International Institute of Information Technology
Hyderabad

Submitting the Thesis Evaluation Request by MS/PhD Students

Materials to be submitted

The thesis evaluation request should contain the following in hardcopy:

  1. The essential details about the thesis, such as name, roll no., advisor(s), key contributions etc., as given in the request form (PDF version, LaTeX version, MSWord version). Hints on the form are given below.

  2. The synopsis or detailed summary of the thesis. This is expected to be 2-3 pages long and can give a bit of background into the thesis as well as key references, if necessary. Write the synopsis to be read by researchers not necessarily very familiar with your problem area.

  3. A research resume of the student, highlighting the research experience, contributions, publications in the standard citation format in reverse chronological order, short abstract of the thesis and other details. This should be the resume you want to send to prospective employers for a research position. This resume with all relevant information may be used as a guide, but feel free to improve it or to use your own styles and appearance.

  4. The expert reviews received from conferences or journals for the key publications and submissions related to the thesis work. Submit the reviews even if the paper was not accepted. They provide critical view on the work from experts outside of the institution and can be a key component in quickly evaluating the research work. Submit a copy of the submitted manuscript if the paper in question is not yet formally accepted.

The process starts with the submission of the above. The complete thesis should be submitted within 2 weeks of this in a spiral bound, draft form.

Time schedules

A representative schedule for a evaluating an MS thesis is given below. The times are given as indications only.

Doctoral dissertations are evaluated more elaborately and may need 2-3 months for evaluation itself. The submission procedure to be followed is the same as given below.

Doctoral dissertation proposals are also evaluated elaborately and will need 1-2 months. The submission procedure to be followed is the same as given below.

Expectations and evaluation of theses

It is difficult to quantify the expectations from a thesis, but research work that expands the horizon of knowledge in the field is expected as part of MS/PhD theses. Critical evaluation of the current scenario, conceptual suggestions towards improving it, and concrete validation of those suggestions are all expected from a thesis, whether it is theoretical, experimental, or system-building.  The nature of the contribution expected is same for both MS and PhD, but they differ in the depth of the contribution. The institute expects the thesis to be accepted by the peer-experts in specific respective area. Peer review can be obtained in two ways:

Instructions to fill the request form

Name, Roll Number, Advisors,etc: Fill in the details. If you are advised by more than one faculty member including external ones, write all the names.

Key contributions: List the major contributions of the thesis. Rather than copying the whole of the abstract of your thesis to this space, you should try to enumerate the contributions of the work undertaken to the prior art. Every thesis should have a section/subsection on its contributions in it anyway. List those out in this space. More details are included in the thesis synopsis being submitted along with this form. The synopsis should briefly describe the problem and its importance, should mention how it is related to the past efforts, and present the key technques and results of the thesis. The synopsis is expected to be 2 pages long for MS theses and 3-4 pages long for PhD dissertations.

Example 1:

  1.   Modified the quick-sort algorithm to speed up sorting when alternate elements are odd numbers.

  2.   Provided a formal proof that the algorithm work and has better average-case complexity.

  3.   Demonstrated the advantages on sorting voter registration list containing 1 million entries.


Example 2:

  1. Pointed out the deficiencies of the proposed IEEE 16514 standard in handling networks containing between 100 and 200 nodes.

  2. Carried out extensive simulations and established that the network indeed performed suboptimally for the range of nodes.

  3. Proposed changes in representing the node-tables to reduce the problem.

  4. Communicated the suggestions to the standards committee.


Example 3:

  1. Developed a system to display weather data on a Rs 5000/- PDA.

  2. Modified memory allocation scheme on PDA so large climatic data could be represented.

  3. Modified display architecture so 24-bit colours could be mapped to 8-bit colours of PDA with minimal loss in information

  4. Implemented it on Taurus 500 low-cost PDA and tested for 25 different conditions.

Nature of the contribution: Contributions of a thesis could be largely theoretical/algorithmic (which involves the development of a new theory or algorithm that alters the state of the art in important ways) or largely experimental (which involves using experimental methods to discover an important and previously unknown characteristic or property and can be the basis for future development of better theory) or largely system building (which involves the development of a system or parts thereof that provides significant improvement in performance or functionality that wasn't available before). A good thesis would have high marks on one of these dimensions and average marks in one or more of the other. It could also be that a thesis gets very high marks in one aspect and zero in others. Average marks in 2 or 3 aspects and high marks in none are not signs of a good thesis. The contributions have to be evaluated in the context of the degree sought. Thus, a PhD dissertation is expected to have a deeper impacnt and significant results than a Masters thesis. The examples given above get high marks respectively in theory/algorithms, experimentation, and system building.

