Igor,
Not 100% sure what you mean by "filter" in this case - but - it sounds
like you need to think about how to express Apriori as a sequence of
operations that explicitly includes counted aggregation followed by
normal filtering on the count field? At least to get something
working? Limit doesn't feel like the right approach here; it's more of
an important operator when you are trying to stop the flow of unwanted
"lower down" data. Think about how you might express Apriori in the
extended relational algebra...?
Cheers,
Mike
On 8/30/19 8:28 AM, Ígor Chagas wrote:
Hi, devs
I started to work with Hyracks a few months ago. I tested several
variations on WordCount code to learn about its paradigm. Now, I'm trying
to implement Apriori Algorithm on it to do some studies.
In some step of this algorithm, I need to filter tuples that have a number
of occurrences higher than N. However, I did not find a proper filter
operator. I tried to use the LimitOperator¹ as a template since it is
simple and deals directly with tuples. However, I didn't figure out how to
compare some tuple field values to a specific Apriori parameter, as support.
Do you guys have some advice or a starting point to implement a filter
operator in this situation? Is there something I'd missed?
Thanks and Regards.
Ígor Chagas Marques
[1] -> dataflow/std/misc/LimitOperatorDescriptor.java
