Geek asks and answers

Matrix multiplication in hadoop

I am trying to create code for the page ranking algorithm, and this is the main difficulty in effectively resolving matrix multiplication, but I did not understand how this task is performed, I read some documents about this, but it is beyond my range. I did not understand the concept that he applies. So, can you give me a concept related to the mapping and reducer function for matrix multiplication. Thanks in advance.

I read the link

+5
source share
3 answers

The idea is that you can break matrix multiplication into subtasks with something like Strassen Algorithm and then send these subtasks to a bunch of different computers.As soon as these subtasks are finished, you can also process them together with summing the various subtasks into a matrix. The key to using Mapreduce is that all the subtasks can basically be computed in parallel, which ... what Mapreduce is for.

+1
source

A couple of frameworks like Apache Hama have a PageRank implementation . Apache Giraph also supports Pagerank .

MapReduce PageRank, Google Pregel paper .

0

The link you provided explains this as clearly as everything (plus includes the source code). If you are still struggling with the concepts of operations, then you should probably start by reading a little more about matrix / linear algebra, so you understand basic math.

-1
source

More articles:


All Articles