User:TimBorgNetzWerk/sandbox/Thesis Writing Guide

Moved to Thesis Writing Guide

This guide was moved to Thesis Writing Guide

Meta-Comment

As a new Wikibooks user, I very much welcome any form of feedback. Collaboration is also appreciated, if possible on User_talk:TimBorgNetzWerk/sandbox/Thesis_Writing_Guide

Abstract

This is an emerging Wikibook to guide through the Bachelor/Master/PhD-Thesis process, with a focus on writing, but also on which work and steps need to be done to get to a good, written thesis.

Specifics

How your specific thesis should and will look like depends on

• Region (e.g. Lower Saxony, Germany)
• Institution (e.g. Leibniz Universität Hannover)
• Faculty (e.g. Faculty of Electrical Engineering and Computer Science)
• (Sub)department (e.g. Data Science and Digital Libraries)
• Evaluator/Supervisor (e.g. your Professor)
• Time (e.g. 2024)

This guide attempts to unify all universal guidelines, while individual rules are separeted into modules:

• Data Science and Digital Libraries at Leibnitz Universität Hannover

Table of Contents

1. Introduction
2. Organization
3. Setup
4. Work
5. Writing

Overview

Introduce the Guide.

Setup

This begins with a establishing a basic setup of Reference Management and Writing Environment . It concludes with practical Lessons Learned: Setup , contributed by individual readers who may or may not be co-authors of the overal guide.

Reference Mangament

TL;DR: On day 0, setup Zotero and bookmark everything you read.

Since references are often an afterthought, but are actually the most valuable currency in science, this point is raised first.

Some rules:

1. Read != saved.
   You won't be using every source you read, but unless you can find them again, you can't even use them. There is nothing worse than knowing you read the essential point you wrote two pages about already, but can't find the right source anymore and have to rewrite or delete this work again.
2. You won't save too many.
   Your thesis will likely have 100 or 200 references at most. Even if you save everything you come across, you will end with maybe 10.000 references in your library. Searching this filtered library for one specific source you read is always much, much easier then having to find it again on the entire web. The average user will save their 350 papers, documents and websites in a Zotero library, cite maybe 50 of them.

Writing Environment

TL;DR: Use Overleaf, preferably one hosted by your university.

Proceed as follows:

1. Check if your supervisor has a template.
   This allows you to don't mind some formal requirements that are already preconfigured in the template.
2. Set up a shared acces working environemnt, preferably over Overleaf and shared acces via invite/link, or using a cloud service such as OneDrive or Google Drive.
   This makes reviewing and providing feedback easier.
3. Set up a backup schedule.
   Back up your thesis every few weeks to a second service, be that a local PC or a different cloud. You can recover two weeks of lost progress much easier than your entire thesis if your primary service is offline when you need it most, like in the final weeks before submission.

If those 3 points are checked, e.g. if you have setup your Overleaf project using the university cloud service and LaTeX template, you can proceed to the next section. If you're looking for more details on particular choices, continue reading this one.

LaTeX

LaTeX

Trend towards Overleaf

Either Overleaf (online) or Visual Studio Code (local):

• Overleaf
• Visual Studio Code + a TeX distribution
  • MikTex + Strawberry Perl
    See:
    • Install Perl. You can use Strawberry Perl in Windows.
    • If you don't have administrator privileges you can install the portable version and add the path to the executable to the PATH environment variable.
    • Install MikTeX. The creator of LaTeX Workshop suggests to use TeX Live instead because it already comes with Perl and you could skip one step in this list. The disadvantages of using TeX Live instead of MikTeX are more (see here).
  • TexLive + LaTeX Workshop

Word

Working with Word in a scientific environment is not recommended over LaTeX, but sometimes the right choice in certain circumstances, particularly when your field or supervisor provide better support for word. This guide recommends giving LaTeX a try and does not further elaborate on Word. If you are a supervisor with Word experience, feel free to supplement this section.

Writing assistance

TL;DR: When in doubt, don't trust (AI) tools.

However, a few tools are recommended:

• Grammarly, especially as a browser plugin.
  The freemium version suffices to provide some feedback for the broadest mistakes. It does not replace careful proofreading and consideration, and should not always be trusted, as it can make mistakes as well.
• LanguageTool, especially as a browser plugin.
  Works in tandem with Grammarly, does roughly the same.

Translation tools are also worth mentioning, for various use-cases. Primarily, most scientific writing is in English, which might not be your first language.

• DeepL, for translating.
  Keep in mind that you should never translate large portions of text just for copy-paste, but rather create your own translation and check with DeepL where you are uncertain. DeepL is also especially helpful to find synonyms or rephrase a sentence.
• Linguee, for translating individual words and see in which context they are used.

ChatGPT

A dedicated section would be worth to discuss ChatGPT, or any LLM for that matter. In short, this guide does not recommend using ChatGPT in general. However, it is being used, see the "delve" phenomena. As such, we need to provide some ground rules:

• Raise this point to your supervisor and get some feedback. Many faculties have their own rules and regulations, some particularly even for ChatGPT or LLMs in general, so just ask.
• When you use ChatGPT in any extent going beyond just finding selected reformulations or synonyms, like e.g. with any tool mentioned above, disclaim that you used it.
• Be aware that you want to be the author of your work. Every bit ChatGPT writes for you is like a colleague writing for you, which would be fraudulent in every other exam. Handle it with care and use it sparingly, and as mentioned above, when in doubt, don't use it at all.

Lessons Learned: Setup

Dear reader, this is your time to shine!

Contribute your own lessons learned, add helpful hints and mistakes made so that others may learn from it.
Just click "edit" and add your own subsection!

Don't setup to late or to long

It's best to start on your setup even before you actually start your thesis. Having your Google Drive, Zotero and Overleaf prepared a week or even a month in advance can help to focus on your task at hand and not stress to much about how 3 new applications work alongside solving an actual scientific problem.

Writing

Background/Related Work

Anything that could be interesting for this thesis guide. Even if it might not end up being used, it should be here.

Approach

How-To knowledge. This is the "Guide" part of the thesis guide.

Present one example workflow of how a general thesis could look like.

Evaluation

Not sure what to do here, yet

Survey

Design

1. NASA TLX Score https://en.wikipedia.org/wiki/NASA-TLX
2. Scales of Measurement - Nominal, Ordinal, Interval, & Ratio Scale Data https://www.questionpro.com/blog/nominal-ordinal-interval-ratio/ https://www.youtube.com/watch?v=LuBD49SFpWs
3. Likert Scale https://en.wikipedia.org/wiki/Likert_scale

Usability

1. https://www.ueq-online.org/
2. https://en.ryte.com/wiki/System_Usability_Scale/
3. https://www.invespcro.com/blog/usability-metrics/

Results

Statistical Analysis

1. Precision and Recall https://en.wikipedia.org/wiki/Precision_and_recall Precision, Recall, Accuracy, F1 Score
2. Inter-rater reliability https://en.wikipedia.org/wiki/Inter-rater_reliability
3. Fleiss' kappa https://en.wikipedia.org/wiki/Fleiss%27_kappa https://datatab.de/tutorial/fleiss-kappa
4. Krippendorff's alpha https://en.wikipedia.org/wiki/Krippendorff%27s_alpha https://real-statistics.com/reliability/interrater-reliability/krippendorffs-alpha/krippendorffs-alpha-basic-concepts/