Cal-Hacks-2025

The Word Tree

With online research and topic exploration becoming increasingly overwhelming, The Word Tree helps users visualize how ideas and concepts connect — instantly generating a branching semantic tree from any given word, topic, or idea.

Usage

git clone https://github.com/Jaaaayden/Cal-Hacks-2025
cd The Word Tree
pip install -r requirements.txt

Download the dataset below too:

https://wormhole.app/6Yjld3#_RVFb2UzWscjAIYbR0h6UQ

How it Works

• Word Embeddings: Uses gensim’s pretrained Word2Vec model to find contextually similar words.

• Linguistic Filtering: Lemmatization + morphological normalization reduces duplicates (e.g., college, colleges, collegiate).

• Tree Construction: Balances similarity between the user input and previous node to create a meaningful hierarchy.

• Async Handling: Built using Flask[async] — ensures smooth asynchronous generation and JSON file I/O.

Challenges

• Balancing semantic weighting between nodes (to avoid random or overly repetitive branches).

• Cleaning unfiltered datasets for stable, human-readable output.

What's next

• Integrating topic definitions and external resources for each node.

• Applying the system to domain-specific datasets (e.g., medicine, law, and physics -- although solid already).

• Displaying numerical similarity scores when hovering over branches.

• Full deployment as a public-facing web app.

Thank you to:

Ethan Zhang

• Project Manager/Lead

Justin Sato

• Back-end Developer

Yuchan Chung

• Front-end Developer