Major Milestones
| What? | Due date | Who's responsible? | Description |
| Detailed research plan | 2/12 | Team leaders | A written report that roughly lays out what's going to happen and who's going to do what. |
| Proposal presentation | 2/25 | Everyone | Part of the cloud computing speaker series. Teams are going to present their project plans. |
| Project update | 3/31 | Everyone | A concise update of what's been going on, timed approximately to coincide with the transition from phase I to phase II of the projects. |
| Final presentation | 5/7,14 | Everyone | Part of the cloud computing speaker series. Teams are going to present the results of their projects. |
| Final project report | 5/14 | Everyone | The final deliverable for the project: a high-quality research report that would be suitable for submission to a major conference or journal. This will be published as a technical report. |
Weekly Updates
At each week's project meeting, the project leader and each project member will agree on a specific deliverable due the following week. The project meeting will also include a review of the deliverable from the previous week. By Monday night (11:59pm) of each week, I expect a short email from everyone: assessment of last week's deliverables and next week's plans. I am requesting emails from everyone so that I receive multiple viewpoints on what's happening (since they serve as a basis for evaluation on my part).
Here are some examples of acceptable deliverables:
- Project x involves solving sub-problem y: sequential algorithm a is the best known method for solving y. Deliverable: come up with an algorithm a' that takes advantage of MapReduce; report back to the team with a sketch of the algorithm, pseudo-code, etc.
- Algorithm a' operates on data structure d. Deliverable: implement a Java class that provides the necessary abstractions. The implementation should pass these unit tests.
- Deliverable: implement one aspect of algorithm a'. Demonstrate correctness of the algorithm with this data.
- Deliverable: run a set of experiments to evaluate the performance algorithm a'. One metric might be speed, e.g., how much faster is it really? Or performance can be measured with respect to some task, e.g., machine translation performance, document retrieval precision, etc.
Blogging
After each invited speaker for the cloud computing speaker series, I expect a blog entry from everyone (due the following day, Thursday, at 11:59pm). Think of this as a response to the talk: don't just summarize what happened---instead, try to relate it to your own project. What's interesting or not? As is generally the case, humor is good. Of course, don't limit blog entries to just these weekly responses; if you're in the habit of blogging, write entries as often as you like. The point here is to generate some "buzz" about this cloud computing work. I will link to everyone's blog from the course homepage.
