Journal of Android and IOS Applications and Testing

Volume 4 Issue 3

A Real Time Video Streaming Web Portal

Anita Mukund Pujar1, Pratibha Arvind Domal2*
1
Professor, 2PG Student
1,2
Department of Computer Science Engineering, WIT, Punyashlok Ahilyadevi Holkar Solapur
University, Solapur, Maharashtra, India
Email: *domalpratibha2138@gmail.com
DOI:

Abstract
Content-based video extracting is very essential now-a-days. The existing data mining
algorithms are not directly applied to videos. This proposed work uses different data mining
algorithms for indexing, clustering, searching and retrieving content-based videos. A system
will be developed in which only admin can upload videos on cloud server. Videos are sorted
based on category and videos are automatically uploaded on Cloud Server on the schedule
provided by the admin. Users can watch videos online, they can download videos based on
video summary and users can rate videos that will be analyzed by the system. This system
automatically removes videos that are less popular so that user’s time will not be wasted for
watching the least rated videos. Users can share videos with other registered users without
using another platform in the proposed system.

Keywords: Clustering, Content-based retrieval, Indexing, Predictive analytics

INTRODUCTION Users can also decide to download the

The multimedia data including text, image,
audio, and video have been produced
massively. The digital videos rapidly
became an important source for education,
advertisements, promotions, and
entertainment [16]. Looking at the need of
users at large digital videos, the system
proposes to maintain the repository of
videos on cloud server.
videos based on the summary provided for
each video. The users can also watch
videos online. Users can rate the videos so
that the system can analyze the popularity
of videos and this analysis is used further
for deleting the least popular videos. Users
can share videos with other registered
users without using another platform in the
proposed system.

Private cloud is a deployment model of The following algorithms are used for
implementing the above features are as
cloud computing where services will be
follows:
accessed by authorized organization or
• The clustering algorithm is used for
owner of private cloud.
sorting the videos based on categories.
Clustering is the process of collecting
Videos are automatically uploaded on
and grouping objects that belongs to
cloud server on the schedule provided by
the same class. The objects which are
the admin and videos are sorted based on
similar grouped in one cluster and
the categories. Users can search for videos
remaining dissimilar objects are
by keywords like title, date, and author so
grouped into another cluster.
that content-based videos will be retrieved.
• Indexing and prediction algorithms are

16 Page 16-20 © MAT Journals 2019. All Rights Reserved

 Journal of Android and IOS Applications and Testing
used for searching and retrieving
content-based videos from huge video
datasets based on the keywords.
An index or database index is a data
structure that is used to quickly locate and
access the data from a database or datasets.
Predictive analytics is the practice of
extracting useful information from existing
data sets to determine patterns and predict
future outcomes and trends.
The Present Proposed Work Includes
The system proposes to maintain the
repository of videos on cloud server and
videos are automatically uploaded on
cloud server on the schedule provided by
the admin. Sorting the videos based on
categories using clustering algorithm.
Users can also watch the videos online or
they can also decide to download the
videos based on the summary provided for
each video.
• Indexing and prediction algorithms are
used for searching and quickly
retrieving content-based videos from
huge video datasets.
• Users can rate the videos so that,
system can analyze the popularity of
videos and this analysis is used further
for deleting the least popular videos.
• Users can share videos with other
registered users without using another
platform in the proposed system.
Related Work
Zhong Hanyang, Song Xin, Yan Zhenguo
[1], a typical clustering algorithm called
K-means is applied to deal with the Space-
based AIS(S-AIS) data received by
“TianTuo-3” satellite developed by
National University of Defence
Technology. In this paper, they used
17 Page 16-20 © MAT Journals 2019. All Rights Reserved
Volume 4 Issue 3
Elbow Rule to determine the optimal
number of clusters and calculate the
normalized standard deviation of course
over ground and speed over ground of
vessels in South Africa area as their
features to conduct clustering. The
proposed method is supposed to evaluate
vessels’ sailing stability and used in
detection of low-likelihood behaviours or
anomalies of vessels. The real-time
performance of one single AIS satellite is
still very poor, which shows the
importance of establishing AIS satellite
constellations in the future.
Shalini L and Gopali Naga Sravya [2]
collected tweets from several records and
analyzed using several sources like fox
news health, CBC health, CNN health,
BBC health, everyday health, GDN health
care, good health. After analysis, tweets
are combined from these 16 records, the
top 5 clusters of negative and positive
words are obtained by using K-means
clustering algorithm. The analysis is done
on the number of negative and positive
words present in those records, and then
corresponding plot for the frequency of top
40 words is obtained. They found that the
views of different people related to the
news on health in different channels.
Igor V Anikin and Rinat M Gazimov [3],
developed DBCSAN clustering algorithm,
which provides security of information
during all stages of the distributed data
mining process. It would be very useful for
Data Mining Techniques in distributed
systems with big data.
Uma Ojha and Dr. Savita Goel [4], found
how precisely these data mining
algorithms can predict the probability of
 Journal of Android and IOS Applications and Testing
recurrence of the diseases among the
patients on the basis of important stated
parameters. Experiments show that
clustering algorithms are worser predictors
than classification algorithms. The result
indicates that the decision tree (C5.0) and
SVM algorithms are the best predictor
with 81% accuracy on the holdout sample
and fuzzy c-means algorithm came with
the lowest accuracy of 37%. M Sivasakthi
[5]. The various supervised data mining
algorithms were applied on the collected
data set to predict programming
performance of the students that are
evaluated based on their predictive
accuracy. The results indicate that the
Multilayer Perception performs best with
93% accuracy and therefore, multilayer
perception proves to be very efficient and
effective classifier algorithm. This analysis
may help the Institutions to identify the
students who are novice programmers in
introductory programming, which further
provide base for deciding special aid or
focus to them.
Peter Brauna, Alfredo Cuzzocreab, Lam
MV Doana, Suyoung Kima, Carson K
Leunga, Jose Francisco A Matundana and
Rashpal Robby Singha [6], explored big
data mining techniques for detecting
outliers or anomalies from YouTube video
viewing history and data-cleaning this
viewing log so that the user-preferred
YouTube viewing patterns or trends can be
recognized and the prediction of user-
preferred YouTube videos can then be
enhanced. They used outlier detection
algorithm for finding anomalies for
YouTube viewing history log so that these
anomalies can be removed.
18 Page 16-20 © MAT Journals 2019. All Rights Reserved
Volume 4 Issue 3
Brian McClanahan and Swapna S Gokhale
[7], examined the interplay between
popularity ratings and video
recommendations. They found that about
40% of the video recommendations come
from categories other than that of the
original video, with Entertainment being
the most preferred cross-linked category,
popularity measures including the number
of views and comments strongly impact
video recommendation, video
categorization has a higher influence on
video recommendations compared to their
popularity ratings.
METHODOLOGY
An Execution of Proposed System
In Fig. 1, the working of proposed system
is described as follows:
Admin authentication process is done for
accessing the video datasets. Videos are
automatically uploaded on Cloud Server
on the schedule provided by the admin.
Videos are sorted based on the categories
using clustering algorithms. User
registration and authentication process is
carried out for allocating permissions [8,
9]. Users can search for videos by
keywords like title, date, and author so that
content-based videos will be retrieved. The
indexing and prediction algorithms are
used for retrieving the content-based
videos [10-12]. Users can also decide to
download the videos based on the
summary provided for each video. The
users can also watch videos online and
users can rate the videos so that system
can analyze the popularity of videos and
this analysis is used further for deleting the
least popular videos. Users can share
videos with other registered users without
using another platform in the proposed
system [13-15].
 Journal of Android and IOS Applications and Testing
Volume 4 Issue 3

Admin Automatic Video Bank

Authentication uploading on the scheduled
Process date

Video Sorting Cloud Server

based-on
Categories

User
Registration Retrieving
Searching
and Content-based
Videos by Videos
Authentication Keywords
Process

Automatically
Share videos system removes Rate Downloading
with registered least popular Videos based
Videos
users videos on Summary

Figure 1: An execution of proposed system.

