Resumen:
This paper describes a scalable and distributed mechanism to generate recommendations for items to any user.
It describes all the tools used, especially those provided by Amazon Web Services (AWS), for handling a massive
amount of data, testing and final implementation of this project, which includes the use of the paradigm
MapReduce through the framework Hadoop. For the generation of recommendations used a Collaborative
Filtering Algorithm based on Items and the calculation of the similarity between two items was applied Pearson
Correlation Coefficient. Also includes an example which measures the level of accuracy of the recommendations
generated, using the method of the Sum of Weights for the calculation of predictions and the Mean Absolute Error
(MAE) to evaluate the degree of similarity between the estimated scores (predictions) and the actual scores.
Although the primary focus is for movie recommendations, this solution can be applied to recommendations of
other items.
