Map/Reduce | Link Walking in JAVA-CLIENT
rajatmathurr at gmail.com
Mon Mar 5 14:53:42 EST 2012
I am new to Map Reduce Technique. I have been using *Riak-Java-Client.*
1.* I wish to know how can I write my own Map functions and Reduce
Functions for querying. Where they have to be stored if I want to use them
the way *NamedEralangFucntions/NamedJSFucntions *are used. How should
reduce functions be written so that they could do aggregation operations on
results provided by map phases.
*2. *During Map Reduce phases can we have an *alternated Filtration and
query process*, i.e. let's say we want to do index search on 4 indices and
have to take OR or AND of them, then instead of calculating keys for all
indices, can we *filter objects* (obtained from querying one of the 2i's )
based on indexes in map reduce.
*3. *If we were to query using Link Walk with Map reduce, but partially
i.e. let's say 2 stages of link walk (and since number of links would be
too high for my app after that, I have stored Keys of next bucket instead
of links to them. Is it fine?) and then from that data, filtering out
desired result, how to get it working ?
ex. Lets say each object in the last step of link walk had a list of keys
for next bucket's objects. Now, Is there a way to retrieve the objects of
last bucket[Who's key's list is contained in objects of last step..] by
continuing MapReduce, OR I have to Iterate each object's list attribute and
fetch objects by key ?
Please suggest some place on WWW where I could see more *CODE *for
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the riak-users