noteorious brings tab complete to note-taking in a whole new way. It allows for faster and better note-taking than ever before, with its customized AI tab-complete suggestions and LaTex integration.
noteorious starts by grabbing the HTML from any Wikipedia page of your choice with requests. Then, it uses BeautifulSoup4 to scrape all the text from the article, and all the text from every article linked in the primary page. It builds a Markov Chain from this data, with the original article weighted, and puts the model into a SQLite database with the peewee ORM. Finally, Python, Flask, and HTML5 are used to transport this data to your browser, where the predictions are updated with each keystroke into an attractive and fully-featured LaTex-enabled WYSIWYG editor.
noteorious predicts your next word based on the last couple of words you typed, and actively improves its predictions with every letter you type. A well-trained instance can save some users more than half their keystrokes, allowing users to keep up with lectures while taking better notes and absorbing more material. noteorious' breadth of knowledge from URL scraping is extremely beneficial to accurate prediction; with this setup, noteorious's predictive model is heavily biased towards words that you are more likely to use, but is also prepared to help you type faster on a range of subjects related to that day's lecture.
noteorious has a wide range of features to make the user experience great, including Bold/Italic/Underline format options, automatic bullet points for making outlines, and even easy LaTex integration. It generally takes under 5 minutes to get a decent grasp on a new subject's vocabulary, even on older hardware. Its HTML + LaTex output is easy to export without losing formatting.
Plans for further work include automatic PDF export, more text features, downloading and reloading learned subjects, cleaner Wikipedia import, and a decrease in training time.