Improve cofounder matching algorithm #88
Description
Current Behaviour: As far as my understanding goes, the current cofounder matching algorithm works by comparing the ratio of tweets to GitHub commits. It tries to find a cofounder who is the opposite of you. For example, if you commit a lot of code, it matches you with someone who tweets a lot, and vice versa.
Proposed Improvement: Instead of matching cofounders based on opposite activities, the algorithm could be improved by matching cofounders who tweet about the same topics. This would ensure that cofounders have common interests and could potentially lead to better collaboration and understanding.
Possible Implementation: This could work using an LLM by fetching the user's tweets and analysing the content of those tweets through an NLP library to identify the main topics and keywords in the user's tweets. For each potential cofounder, compare the topics and keywords and calculate a similarity score. Prioritise matches with higher similarity scores over the current tweets/commits ratio.