{% extends "layout.html" %}

{% block content %}
<div class="topmargin">
    <div class="about">
        <h4>What is this?</h4>
        <p>
            You know the drill. Wake up, coffee, arXiv. Same sections, same keywords, same authors &mdash; every single day.
            Speed-the-arxiv exists so you can do that in seconds instead of minutes, and maybe enjoy the coffee a bit more.
        </p>
        <p>
            Define your searches in <code>.yaml</code> configs (or create them right from the UI), hit a button, and get a clean page
            of results with collapsible abstracts, LaTeX rendering, SciRate scores, and one-click BibTeX &mdash; no copy-pasting DOIs into
            twelve different tabs.
        </p>

        <h4>What it does</h4>
        <ul>
            <li><strong>Configurable searches</strong> &mdash; sections, keywords, authors, logical operators, date ranges, all in tidy <code>.yaml</code> files you can create, edit, duplicate, and delete from the browser.</li>
            <li><strong>Dark &amp; light theme</strong> &mdash; because not everyone wants to stare into the void (or into the sun).</li>
            <li><strong>Sort, filter, paginate</strong> &mdash; sort by date, SciRate score, or title. Filter results in real time. 25 papers per page so your browser doesn't melt.</li>
            <li><strong>SciRate scores</strong> &mdash; fetched asynchronously with concurrency limits and timeouts, so you get scores without hammering the internet.</li>
            <li><strong>BibTeX on demand</strong> &mdash; auto-generated for every paper, copy to clipboard with one click. Need a specific DOI? The Crossref lookup on the landing page has you covered.</li>
            <li><strong>Keyword &amp; author highlighting</strong> &mdash; toggle-able, so the important bits jump out at you.</li>
            <li><strong>MathJax</strong> &mdash; LaTeX in titles and abstracts rendered properly, as nature intended.</li>
            <li><strong>Caching</strong> &mdash; results are cached locally with a visible age indicator so you know exactly how stale your data is.</li>
            <li><strong>Robust error handling</strong> &mdash; timeouts, API failures, and unreachable endpoints all get friendly messages instead of blank screens.</li>
        </ul>

        <h4>Why no ML?</h4>
        <p>
            I thought about it. I even started the branch. Then I dropped it.
            The reason is simple: a recommendation engine narrows your reading to papers that look like what you already read.
            That's great for efficiency, terrible for serendipity.
            Speed-the-arxiv casts a wider net on purpose &mdash; you set the parameters, the search distills the field,
            and every now and then something unexpected catches your eye. That's the good stuff.
        </p>

        <h4>Links</h4>
            <p>speed-the-arxiv &rarr; <a href="https://github.com/mekise/speed-the-arxiv/" target="_blank">github</a></p>
            <p>Stefano Scali &rarr;
                <a href="mailto:scali.stefano@gmail.com">scali.stefano@gmail.com</a> |
                <a href="https://www.stefanoscali.xyz/" target="_blank">website</a> |
                <a href="https://scholar.google.com/citations?user=j7dg-xwAAAAJ&hl=en" target="_blank">scholar</a> |
                <a href="https://github.com/mekise" target="_blank">github</a> |
                <a href="https://www.linkedin.com/in/stefano-scali/" target="_blank">linkedin</a>
            </p>
    </div>
</div>
{% endblock %}