[
https://issues.apache.org/jira/browse/MAHOUT-1420?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13910446#comment-13910446
]
Pat Ferrel commented on MAHOUT-1420:
------------------------------------
Sure, it all depends on what you want to show. One goal I had was to take a
logfile meant for human readability, which almost always needs quite a bit of
scrubbing to feed into Mahout and just use it raw. This comes from a job I did
for a local analytics company who had just this. So I designed and built a
fairly flexible framework--not all options are fully implemented. It allows for
redefining text separators, filename patterns for input files, recursively
searching directories for inputs, defining fields numbers for each of userId,
itemId, actionName, etc.
It splits the input into one directory per action and turns them into Mahout
format (integers everywhere) keeping index/dictionary and the reverse in files.
>From then on it is RecommenderJob or my proto XRecommenderJob, which contains
>a PrepareActionMatrixesJob to create the two input DRMs. Anyway the output of
>these are the standard mahout outputs and temp files. These are input into the
>rest of the Solr pipeline. For instance the similarity matrices are then
>transformed into delimited files with external ids, as they were in the input
>logfiles, ready for Solr to index.
I put up an out-of-the-box Solr server and had it index the output. You can
feed in user history in the same form (also one of the outputs of the above
jobs) and get recs by hand. I went this far but was unable to commit to
packaging up the solr server part. This might be the simplest form of example
that could be deployed easily. I tired to get someone to go this last 100ft but
got no volunteers.
> Add solr-recommender to examples
> --------------------------------
>
> Key: MAHOUT-1420
> URL: https://issues.apache.org/jira/browse/MAHOUT-1420
> Project: Mahout
> Issue Type: New Feature
> Components: Examples
> Affects Versions: 0.9
> Reporter: Andrew Musselman
> Assignee: Andrew Musselman
> Priority: Minor
> Fix For: 1.0
>
>
> Write a new example that builds a solr-recommender based on Pat's code at
> https://github.com/pferrel/solr-recommender and which has the glue scripts
> needed to pipe all the way from start(raw data) to finish(running web service
> and UI page).
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
