FAQs for Prospective Graduate Students

Answered by Prof Amit Sheth

  1. Should I do an MS or a PhD ?

    It really depends on what you want to do. In the increasingly flat world with 3 billion new capitalists, competition is tremendous. I frequently hear company managers say, "we do global sourcing of technical talent". Those with MS degrees will face a lot more competition than those with PhDs. If you are primarily a programmer or doing structured or well defined tasks, someone else can probably do it at lower cost and just as efficiently, so such a work will likely be "packaged" and outsourced.  At the BS/BE and MS/ME levels, the emphasis is on learning fundamentals, technologies, and skills. These things can be taught to hard working students by a few good teachers. People who can innovate, lead, communicate, and motivate are much rarer. These are the things PhDs are expected to do, and the longer time horizon combined with one-on-one advising in a Gurukul-type environment prepares those with reasonable intellect and excellent work ethics to do them. Some careers (such as professorships) are reserved for PhDs, and a growing number of high-end companies prefer to hire PhDs for jobs in the US. That said, the PhD is not for everyone, and there can be many disadvantages, including the investment of significant time and hard work and the delay in earning "real money" If one is primarily interested in development, a PhD is often not necessary. The programs have very different purposes. The PhD is about becoming an independent researcher, thinker, performer, fully capable of becoming a peer of one’s adviser. Students who wish to pursue MS are indeed welcome to join the Center and alumni who completed MS have been very successful. However, typically an incoming foreign student will not likely get research assistantship during the first year (for TAship, contact the department); US applicants do have significant likelihood for being funded. Most or all full time PhD students are funded. A word of caution: if you are evaluating whether to do an MS or a PhD, do not rely on the advice of those who have an MS; also ask the advice of those who have done a PhD.

  2. How do I choose the right program for me?

    In my view, the answer to this is not as easy as looking up the best-ranked university in the US News and World Report list. Ranking matters most for the BS degree, a bit less for the MS, and perhaps the least for the PhD. The higher the degree, the more individualized the education and experience. Let us primarily focus on the choice of PhD program Here are some issues you may want to consider:

    Ranking is complicated: the question is, ranking by whom? One of the last authoritative rankings of computer science was done in 1993/1994 by CRA/National Research Council. But the world has changed a lot since then. A number of very important fields of computer science have emerged; the Web was not a major factor then, and significant research groups have arisen in universities that were not ranked high in those days. For more recent rankings, rely on US News and World Report. But ranking an entire university or even an entire department is a complex issue (here is one example of criticism). How can you weigh the value of a large department versus a small department, a department with several areas of strength versus a department that has one of the best research groups in the world in your area of interest but does not have signicant strength in other areas, and so on?

    Also ask, ranking of what? University, department, research center/group), adviser in the field of choice, funding and resources? Remember that when you do a PhD, you hope to become an expert with a potential to lead in a specific sub-area or even topic within a sub-area of computer science (e.g., in ontology mapping topic of Semantic Web, an important new multidisciplinary area). How do you factor in the opportunity to work with an inspiring faculty adviser, a work environment with a group of excellent colleagues, multidisciplinary research, collaborations with industry, success of alumni, resources, etc.? Naturally the easiest choice would be to go for the best-ranked computer science program (department) that also has one of the best research groups and a top notch faculty advisor in your areas of interest, gives you full financial assistantships (stipend plus tuition waiver), and is in a geographic area of your preference. If you do not have an idea what area interests you, it would be best to choose a higher-ranked department and try to work in an area that the department is good at, but perhaps you are not ready to commit to a PhD program in that case! And if you do not have the best of all, how would you choose from one of these (and other similar alternatives)?

    • Well-known department but no significant strength in your area of interest? Or no financial assistantship?
    • Not a very well known department but a world-class research group led by well-known faculty in the area of your interest and lots of resources with an offer of financial assistantship? Ask this simple question- are you going to learn more from a university, a department, or your advisor?

    In 1980, applying for graduate studies, I went to IIT library and found two professors in my area of interest (which happened to be database management): at the University of Wisconsin-Madison and Ohio State University. I then applied to those universities.

  3. How do you find out whether a faculty or research group is good in a given area?

    Key factors to consider are: funding, publication success, citation record, leadership position, number of researchers and collaborations.

