Academia.eduAcademia.edu

Figure 5 – uploaded by Diana Contreras

See full PDFdownloadDownload figure
Figure 5. TA results: a) tweet level and b) sentence level. Table 3. TA results at the tweet and a sentence level The SA results at the tweet level for the text data commenting only on recovery have the majority of positive polarity, as it is depicted in Figure 6a. The result at the sentence level, having recovery as the main topic, reports also majority of positive polarity in Figure 6b.

Figure 5 TA results: a) tweet level and b) sentence level. Table 3. TA results at the tweet and a sentence level The SA results at the tweet level for the text data commenting only on recovery have the majority of positive polarity, as it is depicted in Figure 6a. The result at the sentence level, having recovery as the main topic, reports also majority of positive polarity in Figure 6b.

Related Figures (9)

We evaluated Nepal's recovery progress based on text data posted on Twitter and Instagram from March 5th to May 2nd, 2020. This period was selected based on SM monitoring and included dates before and after the fifth anniversary of the Gorkha earthquake: April 25th, 2020, 11:56:25 Nepal standard time (NST) (06:11: UTC) (Wilkinson et al., 2019). The methodology comprises six steps, as described in Figure 1.
We evaluated Nepal's recovery progress based on text data posted on Twitter and Instagram from March 5th to May 2nd, 2020. This period was selected based on SM monitoring and included dates before and after the fifth anniversary of the Gorkha earthquake: April 25th, 2020, 11:56:25 Nepal standard time (NST) (06:11: UTC) (Wilkinson et al., 2019). The methodology comprises six steps, as described in Figure 1.
Sentiment and topic analysis can be performed at three primary classification levels: document, sentence, and sub-sentence (MonkeyLearn, 2020). We performed the analyses at the tweet and sentence levels for this work. The recovery assessment can be done exclusively focused on SA. However, without the TA, it would not be possible to determine what is positive or negative in the recovery process. Then another option will be to check the sentiment analysis results per topic. The recovery process can be assessed based solely on the polarity of text data addressing this topic. This refined analysis can be done by combining the SA and TA at the tweet level or the TA at the tweet level, Table 1. Polarity classification rule set for SA.
Sentiment and topic analysis can be performed at three primary classification levels: document, sentence, and sub-sentence (MonkeyLearn, 2020). We performed the analyses at the tweet and sentence levels for this work. The recovery assessment can be done exclusively focused on SA. However, without the TA, it would not be possible to determine what is positive or negative in the recovery process. Then another option will be to check the sentiment analysis results per topic. The recovery process can be assessed based solely on the polarity of text data addressing this topic. This refined analysis can be done by combining the SA and TA at the tweet level or the TA at the tweet level, Table 1. Polarity classification rule set for SA.
Hashtags identified in tweets related to the anniversary were: #Nepal, #earthquake, #Nepalearthquake, #NepalQuake, #5yearson, and #Nepali. The third-party vendor provided 183,361 tweets, 174,641 original (95%), and 8,720 (4.7%) retweets. The number of tweets containing hashtags mentioned before and the activity on Twitter during the observation period with those hashtags are plotted in Figure 2.
Hashtags identified in tweets related to the anniversary were: #Nepal, #earthquake, #Nepalearthquake, #NepalQuake, #5yearson, and #Nepali. The third-party vendor provided 183,361 tweets, 174,641 original (95%), and 8,720 (4.7%) retweets. The number of tweets containing hashtags mentioned before and the activity on Twitter during the observation period with those hashtags are plotted in Figure 2.
Figure 3. Tweets posted around the exact time of the anniversary of the Gorkha earthquake: a) 11:56:25 NST /06:11:25 UTC/07:11:25 CET and b) 11:56:25 UTC Tweets posted around the exact time of the 5" anniversary of the 2015 Gorkha earthquake in Nepal are presented in Figure 3 a) and b), respectively.
Figure 3. Tweets posted around the exact time of the anniversary of the Gorkha earthquake: a) 11:56:25 NST /06:11:25 UTC/07:11:25 CET and b) 11:56:25 UTC Tweets posted around the exact time of the 5" anniversary of the 2015 Gorkha earthquake in Nepal are presented in Figure 3 a) and b), respectively.
The TA at the tweet level indicates that 1011 tweets were unrelated (50%). Apart from those tweets 329 (16%) reported about COVID-19, 286 (14%) discussed restoration progress, 91 (5%) mentione commemoration actions, 85 (4%) emergency response activities, 51 (3%) discussed topics related t governance, only 51 (3%) reported about reconstruction, 22 tweets contains solidarity messages (1% and 18 (1%) emotional distress signals. Other topics discussed in the tweets included in the sample ar listed in Table 3 and plotted in Figure 5a). The topic analysis at the sentence level indicates that 108 sentences were unrelated (29%), 489 discussed restoration progress (14%), 481 (13%) reported abot COVID-19, and 267 (7%) sentences included seismic information. Emergency response actions wer mentioned in 242 (7%) sentences, Injured population and casualties were remembered in 151(4% sentences, and commemoration actions were mentioned in 129 sentences (4%). Governanc capabilities are discussed in 125 (8%) sentences, and solidarity messages are included in 111 (8% sentences. Availability of urban facilities is discussed in 94 (8%) sentences, and progress in th reconstruction in 81 (2%) sentences. The topic of intensity is reported in 69 (2%) sentences, as well a preparedness in 39 (1%) of the sentences. Other topics commented on sentences included in th sample are listed in Table 3 and plotted in Figure 5b).
The TA at the tweet level indicates that 1011 tweets were unrelated (50%). Apart from those tweets 329 (16%) reported about COVID-19, 286 (14%) discussed restoration progress, 91 (5%) mentione commemoration actions, 85 (4%) emergency response activities, 51 (3%) discussed topics related t governance, only 51 (3%) reported about reconstruction, 22 tweets contains solidarity messages (1% and 18 (1%) emotional distress signals. Other topics discussed in the tweets included in the sample ar listed in Table 3 and plotted in Figure 5a). The topic analysis at the sentence level indicates that 108 sentences were unrelated (29%), 489 discussed restoration progress (14%), 481 (13%) reported abot COVID-19, and 267 (7%) sentences included seismic information. Emergency response actions wer mentioned in 242 (7%) sentences, Injured population and casualties were remembered in 151(4% sentences, and commemoration actions were mentioned in 129 sentences (4%). Governanc capabilities are discussed in 125 (8%) sentences, and solidarity messages are included in 111 (8% sentences. Availability of urban facilities is discussed in 94 (8%) sentences, and progress in th reconstruction in 81 (2%) sentences. The topic of intensity is reported in 69 (2%) sentences, as well a preparedness in 39 (1%) of the sentences. Other topics commented on sentences included in th sample are listed in Table 3 and plotted in Figure 5b).
Figure 4. Sentiment analysis results a) tweet level and b) sentence level.
Figure 4. Sentiment analysis results a) tweet level and b) sentence level.
Related topics:
Sentiment AnalysisCommemoration (Memory Studies)Commemoration and MemoryPost Disaster ReconstructionTopical AnalysisPost Disaster RecoverySocial Media Sentiment AnalysisDisaster RecoveryOPINION MINING AND SENTIMENT ANALYSISSocial Media and Sentiment Analysis
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved