The goal of the research performed in this thesis is clear; ideally by stating research questions or hypotheses.
I provided all information that is required such that another student of the same program would be able to understand the thesis without additional literature.
Sufficient information on related methods is provided such that the decision process which lead to the actual choice of methods is understandable.
All work has been described which constitutes are starting point for my own work.
The conceptual and implementation level are clearly separated (don’t mix code snippets with mathematical explanations). Typically, implementation details are not required, if you want to write about them, make a section “Implementation Details”.
It is clearly understandable what my own methods are, and where I ‘just’ reused what others prepared.
The experiments are designed to answer the research questions. Every experiment is clearly explained.
Tables and results are prepared such that the reader can easily understand how they help to get an answer to a research question.
A reader would be able to repeat all experiments by themselves with the available information.
The reader is not left alone with interpreting the results. I talk them through each table and depiction with text.
Every entry is complete; all information is available and it is easy for a reader to understand to which source an entry corresponds.
ArXiv papers (and similar) are only cited if there is no published version of the paper (e.g. in ACL Anthology) available.
All entries are consistent; author’s first names are always (or never) abbreviated; abbreviation always have a ‘.’ behind them (or never), conferences are called what they are called in the official publication (and always the same; but that might change from year to year), journals have numbers and volumes; page numbers are included when available.
I answer the research questions posed in the introduction.
I summarize the main take home messages
All abbreviations are introduced.
Abbreviations are avoided if not used really very frequently.
All work that has been used is cited.
Where somebody’s work has been used to write some text, it is mentioned.
I really checked carefully that I cite peer-reviewed (quality assured) work, I avoided to use blog post or Wikipedia as much as possible (because they can be of questionable quality, because they are not reviewed).
Sections of the same kind are of similar length/degree of detailness (e.g., if SVMs and NN have been used, both received the same attention)
I checked that I am allowed to re-use depictions (with citations); otherwise I have redrawn them and cited the original source.
I checked the book
for free from the university network) and the way how I
write. Particularly relevant are Section 1, 2, 3, and 5.