Multimedia big data computing for trend detection
Tutor / Supervisor
Student
Sulca Correa, Omar Iván
Document type
Master thesis
Date
2015
rights
Open Access
Publisher
Universitat Politècnica de Catalunya
UPCommons
Abstract
The Big data analysis has becoming
increasingly more relevant for the enterprises
because the efficient handling of information represents a unique competitive advantage,
being its application so diverse as the nature of the data. Ejm. Fraud detection, advertising
strategies, web traffic m
onitoring, etc.
Apache Spark is a engine for large
-
scale data processing, intended to be a drop in
replacement for Hadoop MapReduce providing the benefit of improved performance; the
main goal of this project is proof the capabilities of this system, throu
gh the development and
implementation of a distributed pipeline for processing and indexing at high speed and real
-
time multimedia data streams generated by social networks and detect trends in these, using
for this purpose the Spark related projects and l
ibraries: Spark Streaming and Spark MLlib.
To verify the effectiveness of the algorithm, different benchmarks (with different
configurations) will be performed,
these results
will be analyzed.
