More on the curiosity side on property keys, but it could have an impact on stats calculations.
Very simple query to get the property keys on nodes
MATCH (n) RETURN labels(n),keys(n), count(*)
What seemed a bit odd were the results,
The nodes with more than 1 key would have different orders of the keys, though not all the permutations. If I wanted to further process the results I would need to account for that permutation.
|[word]|[pos, term, count]|15103|
|[word]|[count, pos, term]|412|
|[word]|[term, pos, count]|124|
When entering the data it all came from the same CSV file in the same operation.
Is this expected behavior?