    • Peer-reviewed proposal funding from federal sources, especially NSF and NIH (the most competitive), is a good indication of research strength (remember that faculty from top and lesser-ranked universities compete, so whoever gets such grants has proved his or her worth). Many NSF projects have a success rate of around 10%; a typical successful NIH proposals needs to be ranked in top 12% to 15% to be funded. DARPA and DoD are also competitive but are often more dependent on timeliness of research area, development skills, contacts, and collaborations.
    • Check out the proceedings of top conferences in your area of interest and see which places are publishing interesting papers. Broadly speaking, a conference that selects fewer than one in three submissions is reasonably good, while a conference that selects fewer than one in six is probably among the top conferences in that area.
    • Look at the citation record of the faculty who lead a research group. See whether faculty and students have won best paper awards, and see if the faculty is a leader in the field (eg, has your potential faculty advisor served as the Program Committee chair of a major conference in your area of interest?). A simple way to start is to use http://scholar.google.com, look at the h-index of the faculty member, or see whether he or she is a fellow of IEEE or ACM or a member of the National Academy of Science (many such faculty have h-index of over 30, although there are exceptions). For a more detailed look you can use a tool from Harzing.com.
    • Many new areas are interdisciplinary. For example, Semantic Web involves AI (and a number of related subareas including natural language processing, machine learning, knowledge representation, and description logics), text mining and information retrieval, database and Web data management, distributed computing (such as semantic web services), and so on. So having multiple faculty members with complementary skills is very helpful for students. Such research groups can collectively offer a large number of specialized courses in closely related areas to enable a PhD student to get significant strength in that area. Many systems-oriented research areas require contributions from a number of researchers (in top conferences, you will often see papers with three or more authors since systems- oriented studies require access to real data, significant prototyping, and evaluation). In these cases, having a group of collaborating senior and junior students make for an exciting and vibrant work environment. In some areas, collaboration with scientists and industry can be highly valuable. It is also an important validation of the influence of such a research group.
    • Check out accomplishments of current students as well as alumni.

    Once you are serious about a program (be sure to study in detail the web pages of the faculty members, center/lab, current PhD students in the group, and department), you can also consider contacting a faculty (it is best not to send email to many faculty members in the same department) or current PhD students.

  4. How do you choose a faculty advisor?

    There is more to choosing an advisor than simply choosing a faculty member who is most senior, has more funding or has a stellar credentials. A junior faculty member who may not yet be very well known can be an excellent advisor. A key requirement is that the advisor must be able to guide you (and preferably provide networking and other opportunities) in your area of interest. Assuming that you also need financial assistance, research funding can also be an important consideration. While these considerations are necessary, they are not sufficient. Emotional compatibility is important - remember you will work with this person for at least three years, and this person can provide substantial boost to your career and success. Ask if this is the person who you can look up to? Be sure to take at least one course (preferably a seminar or research oriented course) under your potential advisor before you commit. Try to join a research project guided by the potential advisor and get a sense of his or her style, expectations, availability, modes of interactions, and so on. Most PhD students get Teaching Assistantship during their first year or year and half of studies, so there is ample time to learn more about the potential faculty advisor before asking for research assistantship and committing. In this context, it helps if the department has more than one faculty in areas of your interest.

Here are some facts about the Kno.e.sis Center that correspond to some of the considerations above as of May 2008: Kno.e.sis Center has five faculty members (a sixth will join in Fall 2008 and one additional position is open). This allows us to offer a wide variety of courses. The Center’s faculty advise about 20 PhD students. All full-time PhD students are fully funded (receive stipends as well as full tuition waivers). A few undergrads also receive partial funding and participate in research. Prof. Sheth’s advisees intern at the best places (Microsoft Research, IBM Research, HP Research, Yahoo!, Oracle, Amazon, National Library of Medicine, etc.) and are employed at top companies (including Microsoft, IBM, Accenture Labs, Yahoo!, SAP, Oracle, and exciting startups) or universities (e.g., NCSU). Within a five-month period, the Center received competitive research grants and gifts from IBM, Research, Google, HP Research and Microsoft Research. Research is currently funded by competitive grants from NIH, NSF, AFRL, and several other sources. Our collaborators include CCHMC, Boonshoft School of Medicine, NCBO at Stanford University, the Tarleton lab at UGA, CCRC at UGA, eBiquity at UMBC, IBM, HP, NLM, AFRL and a few small companies. We are actively involved in several W3C activities and are its official member. Prof. Sheth’s h-index is 54, and Kno.e.sis faculty members have several best papers in conferences in their areas.

Read even more:

 
The content and opinions expressed on this Web page do not necessarily reflect the views of nor are they endorsed by the Wright State University
377 Joshi Research Center, 3640 Colonel Glenn Highway, Dayton, OH 45435 (+1 937-775-5217)