Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Earthquake Shakes Twitter Users: Real-time Event Detection by Social Sensors

WWW 2010 paper
by

Sunwon Lee

on 22 July 2010

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Earthquake Shakes Twitter Users: Real-time Event Detection by Social Sensors

Earthquake shakes Twitter Users:
Real-time Event Detection by Social Sensors Takeshi Sakaki, Makoto Okazaki and Yutaka Matsuo
WWW 2010 Abstract A popular microblogging service An online social network Twitter asks only one question What's happening? the answer for "what's happening?" Tweet? less than 140 chracters The important characteristic of microblogging service Real-time This paper presents an investigation of the real-time nature of Twitter
and proposes and event notification promptly If some events are occured... "Earthquake!!!" "Now it is shaking!!!" "It's heavy raining!!" "Is it a typhoon?!" Many tweets are posted
"Shaking!!!" "What the f--------" IDEA From these tweets, some events can be notified in real-time Event Detection Typhoon Traffic jam
Target events Authors target events such as.... Earthquake Semantic Analysis on Tweet "Earthquake!!" VS. "I'm attending an
Earthquake Conference" It is necessary to clarify that a tweet is truly referring to
an actual earthquake occurrence Authors prepare 3 groups of features Features A. The number of words in a tweet message,
& position of query word
Features B. The words in a tweet
Features C. The words before and after the query word These features are used in SVM Tweet as a Sensory Value Assumption 2.1 Each Twitter user is regarded as a sensor.
A sensor detects a target event and makes a report probabilistically. Assumption 2.2 Each tweet is associated with a time and location, which is a set of latitude and longitude. Model Temporal Model Spatial Model When a target event occurs,
How can the sensors detect the event? ii. influence people's daily life
i. large scale iii. have spatial and temporal regions First, authors examined the actual data The distribution is apparently an exponential distribution If a user detects an event at time 0, assume that the probability of his posting a tweet in time window T is fixed as lambda. To assess an alarm, the reliability of
multiple sensor values must be calculated. Because of false alram or misclassification of classifier * The false-positive ratio of a sensor is approximately 0.35 * Sensors are assumed to be independent and identically distributed Assuming that we have n sensors The probability of all n sensors returning a false alarm is Therefore, the probability of event occurrence is At time t, the sensors that alarm are Therefore, the number of sensors we expect at time t is Consequently, the probability of an event occurrence at time t is Each tweet is associated with a location.
How we can estimate the location of an event from sensor readings? To define the problem of location estimation,
authors consider the evolution of the state sequence of target. The objective of tracking is to estimate x recursively from measurements where n is a measurement noise sequence
Presuming that is available,
the prediction stage uses the following equation. And some complex equations..... Kalman Filters The Kalman Filter assumes that the posterior density at every time step is Gaussian. It is parameterized using a mean and covariance. Case 1. Location estimation of an earthquake center Case 2. Trajectory estimation of a typhoon Particle Filters A particle filter is a probabilistic approximation algorithm
implementing a Bayes filter, and a member of
the sequential Monte Carlo methods For location estimation, it maintains a probability distribution
for the location estimation at time t, designated as the belief Information diffusion
related to a real-time event Experiments
&
Evaluation Evaluation by Semantic Analysis Evaluation of spatial estimation Earthquake location estimation Typhoon trajectory estimation Earthquake Reporting system "Toretter" Earthquake information is much more valuable if it is received in real time If we have several seconds.... We can shut off gas... We can hide our selves under a desk Vast amounts of work have been done on intermediate-term earthquake prediction Various attempts have also been made to produce short-term
forecasts to realize an earthquake warning system In Japan,
an earthquake early warning service has been operated by JMA since 2007 Toretter It means "we have taken it" in Japanese The system keeps watching tweets... When earthquake occurs.. The system sends e-mail to users Toretter alerted earthquakes in a minutes But, JMA announced 6 mins after earthquake occurs. Toretter VS. JMA Torreter WIN!!!! Discussion Authors says that they plan to expand their system to detect events of various kinds using twitter RAINBOW!! Discussion The model proposed in this paper includes the assumption that
a single instance of the target event exists. No two or more earthquakes or typhoons occur simultaneously This assumption is not reasonable for another target event such as
traffic jams or rainbows And it is necessary to obtain a good set of queries Conclusion Thank you
Full transcript