The matching algorithm has been tuned somewhat, but it's not been tuned very much. It would be nice if we could further test/improve its accuracy. Some ideas include:
Adjusting the hyper-parameters of the matching model.
Implementing other matching algorithms, such as TF-IDF-based cosine similarity.
The matching algorithm has been tuned somewhat, but it's not been tuned very much. It would be nice if we could further test/improve its accuracy. Some ideas include: