Transcript PPT

EFFECT OF CLIMATE AND SEASONALITY
ON DEPRESSED MOOD AMONG TWITTER
USERS
Wei Yanga, Lan Mua, Ye Shenb
a)Department of Geography, University of Georgia, United States
b)Epidemiology and Biostatistics, College of Public Health, University of Georgia, United States
2015 Applied Geography, Elsevier (cross mark)
Background
Method
Result
Discussion
Conclusion
Limitation
› Location-based social media provide an enormous stream of data about
humans’ life and behavior.
› Relationship between depression rates, climate risk factors, and
seasonality are varied and geographically localized.
› New methods are needed to synthesize geo-visual analytics and social
media analytics.
› Time often interacts with space, and therefore maps need to be more
dynamic.
› Considering the continuity and broad coverage of study areas, maps
need to be merged in a creative and meaningful way to enable humans
to visualize the large amount of information intuitively
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Background
Method
Result
Discussion
Conclusion
Limitation
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Background
Method
Result
Discussion
Conclusion
Limitation
Spatiotemporal patterns of tweets related to depression
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Background
Method
Result
Discussion
Conclusion
Limitation
Spatiotemporal patterns of tweets related to depression
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Background
Method
Result
Discussion
Conclusion
Limitation
Spatiotemporal patterns of tweets related to depression
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Background
Method
Result
Discussion
Conclusion
Limitation
Spatiotemporal patterns of tweets related to depression
Humid Continental (Cool Summer)
Humid Continental (Warm Summer)
Humid Subtropical
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Tropical Season
Midlatitude Desert
Mediterranean
Marine Westcoast
Background
Method
Result
Relationship between climatic
factors and depression rate
Humid Continental
(Cool Summer)
Humid Continental
(Warm Summer)
Humid Subtropical
Tropical
Season
Midlatitude Desert
Mediterranean
Marine Westcoast
Discussion
Conclusion
Limitation
Background
Method
Result
Discussion
Conclusion
Limitation
› Midlatitude desert climate has a negative relationship with
depression rate
› Humid continental (warm summer) climate has a positive
relationship with depression rate
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Backgroun
d
Method
Result
Discussion
Conclusion
Limitation
› There are three stages here, consist of geographical data acquisition, data
reduction and text mining, also spatio-temporal analysis of health issues
(included visualization).
› This framework can help us understand how social and behavioral
interventions influence humans’ health and illness.
› Also, this framework can be used to detect major event outbreaks,
such as flu and earthquakes.
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
Backgroun
d
Method
Result
Discussion
Conclusion
Limitation
› This research only includes tweets written in English and users who
identified themselves living in the U.S. Modifying any of those may
change the results
› This research only focus on the depression effects linked with
general climate variables, they will take other depression triggers
into consideration in future research
Effect of Climate and Seasonality on Depressed Mood Among Twitter Users
IOT SOLUTIONS FOR 3-D VISUALIZATION
OF TWITTER DATA
Aman Sharmaa, Rinkle Rania
a)Department of computer Sc. & Engg., Thapar University, India
2015 IEEE International Advance Computing Conference (IACC)
Background
Method
Result
Conclusion
Future Work
› Internet of Things (IoT) is a booming terminology that is used nowadays.
It is interconnection of various objects such as RFID, actuators, smart
› devices, sensors over an internet.
Twitter being a popular social networking site always gives researchers a
›
scope innovation
› Various new ways of data visualization and networks have
influenced many research areas
› Great effort from industry and academia to provide IoT solutions and
develop customer oriented market for IoT devices
› What if 3-D visualization of data using day to day useful things like
lamp, smoothie jar, etc?
IoT Solutions for 3-D Visualization of Twitter Data
Background
Method
Result
Conclusion
Extracting tweets using Arduino Yun.
Arduino is open-source tool used for
developing computing devices that
have sensing and controlling features,
a lot more than a normal computer.
Arduino Yun can be configured to
connect to your WiFi network
IoT Solutions for 3-D Visualization of Twitter Data
Future Work
Background
Method
Result
Conclusion
Arduino Yun
IoT Solutions for 3-D Visualization of Twitter Data
Future Work
Background
Method
Result
Conclusion
Future Work
1. Twitter Mood Light
A lamp will glow a unique
color based on the frequency
of tweets related to particular
mood.
For example red for Anger,
orange for happy, green for
sadness.
Suppose you slept at night and mean while something unusual or
something trendy has happened overnight, and you want that if you
wake-up early in the morning then there should be something seeing
which you can get this information
IoT Solutions for 3-D Visualization of Twitter Data
Background
Method
Result
Conclusion
Future Work
2. Tasty Tweets
A jar which helps users to analyze twitter
trends through their taste buds on a
single press of a button.
Flavor of smoothie changes as the trend
changes, so every time you will get a
unique flavor
Tweets having mentions of keywords of fruits like blueberry, pineapple,
apple and carrot are collected and based on their frequency a blend of
smoothie are made which represents the trending graph for tweets
IoT Solutions for 3-D Visualization of Twitter Data
Background
Method
Result
Conclusion
3. Twitter Enabled Coffee Pot
This device helps you to instruct
your coffee maker to make a
coffee at anytime and anywhere
IoT Solutions for 3-D Visualization of Twitter Data
Future Work
Background
Method
Result
Conclusion
Future Work
› 3-D data visualization extends visual possibilities for data analysis
and data scientists
› It is an attempt to enrich IoT market with smart devices and even
help the social data analyst to dig the new visualization paradigm
IoT Solutions for 3-D Visualization of Twitter Data
Background
Method
Result
Conclusion
Future Work
› Hope this research will boost new and innovative IoT solutions using
social analysis a key research background.
IoT Solutions for 3-D Visualization of Twitter Data
EVENT PHOTO MINING FROM TWITTER
USING KEYWORD BURSTS AND IMAGE
CLUSTERING
Takamu Kanekoa, Keiji Yanaia
a)Department of Informatics, The University of Electro-Communications, Tokyo, Japan.
2015 Neurocomputing, Elsevier
Background
Method
Result
Conclusion
Future Work
› Twitter is a microblogging service which enables people to post and
read not only short messages but also photos from anywhere
› By using such distributed cameras effectively, we can understand
what kind of events happen over the world at this moment visually
and intuitively
› Twitter has not been explored extensively yet, because the amount of
Tweet photo data is too huge to collect and process in general.
› So that their visual analysis including features extraction and
clustering naturally becomes computationally expensive.
Event Photo Mining from Twitter Using Keyword Bursts and Image Clustering
Background
Method
Result
Conclusion
Future Work
› Event Keyword Detection
Detect event keyword candidates which frequently appear in the tweets
posted specific areas in specific days.
› Keyword Unification and Concatenation
Unify and concatenate the detected event keywords
› Event Photo Clustering and Representative Photo Selection
Select geo-tweet photos corresponding to the event keywords by image
clustering. Select a representative photo to each event. Show the
detected events with their representative photos on the map
Event Photo Mining from Twitter Using Keyword Bursts and Image Clustering
Background
Method
Example of representative photos selected incorrectly
(above), by proposed system (below), in US
Result
Conclusion
Future Work
Example of representative photos selected incorrectly
(above), by proposed system (below), in Japan
Event Photo Mining from Twitter Using Keyword Bursts and Image Clustering
Background
Method
Some detected events in US
Result
Conclusion
Future Work
Some detected events in Japan
Event Photo Mining from Twitter Using Keyword Bursts and Image Clustering
Background
Method
Result
Conclusion
Future Work
› The system enables us to discover and understand events visually.
› By integrating the proposed system with the Twitter Streaming API, it can
be expanded into a real-time event photo detection system.
› This research made visual event detection experiments on two largescale datasets: Japan geo-photo tweet dataset and US geo-photo
tweet dataset
Event Photo Mining from Twitter Using Keyword Bursts and Image Clustering
Background
Method
Result
Conclusion
Future Work
› To propose more sophisticated visual event mining methods which integrate
visual, textual, and location information more closely and more
comprehensively.
› Planning to analyze the difference between Tweet photos and Flickr
photos in terms of their characteristic.