EXPECTED RESULT easily access content-based videos based

The proposed system should retrieve
content-based videos based on the
keywords using Prediction algorithms,
remove least popular videos and allow
users for sharing of videos to other
registered users without using another
platform.
CONCLUSION
The proposed work aims to develop a
system that provides a real time video
streaming web portal which maintains
huge amount of multimedia using cloud
server from which users can quickly and
on keywords and by using cloud server
disk space is reduced for storing huge
number of videos locally. Videos are
sorted based on category and are
automatically uploaded on Cloud Server
on the schedule provided by the admin.
The videos can be watched online, videos
can be downloaded based on video
summary, least popular videos are
automatically removed based on users'
rating and users can share videos with
other registered users without using
another platform in the proposed system.

19 Page 16-20 © MAT Journals 2019. All Rights Reserved

 Journal of Android and IOS Applications and Testing
REFERENCES
1. Zhong Hanyang, Song Xin, Yan
Zhenguo (2019), “Vessel sailing
patterns analysis from S-AIS data
based on K-means clustering
algorithm”, IEEE, Suzhou, China.
2. Shalini L, Gopali Naga Sravya
(2018), “Analysis of Health-Tweets
using K-means clustering”, IEEE,
Volume 5, Issue 3, pp. 2074-2077, e-
ISSN: 2395-0056, p-ISSN: 2395-
0072.
3. Igor V Anikin, Rinat M Gazimov
(2018), “Privacy preserving data
mining in terms of DBSCAN
clustering algorithm in distributed
systems”, IEEE, Moscow, Russia.
4. Uma Ojha, Savita Goel (2017), “A
study on prediction of breast cancer
recurrence using data mining
techniques”, IEEE, Noida, India.
5. M Sivasakthi (2017), “Classification
and prediction-based data mining
algorithms to predict students,
Introductory programming
Performance”, IEEE, Coimbatore,
India.
6. Peter Brauna, Alfredo Cuzzocreab,
Lam M.V. Doana, SuyoungKima,
Carson K. Leunga,*, Jose Francisco
A. Matundana, Rashpal Robby Singha
(6-8 September 2017), “Enhanced
prediction of user-preferred youtube
videos based on cleaned viewing
pattern history”, International
Conference on Knowledge Based and
Intelligent Information and
Engineering Systems, KES2017,
Marseille, France.
7. Brian McClanahan, Swapna S
Gokhale (2017), “Interplay between
video recommendations, categories,
and popularity on YouTube”, IEEE,
San Francisco, CA.
8. MdShamsur Rahim, AZM Ehtesham
Chowdhury, Md. Asiful Islam, Mir
Riyanul Islam (2017), "Mining
Trailers Data from YouTube for
20 Page 16-20 © MAT Journals 2019. All Rights Reserved
Volume 4 Issue 3
Predicting Gross Income of Movies",
IEEE, Dhaka, Bangladesh.
9. Ms. Tismy Devasia, Ms. Vinushree
TP, Mr. Vinayak Hegde (2016),
“Prediction of students performance
using educational data mining”, IEEE,
Ernakulam, India.
10. Chenyu Li, Shuxin Ouyang (2016),
“Characterizing and predicting the
popularity of online videos”, IEEE,
Volume 4, pp. 1630-1641.
11. Anand Bhave, Himanshu Kulkarni,
Vinay Biramane, Pranali Kosamkar
(2015), “Role of different factors in
predicting movie success”, IEEE,
Pune, India.
12. Ronit Nossenson, Shuval Polacheck
(2015), “On-line flows classification
of video streaming applications”,
IEEE, Cambridge, MA, USA.
13. Xu Cheng, Jiangchuan Liu, Cameron
Dale (2013), “Understanding the
characteristics of internet short video
sharing: A YouTube-based
measurement study”, IEEE, pp. 1-9.
14. Gonca Gursun, Mark Crovella
Ibrahim Matta (2011), “Describing
and forecasting video access
patterns”, IEEE, Shanghai, China.
15. Milan Mirkovic, Dubravko Culibrk,
Symeon Papadopoulos, Christos
Zigkolis, Yiannis Kompatsiaris,
Gavin McArdle, Vladimir Crnojevic
(2011), “A comparative study of
spatial, temporal and content-based
patterns emerging in YouTube and
Flickr”, IEEE Salamanca, Spain.
16. D Saravanan, S Srinivasan (2011),
“Data mining framework for video
data”, IEEE, Chennai, India.
Cite this article as: