QuikiWiki

Inspiration

playing wikirace in middle school history class

What it does

uses NLP and A* to find shortest path between 2 wikipedia pages

How we built it

python, spaCy for NLP

Challenges we ran into

heuristic took a long time to compute, the branching factor was super high

Accomplishments that we're proud of

we kept improving over the night, but got to a point where it wasn't feasible to keep cutting it down. we are proud that we made improvements consistently and learned a lot about AI and web scraping.

What we learned

We learned about string similarity algorithms like Jaccard and Cosine, we learned about AI models for natural language processing like pandas and spaCy. We did a lot of trial and error for designing our algorithms, so we got our hands dirty

What's next for QuikiWiki

• train an AI model to self-improve the heuristic and find paths more efficiently
• implementing different algorithms like bidirectional DFS and Greedy searches to test which algorithms work best
• designing new heuristics that are much faster
• making a user interface so that "racing"(getting to a page in less steps) is interactive for a user.

Built With

• python
• pywikibot
• spacy