Event Photo Mining from Twitter Using Keyword Bursts and Image Clustering
SENTICOMPASS: INTERACTIVE
VISUALIZATION FOR EXPLORING AND
COMPARING THE SENTIMENTS OF TIMEVARYING TWITTER DATA
Florence Ying Wanga, Arnaud Sallaberrya, Karsten Kleinb, Masahiro Takatsukac, Mathieu Rochea,
a)LIRMM & Universite de Montpellier, France
c)Monash University, Australia
d)The University of Sydney, Australia
2015 IEEE Pacific Visualization Symposium
Background
Method
Result
Case Study
Conclusion
› Through posting microblogs, people are able to express and share
their opinions almost instantly.
› Twitter data can be used as an important resource for monitoring
people’s attitudes on global events such as sports matches or
political elections.
› Representations of sentiments and temporal information still hard
to be visualized.
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
› Hybrid Approach for Sentiment Analysis
We adopt traditional classification named Affective Norm for
English Words (ANEW) dictionary to estimate both dimensions of
sentiment: valence and arousal.
Meanwhile, to compensate the affective dictionary approach, we
calculate the polarity (i.e. negative or positive meaning) of the
tweet using Naïve Bayes classifier.
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
› The Visual Metaphor of SentiCompass
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
Elementary Task (ET)
No
Requirement
ET1
Visualize the number of tweets and the strength of a sentiment at a given time segment
ET2
Compare the number of tweets and strength for different sentiments at a given time segment
Synoptic Task (ST)
No
Requirement
ST1
Visualize both dimensions of sentiments (i.e. valence and arousal) and be able to look up their
semantic meanings
ST2
Visualize and be able to compare the distances of different sentiments in affective space
ST3
Visualize and be able to compare the volume of tweets over time
ST4
Visualize the temporal variations of sentiment patterns on one specific topic
ST5
Visualize the dominant sentiments of one specific topic at different time frames
ST6
Compare temporal sentiments variations of different topics at different time frames
ST7
For tweets on one or more topics, find out if any time frames have similar sentiment patterns
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
SentiCompass of 122.393 tweets collected during 2013 Australian election period. The election day is highlighted
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
Background
Method
Result
Case Study
Conclusion
› This research combines the circumplex model of affect with the time
tunnel representation
› These two case studies show the effectiveness of SentiCompass in
achieving various tasks related to temporal sentiment and affective
analysis of tweets
SentiCompass: Interactive Visualization for Exploring and Comparing the Sentiments
of Time-Varying Twitter Data
NETWORK-BASED VISUALIZATION OF
OPINION MINING AND SENTIMENT
ANALYSIS ON TWITTER
Alemu Molla, Yenewondim Biadgie, Kyung-Ah Sohn
Department of Computer Engineering, Ajou University, South Korea
2014 IEEE IT Convergence and Security (ICITCS) International Conference
Background
Method
Result
Conclusion
Future Work
› Visualizing the result of users’ opinion mining on twitter using social
network graph can play a crucial role in decision-making, especially
for companies
› The edges in the social network may reflect different kinds of relations
among people: friendship, cooperation, contact, conflict, etc.
› It would be useful for companies to trace the overall opinion trend in
the social network as well as to identify key players.
› Available data visualizing tools, such NodeXL, use a specific file format
as an input to construct and visualize the social network graph. Also
available open source libraries for machine learning
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
Background
Method
Result
Conclusion
Future Work
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
Background
Method
Result
Conclusion
Future Work
General Social Graph
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
Background
Method
Result
Conclusion
Future Work
Social Graph with Time Stamp
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
Background
Method
Result
Conclusion
Future Work
Social Graph with Sentiment Score
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
Background
Method
Result
Conclusion
Future Work
› This research focused on the opinions contained in the companies’
official twitter account
› The main purpose of this system is to identify opinion in the tweet
and classify them automatically into positive, negative, and neutral
sentiments and then visualize their communication network
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
Background
Method
Result
Conclusion
Future Work
› This research did not include the emoticons of users, which may be
helpful for understanding the behavior of users.
› The researchers would focus on the inclusion of emoticons and
location information on each tweet.
Network-based Visualization of Opinion Mining and Sentiment Analysis on Twitter
THANKS FOR ATTENTION