Research Project¶
The semester project is the main graduate/PhD component of the course. It should produce a research artifact, not only a software demo.
Acceptable project modes¶
Replication¶
Reproduce a meaningful subset of a published design-analysis or automated-refactoring technique and investigate whether the result holds on the same or newer subject systems.
Extension¶
Re-implement a baseline and add one clearly motivated extension: a new representation, objective, ranking model, constraint, language, or evaluation dimension.
New prototype¶
Develop a small original method addressing a narrow, testable research question.
Expected outputs¶
- proposal;
- version-controlled implementation;
- reproducible dataset/subject-system acquisition;
- experiment configuration;
- raw + analyzed results;
- research report (conference-paper style);
- final technical presentation/demo;
- replication package.
Scope test¶
A good project question can usually be written as:
Under condition C, does technique X improve/identify/recommend Y compared with baseline B according to measures M?
Avoid topics whose only question is “Can we build an app that uses an LLM?”