---
title: Title of your thesis
author: Your name
date: 2026-06-01
toc: true
toc_depth: 3
number_sections: true
chapter_break: 1
theme: newsprint
pdf_ua: true
---

# Abstract

One paragraph, no citations: what question the thesis answers, how it answers
it, and what the result is. If a reader stops after this page, this is all they
get.

# Acknowledgements

Optional. Keep it short, and remove the heading if you do not need it.

# Introduction

Research question, why it matters, and the shape of the argument. Cite with a
numbered reference, for example [@vaswani2017], and the list at the end matches.

## Research question

State it as one sentence. Everything below should serve it.

## Contribution

- What is new compared with the closest previous work.
- What it is measured against.
- What it does not cover.

# Background

Definitions and prior work. Formula support works inline ($\alpha + \beta$) and
as a display block:

$$
\mathrm{score}(q, d) = \sum_{i=1}^{n} w_i \cdot \mathrm{sim}(q_i, d)
$$

Keep variable names consistent between text and formulas — a reader who sees
`w_i` in one place and `\omega` in another has to re-learn the notation.

# Method

Describe the setup so that someone else could repeat it. A table usually reads
better than a paragraph for parameters:

| Parameter | Value | Why |
| --- | --- | --- |
| Sample size | 1,200 documents | Two weeks of submissions |
| Window | 2024-01-01 – 2024-12-31 | Matches the archive release |
| Baseline | Previous ranking model | Same training data |

## Procedure

1. Collect the documents described above.
2. Run the pipeline; the code that matters is:

```python
def score(q, d, weights):
    return sum(w * sim(q, d) for w in weights)
```

3. Record the result before looking at the evaluation set.

# Results

![Scatter plot: model score against human rating, showing agreement above 0.7](figure.png)

Replace the path with your own image and keep the alt text descriptive. Report
what the figure shows in one sentence before the figure, so the argument still
works when the image is skipped.

# Discussion

What the result means, and what would falsify it. Separate "we observed" from
"we think"; the second one belongs here, not in Results.

# Conclusion

One paragraph that answers the research question from the Introduction. No new
claims, no new citations.

# References

- Vaswani, A. et al. (2017). *Attention is all you need.* [arXiv:1706.03762](https://arxiv.org/abs/1706.03762)
- Replace these two entries with your own; the numbering follows the order of
  first citation in the text.
