Skip to content

Latest commit

 

History

History
43 lines (30 loc) · 2.03 KB

Recommendation_Integration.rst

File metadata and controls

43 lines (30 loc) · 2.03 KB

Recommendation Engine Integration

.. index::
 single: MyMediaLite

The Book Recommendation Application uses an open-source recommendation engine named MyMediaLite. This section provides more detail on how integration with the recommendation engine occurs.

Installation of MyMediaLite

The general steps are:

  1. Download MyMediaLite and install in the /opt/book directory
  2. MyMediaLite uses the .NET framework. If you are running on a non-Windows operating system you can use the Mono framework to emulate .NET. The command to install Mono on Ubuntu Linux is: sudo apt-get install mono-complete
.. index::
 single: rating.txt
 single: item_attributes.txt
 single: item_predict.txt

Using MyMediaLite

In order to generate book recommendations, MyMediaLite uses several different files as shown below.

File Name Direction Description
rating.txt Input Contains a record for each book and a computed rating based on total downloads
rating_test.txt Input Contains a record for each user, the book id, and their rating for the book.
item_attributes.txt Input Contains a record for each book and a unique attribute id for the language and subject
item_predict.txt Output Contains a record for each user and recommended books

Each of these files is generated by the Book Utility Java code discussed in the previous section.

Note

MyMediaLite can use a variety of statistical methods for making recommendations. More information can be found on the project's homepage.