arXiv subjects

Md. Rafiul Biswas

Publications and source records attributed to Md. Rafiul Biswas.

21 records · Page 2Linked to original sources

arXiv2023

Potentials of ChatGPT for Annotating Vaccine Related Tweets

This study evaluates ChatGPT's performance in annotating vaccine-related Arabic tweets by comparing its annotations with human annotations. A dataset of 2,100 tweets representing various factors contributing to vaccine hesitancy was examined. Two domain experts annotated the data, with a third resolving conflicts. ChatGPT was then employed to annotate the same dataset using specific prompts for each factor. The ChatGPT annotations were evaluated through zero-shot, one-shot, and few-shot learning tests, with an average accuracy of 82.14%, 83.85%, and 85.57%, respectively. Precision averaged around 86%, minimizing false positives. The average recall and F1-score ranged from 0.74 to 0.80 and 0.65 to 0.93, respectively. AUC for zero-shot, one-shot, and few-shot learning was 0.79, 0.80, and 0.83. In cases of ambiguity, both human annotators and ChatGPT faced challenges. These findings suggest that ChatGPT holds promise as a tool for annotating vaccine-related tweets.

cs.SI↗

arXiv2022

Influences of social media usage on public attitudes and behavior towards COVID-19 vaccine in the Arab world

Background: Vaccination programs are effective only when a significant percentage of people are vaccinated. However, vaccine acceptance varies among communities around the world. Social media usage is arguably one of the factors affecting public attitudes towards vaccines. Objective: This study aims to identify if the social media usages factors can be used to predict attitudes and behavior towards the COVID-19 vaccines among the people in the Arab world. Methods: An online survey was conducted in the Arab countries and 217 Arab people participated in this study. Logistic regression was applied to identify what demographics and social media usage factors predict public attitudes and behavior towards the COVID-19 vaccines. Results: Of the 217 participants, 56.22% of them were willing to accept the vaccine and 41.47% of them were hesitant. This study shows that none of the social media usages factors were significant enough to predict the actual vaccine acceptance behavior. Whereas the analysis showed few of the social media usage factors can predict public attitudes towards the COVID-19 vaccines. For example, frequent social media users were 2.85 times more likely to agree that the risk of COVID-19 is being exaggerated (OR=2.85, 95% CI=0.86-9.45, p=0.046) than infrequent social media users. Whereas participants having more trust in vaccine information shared by their contacts are less likely to agree that decision-makers have verified that vaccines are safe (OR=0.528, 95% CI= 0.276-1.012, p=0.05). Conclusion: The use of social media and information shared on it may affect public attitudes towards COVID-19 vaccines. Therefore, disseminating correct and validated information about COVID-19 and other vaccines on social media is important for increasing public trust and countering the impact of incorrect and misinformation.

cs.SI↗

arXiv2021

Characterising Communities of Twitter Users Who Posted Vaccines Related Tweets by Information Sources

Objective: We formed community structure based on web page credibility and we measured the types of information for characterizing communities of tweeter users who posted about tweets related to vaccine. Methods: We performed the experiment on only Twitter data (tweets) regarding vaccine. The duration of data collection was between 17 January 2017 and 14 March 2018. We formulated cluster based on the information on its contents and sources it resides (i.e., website domains). We only focused the topics which were related to vaccine. To detect the structure and network of community, we applied Louvain community algorithm along with Random walks called Info map method over vaccines related tweeter user. We defined the communities based on various measures derived from the information shared by Twitter users. Representations and visualizations of the communities based on these derived measures help the public health organization to make understand the possible cause of the rejecting the safety and efficacy of vaccines. Results: To analyse people perception over social media, we downloaded 6,591,566 tweets. We distinguished 1,860,662 users who were posting related to vaccine. We applied Louvain community detection algorithm along with DMM values and we found 192 communities. It also produced higher alignment values as long as the number of topics were low. With the increase of topics number, the alignment becomes lower. The total tweets were divided into two parts based on their characterizes. The first category contained 163,148 (57.16%) tweets which were based as evidence and advocacy and the second category contained 6244 (2.19%) tweets which were based on the experiences and opinion. We observed 4548 users were posting about experiential tweets about vaccine and of them 3449 users (75.84%) were posted evidence and advocacy related.

cs.SI↗