Abstract:
The emergence and rapid proliferation of
various social media networks have reshaped the way how video contents are
generated, distributed and consumed in traditional video sharing portals.
Nowadays, online videos can be accessed from far beyond the internal mechanisms
of the video sharing portals, such as internal search and front page highlight.
Recent studies have found that external referrers, such as external search
engines and other social media websites, arise to be the new and important
portals to lead users to online videos. In this paper, we introduce a novel
cross-network collaborative application to help drive the online traffic for
given videos in traditional video portal YouTube by leveraging the high propagation
efficiency of the popular Twitter followees. Since YouTube videos and Twitter
followees distribute on heterogeneous spaces, we present a cross-network
association-based solution framework. In this framework, we first represent
YouTube videos and Twitter followees in the corresponding topic spaces
separately by employing generative topic models. Then, the cross-network topic
spaces are associated from both semantic-based and network-based perspectives
through the collective intelligence of theo bserved overlapped users. Based on
the derived cross-network association, we finally match the query YouTube
videos and candidate Twitter followees in the same topic space with a unified
ranking method. The experiments on a real-world large-scale dataset of more
than 2.2 million YouTube videos and 31.8 million tweets from 38,540 YouTube
users and 39,400 Twitter users demonstrate the effectiveness and superiority of
our solution in which network-based and semantic-based association are
integrated.
Existing system:
The rise of social media, the way people
can get access to the video contents is changing. Instead
of only relying on the internal mechanisms
provided by the traditional video sharing portal to access the videos, more and
more people now prefer to directly watch videos from their involved social
media networks . Fore See has reported that more than 18% users are influenced
by the social network when accessing video contents1 and a significant portion
of what users watch is being increasingly referred by social media. Some recent
research work has also begun to investigate into the interaction between video
sharing and social media networks .
Disadvantages:
Social media brings in much possibility
to the traditional video sharing portals and it is very fascinating to explore
the innovation sparkles generated when these two meet with each other.
Proposed System:
The key lies in how to establish a
reasonable cross-network association between YouTube
and Twitter. Inspired by the fact that
the same individual usually involves with different social media networks and
different social media networks share remarkable percentage of overlapped users
3, if we know the corresponding Twitter accounts of YouTube users who show
interest to a given video (e.g., upload, favorite, add to playlist), it is
confident to identify the Twitter followee that these Twitter accounts jointly
followed as the optimal promotion referrer. Therefore, we propose a brand new
way to establish the cross-network association by leveraging the collective
intelligence of the observed overlapped users. Since YouTube video generally
distributes on specific semantic level, a direct way to associate the YouTube
and Twitter spaces is from the content level where the YouTube video content
and Twitter users’ generated tweets are
used to capture the association. However, this kind of association can only capture
the semantic correlation between the two spaces and still suffers from the
discrepancy issue in topic granularity. Therefore, we also make further
exploration with a more flexible and heterogeneous kind of association under
which the YouTube video content and the network structure of the Twitter users
are correlated.
Advantages:
A cross-network association-based solution
framework is presented, under which different kinds of associations are
explored and integrated To the best of our knowledge, this is the first attempt
to mine the cross-network association under a user-bridged scheme.
Algorithum:
Implementation Modules:
1. Cross-network
Collaboration
2.
Social Media Influencer Mining
3.
Heterogeneous Topic Association
4.
Heterogeneous Topic Modeling
Cross-network Collaboration:
With various social media networks
growing in prominence are using a
multitude of social media services for social connection and information
sharing. Cross-network collaborative applications have recently attracted
attentions. One line is on cross-network user modeling, which focuses on
integrating various social media activities.The authors introduced a cold-start
recommendation problem by aggregating user profiles in Flickr, Twitter and
Delicious. Deng et al. has proposed a personalized YouTube video recommendation
solution by incorporating user information from Twitter . Another line is
devoted to taking advantage of different social networks’ characteristics
towards collaborative applications. exploited the real-time and socialized
characteristics of the Twitter tweets to facilitate video applications in
YouTube. Twitter event detection is
conducted by employing Wikipedia pages as the authoritative references. Our
work belongs to the second line, where a collaborative application is designed
to exploit the propagation efficiency of Twitter to meet the YouTube video
promotion demand.
