceive the tool as useful in practice. To address this, we will conduct a case study observing developers as they use JupyRecSys to build an ML solution prototype in a Jupyter Notebook.

The tasks of this thesis include:

• Adapting JupyRecSys to log user interactions (e.g., copying and pasting recommended cells)
• Planning and conducting the case study
• Analyzing the resulting quantitative and qualitative data

Requirements:

• Moderate Python and React skills
• Communication skills, as you will interact with multiple study participants. Ideally, you can also help recruit participants for the study.

References:

Selin Aydin, Dennis Mertens, Ouyu Xu: An Automated Evaluation Approach for Jupyter Notebook Code Cell Recommender Systems. In 12th Workshop on Quantitative Approaches to Software Quality (QuASoQ 2024): co-located with APSEC 2024, Chongqing, China, Dec 3, CEUR-WS.org, CEUR Workshop Proceedings, Vol. 3864, 4-11.