Last updated on September 2, 2024
(This post is especially for more junior researchers)
Some forms of contributions in academic research:
- Empirical findings — you show something novel based on the data you collected and analyzed (e.g., “x increases y”). These are based on hard data and evidence, typically obtained using statistical analysis.
- Novel concepts — you introduce and define a new concept (e.g., “Remora’s dilemma”), meaning you conceptualize (i.e., label, give name to) a phenomenon that you have observed. You clearly define it and delineate its boundaries (or what is known/believed about the boundaries). These are based on observation and creative thinking.
- Method or approach — you introduce a new approach (e.g., algorithm, model) to solve a problem or task (e.g., “zig-zag strategy for hyperparameter selection”) and show that it performs better than a baseline, which is usually a more naïve, simpler way of solving the task, or it can be a state-of-the-art (SOTA) approach which represents the best known way of solving the task. These contributions are based on testing and experiments.
- Synthesis — you summarize what is known about a topic and present gaps in our body of knowledge (this is usually a systematic literature review). This is based on reading and systematically processing/coding/interpreting previous work, often both empirical and conceptual.
- Theory or framework — you try to explain a phenomenon (e.g., LLM behavior, human behavior) in a more or less generalizable way, or in the case of framework you describe a process or structure that has non-trivial degree of complexity (often done in a diagram showing relationships between different concepts). These contributions can be based on empirical findings, creative thinking, synthesis, novel concepts — basically on all other types of contributions that you leverage for a bigger “story” of how things are (or in some cases, how they could be).
You can have multiple contributions in one paper, though often a paper focuses on one type of contribution (it’s hard enough to argue for one!). Even if you wouldn’t make multiple types of contributions in one paper, it helps to think if there is also a conceptual contribution, for example, in addition to empirical ones.
All contributions need to be interesting and novel. “Interesting” means people in your topical field find them interesting. “Novel” means people in your topical field didn’t already know what you presented.
UPDATE (Sep 2, 2024—-today, stumbled across with these seven contributions, which is a more extensive version than what I wrote above): In their 2016 paper, Wobbrock and Kienz outline seven types of contributions in Human-Computer Interaction (HCI). Here’s a summary of these contributions:
- Conceptual Contributions: Introducing new ideas, theories, or frameworks that enhance understanding of HCI.
- Empirical Contributions: Providing new data or findings through experiments, surveys, or observations that inform HCI practices.
- Technical Contributions: Developing new tools, technologies, or systems that advance HCI methodologies or user experiences.
- Methodological Contributions: Offering new methods or techniques for conducting research or evaluation in HCI.
- Design Contributions: Creating novel designs or interfaces that improve user interaction or experience.
- Theoretical Contributions: Expanding or refining existing theories within HCI through critical analysis or synthesis of past research.
- Application Contributions: Demonstrating the practical application of HCI principles in real-world settings, showcasing the impact on users or communities.