ABSTRACT:
It is really popular to detect hot topics, which can benefit many tasks
including topic recommendations, the guidance of public opinions, and so on.
However, in some cases, people may want to know when to re-hot a topic, i.e.,
make the topic popular again. In this paper, we address this issue by
introducing a temporal User Topic Participation (UTP) model which models users’
behaviors of posting messages. The UTP model takes
into account users’ interests, friend-circles, and unexpected events in online
social networks. Also, it considers the continuous temporal modeling
of topics, since topics are changing continuously over time. Furthermore, a
weighting scheme is proposed to smooth the fluctuations in topic re-hotting prediction. Finally, experimental results conducted
on real-world data sets demonstrate the effectiveness of our proposed models
and topic re-hotting prediction methods.
PROJECT OUTPUT VIDEO: (Click the below link to
see the project output video):
EXISTING
SYSTEM:
·
Wang et
al. propose an algorithm to predict topic trends, which addresses the problem
of short life circles of topics. Furthermore, a methodology is presented to
detect the topic of epidemics based on Twitter. However, all these research
work just concentrates on detecting popular topics, and they cannot be directly
used to deal with the problem of TRP.
·
Zhang et
al. propose a new method to detect events and to predict their popularity
simultaneously. Specifically, they detect events from online microblogging stream by utilizing multiple types of
information, i.e., term frequency and users’ social relation. Meanwhile, the
popularity of detected event is predicted through a proposed diffusion model
which takes both the content and user information of the event into account.
·
Zhang et
al. address the problem of inferring continuous dynamic users’ behavior by utilizing both the social influence and the
personal preference.
DISADVANTAGES
OF EXISTING SYSTEM:
·
Topic re-hotting prediction is more difficult than topic detection.
·
The
methods of topic detection only justify whether or not a new topic is emerging,
however the topic re-hotting prediction approaches
should tell exact time points when a given topic will re-emerge.
·
Unfortunately,
to the best of our knowledge, few studies considered when to re-hot topics so
far.
·
It is
nontrivial to formalize the problem of topic re-hotting
prediction and reasonably model the mechanism of topic participation.
·
It is
very difficult to precisely obtain opportune time points for re-hotting a given topic.
·
It is not
easy to propose an effective topic re-hotting
prediction approach.
PROPOSED
SYSTEM:
·
This
paper addresses the problem of topic re-hotting
prediction. As shown in Fig, we could consider the following two strategies to
deal with the topic re-hotting prediction problem.
·
The
discrete modeling strategy divides the whole time
domain into contiguous non-overlapping time windows, and then uses the trained
data (depicted as blue broken lines) to predict whether the topic will re-hot
in the next time window (i.e., during the period from t5 to t6). Although this
strategy is easily understandable, it cannot predict accurate time points for
re-hotting a given topic. Furthermore, it is hard to
describe the changing trends of topics in a fine-grained manner.
·
The continuous
modeling strategy argues that topics are continuously
changing in the time domain. Based on the trained data (depicted as red solid
lines), it predicts accurate time points when the topic will re-hot, e.g., at
the time point t’5. Please note that this strategy could predict the re-hotting time points over a long period of time (depicted as
red dotted lines) instead of just the next time window.
ADVANTAGES
OF PROPOSED SYSTEM:
·
We
present and formalize the problem of topic re-hotting
prediction (TRP) in OSNs at the first time. It facilitates a better
understanding of the topic characteristics when the focusing topics are
dwindling, as well as benefits many related issues, such as topic detection and
topic tracing.
·
We
propose a novel temporal model, i.e., User Topic Participation (UTP) model, for
the TRP problem. UTP can effectively explain users’ behaviors
of participating in the topic discussions in OSNs. Also, we bring forward an
improved EM algorithm called EMG to effectively infer the UTP model.
·
We design
a method based on the UTP model to appropriately predict the re-hotting time points for given once-hot topics, i.e., the
topics which had been hot before.
·
We
evaluate the performance of our methods on three different real-world data sets
collected from OSNs. Experimental results demonstrate the effectiveness of both
the proposed UTP model and TRP method.
SYSTEM
ARCHITECTURE:
SYSTEM
REQUIREMENTS:
HARDWARE
REQUIREMENTS:
·
System : Pentium Dual Core.
·
Hard Disk : 120 GB.
·
Monitor :
15’’ LED
·
Input
Devices : Keyboard, Mouse
·
Ram : 1
GB
SOFTWARE
REQUIREMENTS:
·
Operating
system : Windows 7.
·
Coding
Language : JAVA/J2EE
·
Tool : Netbeans 7.2.1
·
Database
: MYSQL
REFERENCE:
Chaokun Wang,
Member, IEEE, Xin Xin, and Jingwen Shang, “When to Make a Topic Popular Again? A Temporal
Model for Topic Re-hotting Prediction in Online
Social Networks”, IEEE
Transactions on Signal and Information Processing over Networks 2017.