Computer Science Master's Thesis/Project Guidelines

Project

The goal of a project is to develop some application of interest to you. Most projects will involve the development of a large piece of software, though you are not restricted to purely software projects (e.g., embedded systems are fine). Most projects will have many of the following characteristics:

• A level of effort comparable to what is required in a 3 credit course. Thus, it will be substantially larger than any single project you have written in any course.
  
• Regular meetings with your mentor, as decided by you and your mentor
• Interfaces with a front end GUI
• Interfaces with a back end package (e.g., databases, domain specific solvers)
• Documentation for the user, and if applicable developers
• Adherence to standard software engineering (SWE) guidelines and usage of SWE tools

Although the goal of a project is not to do research, some project tasks might involve reading research papers and implementing ideas based on these papers.

Picking a topic/mentor:

There is no single way to pick a topic, and it ultimately comes down to something that you and your mentor agree on. Some models are:

• I took class X, and would like to do further work on topic T that we learned in class: This is the best (and most common) model. Talk to the professor and discuss your interests. Do it soon after the semester when everything is still fresh!
  
• I would like to write this cool app to do Y: talk to a professor who is likely to have an interest in the area and discuss. You might refine your basic idea at this point. If you're not sure who would be an appropriate mentor for that area, ask the masters program director for suggestions.
  
• I am interested in learning Y: Talk to the masters program director for suggestions on who might be an appropriate mentor.

You and your mentor should decide upon the 'rules' for the project. While it is ultimately decided between you and your mentor, we recommend that you write a short proposal with the following:

• Statement of problem, including brief overview/comparison of related work
  
• Your approach to implementing it, including the following:
  • Platforms: Which languages/compilers, operating systems, hardware will your project work on?
    
  • Input format and content
  • Software components that you will be writing, and algorithms for these if applicable
  • External components that you will be interfacing to
  • Output format and content
    
• Risk plan: what are the risks in your approach that might make your project fail, and how will you address each of these risks (e.g., alternate approaches). The risks could be both engineering (e.g., issues integrating external systems) and science (e.g., scalability)
  
• List of deliverables: this includes software, documentation, reports
  
• Management plan: when/how will you meet with your mentor? If working with others, how will you interface with them?
  
• Anything else appropriate to your topic!

Timeline:

• [-1]: final version of proposal alluded to above completed and agreed to by you and your mentor (form to register).
  
• [0-12]: work with mentor to perform project
  
• [13-14]: submit final project to mentor, make final revisions as needed
  
• [15] Administrative: Submit project to mentor and director, mentor submits grade, mentor confirms approval to director

The deadline for everything is the end of semester. If you are taking the project in your semester of graduation, the administrative deadline is particularly tight.

Thesis

The goal of a thesis is to do independent research in some [possibly interdisciplinary] subarea of computer science. It should be in the area of your mentor, though some great theses might integrate your background with your mentor's area; for example solving problem X that you have experience in, using techniques that your mentor does research on.

The final deliverable for the thesis is a document following guidelines by the School of Arts & Sciences; in computer science, resulting code/experiments is also often placed in a public repository. These theses are stored in the Hunter library and accessible around the world, so you may wish to see some other theses.

Picking a topic/mentor:

Some common models are:

• If you already know which professors do work in the area, talk to them and identify a topic. Most research faculty teach courses in their research areas, and they may ask you to take their course to gain the needed background. This means you should take any elective courses when offered, as they might not be offered again in time. For some topics, you may also need to do some preliminary reading and/or work prior to proposing a topic.
  
• If you have an area of interest but don't know who is an appropriate mentor, talk to the masters program director for suggestion.
• As with the project, thesis topics often arise from elective courses. Take elective courses that you can see as potential topics, and talk to the professors if you have an interest.
  

• Statement of problem
• Literature review: other approaches (including their shortcomings and benefits)
  
• Proposed approach: how do you solve the problem, how is it different from what has been done, why is it likely to benefit
• Risks: What can go wrong with your approach, and how do you propose to address it it happens?
• Evaluation: How will you evaluate the success of your approach?
  
• Anything else appropriate to your topic!

You also want to come to agreement with your mentor on how/when you will interact (possibly in the context of a larger group). These interactions typically take the forms of  individual meetings, progress reports, research group meetings, meetings to present/discuss research papers, etc. (all done on a regular schedule).

Thesis Process and Timeline:

Although your mentor is the primary overseer of the thesis, you will need to form a committee composed of:

1. Mentor: Primary overseer of the thesis
   
2. Reader: As the title suggests, this person will be a secondary reader of your thesis. If your thesis spans two research areas, you should make sure to pick the mentor and reader from the two different areas. You may also choose to have >1 reader. 
   
3. Director: General overseer. If the program director is a mentor or reader, an alternate acting director will be selected.
   

The administrative process is:

1. Write proposal
   
2. Perform research
3. Write thesis; get preliminary approval by mentor
   
4. Submit thesis to committee, schedule defense at least 2 weeks later.
   
5. Defense: present thesis to audience (committee + anybody interested in attending). The committee decision will be either accept, accept with minor revisions (most common if you have been working closely with mentor), accept with major revisions, or reject.
   
6. Revise thesis as per above and get approval by committee. 
7. Submit thesis to Hunter with committee approval
   

It takes time to go through all these stages! The deadline for submitting final theses is set each year by the school of Arts & Sciences (see guidelines). A sample timeline for a September-May thesis might be:

• [Aug 20] perform preliminary work, write proposal under guidance of mentor (form to register)
  
• [Oct 30] Finish version 1 of research, identify problems
  
• [Feb 15] Repeat (multiple iterations): fix problems by trying other approaches, reading research papers solving similar problems, etc; implement fixes and identify new problems
• [Feb 25] Generate final experimental data to show results of evaluation (which you have been doing all along above)
  
• [Mar 20] Finish writing thesis, making sure to keep regular contact with mentor. You have of course been writing little parts all along, and your excellent proposal can be adapted to be the first part of your thesis.
  
• [Apr 01] Submit thesis to committee and schedule defense
  
• [Apr 15] Defend thesis
  
• [Apr 25] Submit revised thesis to Hunter
• [Apr 26] Well-earned celebration
  

The above dates are just a general guideline and will of course be decided by you/mentor (and also vary widely by area); however, the final tasks have strict date constraints.

Unlike PhD theses, masters theses don't require published papers; however, you should strive to publish a paper especially if you plan on entering a PhD program. This imposes additional scheduling constraints, as most good CS conferences have a 2 month gap between submission and notification of acceptance - you can do the math.