Publications: Peer reviewed publications are the best means to establish the depth and importance of one's research contributions. Theoretical, experimental, and system building efforts are also published in top journals and conferences. Journals follow elaborate review process and form the best way to establish the utility and novelty of the research for the long term. Conferences provide the means to place the results in front of the peers in a shorter time frame. They also provide the opportunity to present your work in person and to interact with the peers. There are many journals and conferences. Conferences are taking place in large numbers in our areas and tend to be of varying quality. Publications can take a few weeks to get accepted and a few months to appear at the conference. For an MS thesis, there may not be enough time to get a journal paper accepted there usually is enough time to get a feedback on a conference submission. PhD dissertations take a long time for fruition and a student has ample time to get a journal paper or two accepted. Roughly, the contributions of a Masters thesis should be equivalent to a paper in a good journal or a  domain-focussed conference. The contributions of a PhD dissertation should be equivalent to two good international journal papers: one accepted already and at least one more on significantly different work in the review process. The different publication categories could be described as follows.

International Journals are journals with international audience and reach, published by reputed publishers such as IEEE, ACM, Elsevier, Springer, etc. They have extensive review processes with the participation of the top experts in the field and can take several months to decide on a paper. They provide the stamp of the highest academic quality to the research. They are also available at the libraries of most institutions of higher learning in the world, bringing wide dissemination of the research work. Examples include the Transactions of IEEE and ACM on various topics and other journals by Elsevier, Springer, etc., that have been existing for several years.

Top conferences are conferences that are conducted and attended by many top researchers in its focus area. A conference's quality depends primarily (if not solely) on the kind of people running it and the effort they put into it. Tradition and history are perhaps the best indicators for a conference's quality. No other factor (the "international" adjective, location of the conference, sponsorship by "IEEE", low ratio of acceptance, etc) seems to have a direct relation with its quality. Pressure to publish has resulted in a mushrooming of conferences which makes the determination of quality all the more difficult. Top conferences are known to the researchers from the tradition. Examples include: IJCAI, SIGGRAPH, FOCS, SIGMOD, ICCV, STOC, ICPR, ISCA, etc.

Domain-focussed conferences are focussed on an area and is attended by a number of experts who work in it. These could be a national or international conference or one held at the same place every year. They follow a good review process and exercise quality control. They are, again, distinguished by its tradition, the kind of people involved with them and the effort they put in and not by being focussed narrowly. Examples of these include SPCOM, FSTTCS, ICON, ICVGIP, etc.

National Journals have limited scope and reach but with a review process involving local experts. These provide archival but probably do not reach many universities outside the country. Examples include journals by IETE, CSI, etc.

Dissemination conferences are meant for disseminating the results of the research to a wide and often widely focussed audience. Their areas of interest are wide and the review process is often inclusionary. Many such conferences have been taking place in India of late. These conferences are helpful in meeting a large cross-section of researchers from the country and establishing a proof of publication. Their value as publications in one's resume may be low in general. Examples of these include CIT, IICAI, ADCOM, ICSCI, etc.

Other conferences abound too which are conducted by institutions occasionally and follow minimal review etc. They may be useful in meeting some of the researchers but may not help in interacting with the top researchers. These include many workshops conducted by institutions, seminars in connection with specific events, etc. Their value as publications in one's resume is zero. Some of them, like the international multiconference that seemed to be held in fun places every year, appear to be commercial ventures whose academic value may be negative!

System Building:Building large, complex, and useful systems involve serious effort and learning. MS/PhD students are encouraged build such systems. However,  innovation and critical analysis are essential for a research thesis. A system-building thesis should provide new insight into the system, the process of its design, conceptualization, and implementation, and report the performance of the system on all kinds of test cases. A mere study of the the problem and the description of the system that was built does not qualify for even an MS thesis unless accompanied by the above elements. Most conferences, if not all of them, accept system-building papers. Publishing them is the surest way to prove the starnding of your work in the community to the world and to the committee.

Patents: Research work can generate patents on innovative aspects of the research or system. These are very valuable. Typically, patents take years to grant but should be applied for at the earliest opportunity and legally before any open publication of the work appears.

Demonstrations: Demonstration of parts or whole of the research at important forums is a way to disseminate the activity as well as to get recognition for it. These could be at different academic/technical events or forums. at IIIT's R&D Showcase, etc. Some conferences have official demonstrations to which you can submit your system. These are reviewed by peers and are given time/space at the conference. They get attention from peers and is a good way to publicize your work as well as to get constructive opinion on it.

Software packages: Other forms of output of research could be useful software tools and packages made available to the community either in the open source model or only as executables. These packages may gain widespread acceptance bringing recognition to the work and individuals behind it. Such packages could be directed at specific groups or could be of wide appeal. Naturally, one can make many packages available for download, but nobody will use them unless they are useful. The real achievement is in getting people to use it, appreciate it, and give you feedback on it; the number of downloads is not the achievement.

PJN. May 2005; Jul 2006; Sep 09