Recommendation for Alfresco?

Question

I want to implement recommendations like Amazon in Alfresco .

For example, if an employee searches for “2007 financial reports,” the search user interface will display related documents, such as documents that have been downloaded or viewed by users who previously searched for the same.

It can display documents that were not found by Lucena (which uses Alfresco).
For example, did anyone integrate Alfresco with Apache Mahout or pysuggest ?

Answer 1

We included Mahout in Alfresco to provide Content Recommendations based on similar users of the content, as well as on how users rated the content. Alfresco Mahout Integration Code Available at

https://github.com/zaizi/alfresco-recommendations

This provides Amazon style advertising guidelines. It can be expanded to recommend similar search phrases.

Answer 2

Interesting topic! I recently read about Mach in the context of Lutsen / Sol. There are some people deeply involved in Mach in the Lusimimination, see:

• https://lucidworks.com/blog/2011/11/08/apache-mahout-scalable-machine-learning-for-everyone/
• https://lucidworks.com/blog/2010/03/16/integrating-apache-mahout-with-apache-lucene-and-solr-part-i-of-3/

Since Lucene / Solr is part of Alfresco, you might consider integrating it at the search engine level. Alternatively, you can ask for canoo (Basel, Switzerland). Previously, they offered us an interesting solution for a multi-platform document developed on the basis of Solr.

Answer 3

It is good that alfresco by default supports links (associations). This way you can define many useful relationships between documents. For instance:

Document-> User => Viewed

Document-> User => Search

Document-> User => uploaded

Document-> Document => Related

Document-> Document => Same year

...

You can catch / implement most events using policies / behavior in alfresco (http://wiki.alfresco.com/wiki/Policy_Component). For example: when the onCreate event occurs (a document is created), search for documents with the same author and attach this document (add associations) to them.

Then you can implement a custom search (possibly webscript) that will return results, and for each result also return links (associations).

The only thing that bothers me is that some events are likely to be available only through the audit log, which I have no idea how to programmatically use java.

In the end, you can pass this stuff on to your engine that finds out about it.