I have a lot of unstructured data in millions.
Some of them have {...data, value: <float>} and some have {...data, value: <long>} in the data.
All I'm trying to say is, the data type of value can differ.
In addition to it, if you suggest me to put a "float/double" mapping for the above field in the index, that's difficult too, as the attributes are dynamic in nature.
And this issue is not just with numbers.
Certain fields could be string sometimes, number sometimes or object sometimes. -> This case is still ok.. We can ensure that data doesn't go out of its way somehow.
But wrt numbers, maintaining strictness is difficult.
Please suggest a workaround, if it exists, for the above scenario