This is not a comprehensive tutorial; however, it should help you if you are doing a master thesis in computational linguistics at Matfyz. Especially foreign students are sometimes unaware of some details that their Czech schoolmates may already know. This help focuses on LCT students who typically have just one academic year to write their thesis.

LCT students typically select the topic at the beginning of their second year, i.e. during October. At the same time they must find their supervisor (typically among ÚFAL professors, researchers and Ph.D. students, although the supervisor may in theory come from other departments of Charles University). Those doing their second year at another university still must find a co-supervisor at ÚFAL. They will also have to come in person and defend the thesis at Charles University, regardless what other procedures are required at their second university. Along the same lines, students doing their second year at Charles University should check with their first-year university and/or with the program coordinators (drs. Lopatková and Kuboň) what additional steps might be required by the other university and whether they need a co-supervisor there.

The student and their supervisor put together the title of the thesis and the description of the work (abstract): this will constitute the official assignment. It should be broad enough to allow deviations once the student gets their first results and realizes that the originally anticipated course of experiments is not the best one to follow. Later changes of title and topic are in theory possible but it is an administrative hassle that is better to avoid.

The supervisor puts the assignment into SIS (the information system) and also prints three hardcopies, signs it and gives it to Ms. Brdičková, the ÚFAL secretary. This must be done before the deadline announced for each academic year (typically first half of November – look for “Academic Calendar” at http://www.mff.cuni.cz/to.en/studium/obecne/harm.htm). The student office will later send back the hardcopy stamped and signed by the vicedean. Ms. Brdičková will keep it until the end of the academic year. The student will then have to pick it up and include it in one of the printed copies of the thesis (mandatory part, this copy goes to the faculty library).

When the student plans their classes for the academic year (and has them registered in SIS), they should also register for the three pseudo-courses reflecting their work on the thesis: NSZZ023 Diploma Thesis I, NSZZ024 Diploma Thesis II and NSZZ025 Diploma Thesis III. At the end of the summer semester they will (hopefully) get the credits (in SIS) from their supervisor. This is not directly connected with the fact that the thesis will be (can be) defended. The credits just reflect the fact that the student has invested significant time and effort into doing the research and preparing the thesis. The supervisor will not be able to award the credits if SIS does not know that the student wants them, i.e. if the student has not registered for these pseudo-courses! While the registration can be completed just before the submission of the thesis, it may complicate the situation because the assistance of the student office is needed, the staff may be out of office (vacation time!), you may be traveling from the other end of Europe, trying to find all the people and get all the signatures within one afternoon etc.

There are official templates and instructions, see http://www.mff.cuni.cz/to.en/studium/bcmgr/prace/, especially Guidelines for Writing a Master Thesis https://www.mff.cuni.cz/to.en/studium/bcmgr/prace/dp_pruvodce.htm (this page also summarizes links with respectíve university regulations). If you order the binding in Prague, the stores typically know how the product should look like, so they will only ask you your name and the name of the faculty. You will have to submit three hardcopies of the thesis: one for the supervisor, one for the opponent and one for the library. The hardcopy for the library must also contain the original assignment, signed and stamped from the dean's office (at this time it is probably being kept for you by Ms. Brdičková – see above). The hardcopies typically also contain a CD/DVD with the electronic version of the text (PDF) and all data and software that you created during the work, including documentation and possibly third-party software, if it is relevant and redistributable.

The submission deadline is specified in the Academic calendar https://www.mff.cuni.cz/to.en/studium/obecne/harm.htm . The hardcopies are submitted to Ms. Brdičková (secretary of ÚFAL); you must also upload the electronic version to the SIS (and all uploads are time-stamped, so any violation of the deadlines will be visible there). The deadline for the hardcopies is slightly less strict. If the official deadline is Friday, you have to upload the electronic version to SIS at or before 23:59 that day; however, bringing the hardcopies to ÚFAL at Friday evening does not make sense, there won't be anybody here anyway. If you are not able to come earlier, ask Ms. Brdičková whether you can bring the hardcopies on Monday morning.

Note that you have to finish all your courses and get necessary credits before you are allowed to submit your diploma thesis and/or registered for the state exam.

You should follow the instructions for the thesis formatting as strictly as possible; however, some slight deviations commonly occur for LCT students and are tolerated. You can have your other university's name and logo on the title page. If you have co-supervisor from the other university, you can include their name as well. For the purpose of defense at Charles University, you should show your CUNI supervisor (i.e. the one that is indicated at your CUNI official assignment) as the main one. Name your CUNI supervisor on the abstract page.

It is strongly recommended that the abstract in your thesis differs from the abstract in the official assignment. Even though in theory you could claim that you did exactly what was in the assignment, it is better to show the reviewer that it is not just a copy. Typically, the assignment is more vague because you do not know what exactly you will do after you see the results of the first experiments. In contrast, the abstract should summarize what you actually did. Even if your thesis closely matches the assignment, the abstract probably should highlight your main achievement(s) (e.g. “we were able to improve the state of the art by 50%”).

PDF-A requirements:
there is an ufal-wide installation of the verapdf tool available at /lnet/ms/projects/verapdf, with a useful wrapper script /lnet/ms/projects/verapdf/checkpdf with built-in link to the university validation profile; but make sure that the link in the wrapper script is the same that is linked from https://cuni.cz/UK-7987.html because the university updates the profile from time to time; you can also try ~vernerova/bin/checkpdf-talkative, which is a more talkative version of the same wrapper

Prepare a presentation (PDF, Microsoft PowerPoint or LibreOffice Impress). A laptop will be available (or you can use your own), with a dataprojector.

You will have only 10 minutes, which is very short time! The committee will be NLP-literate, but not necessarily specialized in your problem. Keep the introduction very short: the area I selected is XXX, the concrete problem is YYY, it is a problem because ZZZ (impact, motivation). Then another slide introducing methodology, then possibly a few interesting details from your research, then results. As with every presentation, avoid slides with too much text or too many numbers (or if you believe you need many numbers to be able to answer questions, highlight the one or two numbers that the audience should not miss). Avoid complicated formulas or anything that may take some time for the audience to grasp (remember, you will not give them much time because you will not have it). Avoid too many slides (1 minute per slide should be minimum).

Present the slides to yourself aloud, possibly several times. Try to arrange a dry-run with your supervisor. Get confident about what you are going to say. Check the timing.

There will be two written reviews of your thesis: one by your supervisor, the other by an opponent. Both reviews shall be available to you at least one week before the defense. The reviews may contain questions that you will have to answer during the defense. If so, prepare the answers. You may prepare slides to back the answers if applicable. In that case, don't include these slides in your main presentation. They don't count towards your time limit. Wait until asked, then show them (they may be part of the same presentation file, e.g. you may put them after the Thank You slide).

Typically several students defend their theses before the same committee on the same day. The defense is open to public (except for the part when the committee discusses your grade). The scenario is as follows: the chair shortly introduces you, then you get your 10 minutes to present the work. Then the supervisor reads (or summarizes) their review, possibly asks questions, you answer them. Then the same for the opponent. Then the other committee members may ask questions (yes, these will be questions you did not know about in advance), then the other guests. Then you and all other non-committee-members will be sent out of the room, the committee will negotiate, call you back and announce the verdict. Altogether it should fit within 45 minutes.