Part one of a presentation about Mahout system. It is based on http://my.safaribooksonline.com/9781935182689/

1. Mahout in Action
Part 1
Yasmine M. Gaber
28 February 2013

2. Agenda

Meet Apache Mahout

Part 1: Recommendation

Part 2: Clustering

Part 3: Classification

3. Meet Apache Mahout

It is an open source machine learning library
from Apache

It is scalable

It is a Java library

It can be used with Hadoop to deal with large
scale data.

4. Famous Engines

Recommender engines:

Amazon.com

Netflix

Dating sites like Líbímseti

Social networking sites like Facebook

Clustering engines:

Google News

Search engines like Clusty

Classification engines:

Spam emails

Google’s Picasa

Optical character recognition software

Apple’s Genius feature in iTunes

5. Recommendations

6. Recommender Input

A preference consists of a user ID and an item
ID, user’s preference for the item

It is .csv file

7. Create Recommender

8. Recommender Evaluation

Average difference vs Root-mean-square

9. Mahout RecommenderEvaluator

10. Precision and Recall

11. RecommenderIRStatsEvaluator

12. Representing Recommender Data

Preference object
− new GenericPreference(123, 456, 3.0f)

Preference Array

13. Representing Recommender Data

Preference Array

FastByIDMap and FastIDSet

14. In-memory DataModels

GenericDataModel

File-based data

Refreshable components

Database-based data

15. Coping without preference values

16. Coping without preference values

17. User-based Recommender

The algorithm
for every item i that u has no preference for yet
for every other user v that has a preference for i
compute a similarity s between u and v
incorporate v's preference for i, weighted by s, into a running
average
return the top items, ranked by weighted average

18. Recommender Components

Data model, implemented via DataModel

User-user similarity metric, implemented via
UserSimilarity

User neighborhood definition, implemented via
UserNeighborhood

Recommender engine, implemented via a
Recommender (here,

19. GenericUserBasedRecommender

20. User Neighborhoods

Fixed-size neighborhoods

Threshold-based neighborhood

21. similarity metrics

Pearson correlation–based similarity
− It is a number between –1 and 1 that measures
the tendency of two series of numbers, paired up
one-to-one, to move together
− Problems:

It doesn’t take into account the number of items in
which two users’ preferences overlap, which is probably
a weakness in the context of recommender engines.

If two users overlap on only one item, no correlation can
be computed because of how the computation is
defined

22. similarity metrics

Euclidean distance similarity
− 1 / (1+euclidean distance)

Cosine measure similarity
− between –1 and 1

Tanimoto coefficient similarity
− The ratio of the size of the
intersection to the size of
the union of their preferred items

23. Item-based recommendation

The algorithm
for every item i that u has no preference for yet
for every item j that u has a preference for
compute a similarity s between i and j
add u's preference for j, weighted by s, to a running average
return the top items, ranked by weighted average

24. GenericItemBasedRecommender

25. Slope-one recommender

The algorithm
for every item i the user u expresses no preference for
for every item j that user u expresses a preference for
find the average preference difference between j and i
add this diff to u's preference value for j
add this to a running average
return the top items, ranked by these averages

26. Taking Recommender to Production

27. User-based recommenders

28. Thank You
Contact at:
Email: Yasmine.Gaber@espace.com.eg
Twitter: Twitter.com/yasmine_mohamed