Abstract:
The daily recommendation systems become essential to filter the vast amount of information available on the
Web. One of the main objectives of these systems is to assist users in their information seeking process on the Web.
This paper presents a basic overview of key aspects related to design, implementation and module structure of a
recommendation to Wikipedia, using Map / Reduce Framework implemented in the Apache Hadoop to process
large amounts of information. At Wikipedia you can categorize two types of users, the contributor and the user
making the query, users often contribute to several wikis are these or not in the same category or content.
Module was created based on recommendations wikis that have contributed to the users. Given a wiki that users
were consulted contributed to this wiki, then search for each user to have contributed to other wikis and thereby
get for such a list of possible wiki wikis recommended.
To do whatever is possible not to lose the consistency of information between the user wiki wikis recommended
consulted and applied the Jaccard similarity coefficient between the wiki and consulted each of the wikis to
recommend possible. This ratio allows us to obtain a value that indicates the percentage of similarity between two
wikis based on the number of users who have contributed so much to one or both wikis, the end user are presented
to those wikis that the percentage of similarity obtained more high.
We can say that the result of the recommendations for a wiki topics also shows other users like you contributed
to the wiki, but not necessarily treated on the same topic.