2.Social Media Influencer Mining
Previous analysis on Twitter has found
that popular users with high in-degree are not necessarily influencers for
propagation , which calls for research onto the problem of influencer mining.
One line is to identify the domain or topic experts. Our introduced problem of Twitter followee
identification can be viewed as a special case of influencer mining. The
existing influencer mining methods mainly focus on single network and need an
explicit relevance metric, e.g. the topical relevance between follower and followee, and the accept rate between the
propagation item and follower. In our problem, the relevance of influencer is
designed by items distributed on another network. It is difficult to explicitly
define the relevance metric between cross- network knowledge. Moreover, to
focus on addressing cross-network association, we pay no attention to the
complicated social network structure as in the standard maximizing influence
problems. What we care is actually about the propagation efficiency in the
first level of followee -follower network.
3. Heterogeneous Topic Association:
The propagation efficiency in the first
level of followee-follower network. Heterogeneous Topic Association The
core of our solution lies in the heterogeneous topic association between
Twitter followee and YouTube video. Typical applications of existing
heterogeneous topic association work include heterogeneous face recognition and
cross-media retrieval, where invariant feature extraction and subspace learning
based solutions are extensively investigated. Invariant feature extraction
methods are devoted to reducing the heterogeneous gap by exploring the most
insensitive feature patterns. proposed to extract the SIFT and Multiscale LBP
for forensic sketch and mug shot photo matching .The intra- difference and
inter-difference are jointly considered into a discriminant local feature
learning framework. The basic idea of subspace learning is to learn a new space
where the observed heterogeneous data can be well represented, among which
subspace clustering has proved to be a very effective method to represent data
from multiple sources and modalities
4.Heterogeneous Topic Modeling:
YouTube Video Topic Modeling: In YouTube, the video topics are
expected to span over both
textual and visual spaces. We introduce
a modification to the multi-modal topic model, Corr-LDA. Corr-LDA is proposed
for the problem of image annotation, by modeling the correspondence between image segments and
caption words. It assumes a generative process that first generates the segment
descriptions and subsequently the caption words.
Semantic-based
Twitter User Topic Modeling:YouTube
videos distribute more on semantic level, it is natural to also represent
Twitter users in some semantic topic space. Therefore, we aggregate each
Twitter user’s generated tweets and keep only the nouns and hashtags5. Then the
standard LDA model is applied to each user for topic modeling, with user as document, the nouns or hash tags of
his/her tweets as word. In this
way, the derived Twitter topics can capture some co-occurred semantic concepts
frequently used by many users which may also be found in YouTube video semantic
space.
Network-based
Twitter User Topic Modeling:The properness of
Twitter followee is decided by the followers, we are also interested in
investigating into the followee-follower architecture in Twitter. Therefore, we
represent each Twitter user (document)
with all his/her followees (words)
and apply thestandard LDA for topic modeling. Since topic modeling exploits
co-occurrence relationships, like the YouTube video topics capturing the
frequently co-occurred visual features and textual words in videos, the derived
Twitter topics actually capture the shared followees by a subset of Twitter
users.
Configuration:-
H/W System Configuration:-
System -
Pentium –IV 2.4 GHz
Speed -
1.1 Ghz
RAM -
256MB(min)
Hard Disk - 40 GB
Key Board - Standard Windows Keyboard
Mouse -
Logitech
Monitor - 15 VGA Color.
S/W
System Configuration:-
v Operating System :Windows/XP/7.
v Application
Server :
Tomcat5.0/6.X
v Front End :
HTML, Java, Jsp
v IDE :Eclipse
v Scripts : JavaScript.
v Server side Script : Java
Server Pages.
v Database : Mysql 5.0
v Database Connectivity :
JDBC.
Comments
Post a Comment