I think treating all parts of the metadata is hurting out query performance.
I'd like to see if weighting the metadata (i.e. assigning a weight to each word based off of what part of the metadata it came from) could help this.
My intuition is that the words in the title of the dataset are more relevant than the words in the column descriptions.
I think treating all parts of the metadata is hurting out query performance.
I'd like to see if weighting the metadata (i.e. assigning a weight to each word based off of what part of the metadata it came from) could help this.
My intuition is that the words in the title of the dataset are more relevant than the words in the column descriptions.