AN INDEPENDENT PROJECT FOR CURIOUS READERS

Good books.
Open possibilities.

LexiMind is a small place to discover books through the things that interest you, and the stories you already love.

It started as an undergraduate research project exploring how a single model could learn to summarise text and recognise its topics and emotions. The original idea was simple: help people find a book that speaks to them. This library returns to that idea.

Follow an idea.

Search for a title, an author, or a subject. Browse a genre. Choose “More like this” on a book that catches your eye. Recommendations look for connections in book descriptions, subjects, and genres, with room for a different author or a new direction.

These are suggestions from a small catalogue, not a complete map of what you might love. Every book links to its source so you can explore further.

A little shelf of your own.

Save books to a reading list, heart your favourites to shape your recommendations, and hide books you’re not interested in. Hidden books can always be restored from your shelf.

Your shelf is stored only in this browser. There is no account, and it won’t sync between devices. Clearing your browser’s site data will clear your shelf. You can export a backup from My shelf and import it in another browser. Importing adds to the books you already saved.

Sources you can follow.

Book records link to Open Library. Descriptions and available covers come from the source records; when a cover is unavailable, a simple title card takes its place. This is an early, curated catalogue, and coverage will grow over time.

The research continues.

The question behind LexiMind is still interesting: can learning several language tasks together help a model understand books better? The website is a useful place to explore what better book discovery should feel like. The research models are not currently used to assign emotions or make recommendations here.