Academia.eduAcademia.edu

Figure 26 – uploaded by Shamanth Kumar

See full PDFdownloadDownload figure
Figure 6.5: A Comparison of the Weighted AUC Score tistically significant, we conduct the Wilcoxon-Signed Rank test. As the results are

Figure 6 5: A Comparison of the Weighted AUC Score tistically significant, we conduct the Wilcoxon-Signed Rank test. As the results are

Related Figures (44)

Figure 2.1: TweetTracker: The central pane is focused on geographic region (blue
Figure 2.1: TweetTracker: The central pane is focused on geographic region (blue
Figure 2.2: Search interface in TweetTracker. Search parameter choices are visible at
Figure 2.2: Search interface in TweetTracker. Search parameter choices are visible at
Figure 2.3: TweetXplorer: The top-left panel shows the User x User retweet network.
Figure 2.3: TweetXplorer: The top-left panel shows the User x User retweet network.
Figure 2.4: A Closer Look of the Keywords Used to Generate the Views bottom-right panel shows the grouped keyword queries.
Figure 2.4: A Closer Look of the Keywords Used to Generate the Views bottom-right panel shows the grouped keyword queries.
Figure 2.5: Retweet Network Identifying the Direction and Involved Nodes in Infor-
Figure 2.5: Retweet Network Identifying the Direction and Involved Nodes in Infor-
Figure 2.6: Parameters Used to Collect Tweets Related to Hurricane Sandy
Figure 2.6: Parameters Used to Collect Tweets Related to Hurricane Sandy
Figure 2.7: Traffic Trend for Severely Affected Areas of Hurricane Sandy regions on Sandy’s path. In Figure 2.7, I present a comparison of the traffic from
Figure 2.7: Traffic Trend for Severely Affected Areas of Hurricane Sandy regions on Sandy’s path. In Figure 2.7, I present a comparison of the traffic from
Figure 2.10: Analyzing Discussion During the Recovery Phase of Hurricane Sandy
Figure 2.10: Analyzing Discussion During the Recovery Phase of Hurricane Sandy
Table 3.1: Major Earthquakes in 2011 and 2012 Evaluating Scalability
Table 3.1: Major Earthquakes in 2011 and 2012 Evaluating Scalability
Table 3.2: | Efficiency of | Fivent Detection: Earthquake Quality of Detected Events
Table 3.2: | Efficiency of | Fivent Detection: Earthquake Quality of Detected Events
Table 3.3: ] Fivents Detected in the ] Marthquake Dataset To quantify the effectiveness of our approach in detecting events, I compute the
Table 3.3: ] Fivents Detected in the ] Marthquake Dataset To quantify the effectiveness of our approach in detecting events, I compute the
Figure 3.1: A Comparison of the Tweet Collection Rate and the Tweet Processing approach. Here I set H; = 6.3 due to the larger volume.
Figure 3.1: A Comparison of the Tweet Collection Rate and the Tweet Processing approach. Here I set H; = 6.3 due to the larger volume.
Table 3.4: Events Detected in the Random Stream large amount of noise.
Table 3.4: Events Detected in the Random Stream large amount of noise.
Table 3.5: ] Efficiency of Threading Technique: Earthquake technique was 0.64 compared to 0.77 for the proposed approach. As the Earthquake dataset was the smallest among our datasets, the results show that the framework
Table 3.5: ] Efficiency of Threading Technique: Earthquake technique was 0.64 compared to 0.77 for the proposed approach. As the Earthquake dataset was the smallest among our datasets, the results show that the framework
Table 3.6: ] Fvolution of the Crisis ] Event on April 11, 2012
Table 3.6: ] Fvolution of the Crisis ] Event on April 11, 2012
Table 4.1: Parameters Used to Collect the Tweets roblem Statement. Given a social media site, and an event EF, let C be the content associated with E' and U be a set of corresponding users; find “informatio
Table 4.1: Parameters Used to Collect the Tweets roblem Statement. Given a social media site, and an event EF, let C be the content associated with E' and U be a set of corresponding users; find “informatio
Table 4.2: Characteristics of the Arab Spring Dataset the course of 7 months starting from February 1, 2011 to August 31, 2011. A ful
Table 4.2: Characteristics of the Arab Spring Dataset the course of 7 months starting from February 1, 2011 to August 31, 2011. A ful
Their Selection. Table 4.3: Sample of Words from a Subset of Topics in Tunisia with Justification for Tweets can be considered as small documents of length at most 140 characters. The topic of discussion of the tweets can be manually labeled as one of several topics
Their Selection. Table 4.3: Sample of Words from a Subset of Topics in Tunisia with Justification for Tweets can be considered as small documents of length at most 140 characters. The topic of discussion of the tweets can be manually labeled as one of several topics
Figure 4.1: User Visualization of Geo-Relevancy and Topic Affinity for a Topic in the crisis region and a tweet_loc; = 0 indicates that the user was outside or that the location information was missing. Then, we can compute the geo-relevancy
Figure 4.1: User Visualization of Geo-Relevancy and Topic Affinity for a Topic in the crisis region and a tweet_loc; = 0 indicates that the user was outside or that the location information was missing. Then, we can compute the geo-relevancy
Random Users UpRgng. All Results are Extremely Statistically Significant with p-value Table 4.4: Comparison of the Quantity of Tweets Generated by Q1 Users and a set of
Random Users UpRgng. All Results are Extremely Statistically Significant with p-value Table 4.4: Comparison of the Quantity of Tweets Generated by Q1 Users and a set of
Table 4.5: The Divergence of Q1 Users from the General Topic MWe can generalize the distance to vector format using any vector norm. Here, I use the L4-norm,
Table 4.5: The Divergence of Q1 Users from the General Topic MWe can generalize the distance to vector format using any vector norm. Here, I use the L4-norm,
Figure 4.2: Group embedding (using Isomap) of influentials and Q1 users. Th
Figure 4.2: Group embedding (using Isomap) of influentials and Q1 users. Th
tistically Significant with p-value < 0.0001 Table 4.6: Evaluation of Tweet Quantity by Q1 and Followers. All Results are Sta- make this comparison,
tistically Significant with p-value < 0.0001 Table 4.6: Evaluation of Tweet Quantity by Q1 and Followers. All Results are Sta- make this comparison,
Figure 5.1: Tumblr Profile of BBC News
Figure 5.1: Tumblr Profile of BBC News
Table 5.1: Dataset Characteristics unpopular due to the absence of an explicit dislike function in Tumblr. From this
Table 5.1: Dataset Characteristics unpopular due to the absence of an explicit dislike function in Tumblr. From this
Figure 5.2: Distribution of likes in Tumblr blogs in log-log scale. The y-axis represents
Figure 5.2: Distribution of likes in Tumblr blogs in log-log scale. The y-axis represents
Figure 5.3: Effect of the Number of Special Characters on Popularity
Figure 5.3: Effect of the Number of Special Characters on Popularity
Figure 5.4: Effect of Numeric Sequence Length on Popularity users with blognames containing numeric sequences greater than 8 characters hac
Figure 5.4: Effect of Numeric Sequence Length on Popularity users with blognames containing numeric sequences greater than 8 characters hac
Figure 5.5: Effect of the Length of the Username on Popularity
Figure 5.5: Effect of the Length of the Username on Popularity
Figure 5.6: Effect of the Length of the Blogname on Popularity
Figure 5.6: Effect of the Length of the Blogname on Popularity
Figure 5.7: Effect of the Number of Unique Characters on Popularity
Figure 5.7: Effect of the Number of Unique Characters on Popularity
Figure 5.8: Effect of Similarity Between Blogname and Username on Popularity DFI as the dot product between the distribution of two texts
Figure 5.8: Effect of Similarity Between Blogname and Username on Popularity DFI as the dot product between the distribution of two texts
Table 6.2: Dataset Characteristics
Table 6.2: Dataset Characteristics
Figure 6.1: Distribution of Client Usage in Crisis Data
Figure 6.1: Distribution of Client Usage in Crisis Data
Figure 6.2: Crisis Data Generated from the Top 100 Clients
Figure 6.2: Crisis Data Generated from the Top 100 Clients
Table 6.3: Behavioral Characteristics in Tweets: I (* Indicates p-value < 0.05 and * more likely to have access to information that users outside the crisis regions may not
Table 6.3: Behavioral Characteristics in Tweets: I (* Indicates p-value < 0.05 and * more likely to have access to information that users outside the crisis regions may not
Specifically, we attempt to answer the following questions:
Specifically, we attempt to answer the following questions:
Table 6.5: Behavioral Characteristics in Tweets: III (* indicates p-value < 0.05 and
Table 6.5: Behavioral Characteristics in Tweets: III (* indicates p-value < 0.05 and
Figure 6.3: Average Probability of Tweets in Various Crises Data
Figure 6.3: Average Probability of Tweets in Various Crises Data
Figure 6.4: A Comparison of the F, Score
Figure 6.4: A Comparison of the F, Score
Table 6.6: Arizona Wildfires Summary Created Using Z; crisis. In this case study, we will use the task of event summarization to demonstrate that the application of our approach enables the generation of a more meaningful
Table 6.6: Arizona Wildfires Summary Created Using Z; crisis. In this case study, we will use the task of event summarization to demonstrate that the application of our approach enables the generation of a more meaningful
Related topics:
Crisis Response
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved