Homework - 5

Due: March 1, 2026

Metrics in CS papers

1. Make a list of all the metrics used for measurements in the 10 papers in your area of research. Categorize your list into a few logical groups. You may come up with your own criteria to group the metrics (e.g., performance metrics like latency/throughput, accuracy metrics like precision/recall, or resource-usage metrics).

2. Define each metric in a few sentences. Write these definitions so they are accessible to a computer scientist who is not an expert in your specific subfield.

3. Identify the most common and the most uncommon metrics used across your selected papers. Why do you think these specific metrics are the most and least frequently used in this area of research?

4. Choose one specific metric from your list. Explain how it is used to directly support a core claim made in one of the papers.

5. Show at least one instance where a metric is used in a way that provides an incomplete picture, introduces bias, or limits our understanding of the results. (For example: reporting an average without the variance, using a metric that ignores edge cases, or choosing a baseline that artificially inflates the perceived improvement).

Please also include References section of hw3 as the last page of your submission.

If you used AI assistance for this assignment, please write a paragraph on what you discovered about the strengths and weaknesses of AI systems to do this assignment.

Submission

Please submit your homework as a pdf 1-3 pages in length. The name of the file should be cosc6321-lastname-hw5.pdf