A correlation will be found between significant events during the COVID-19 pandemic and fluctuations in the volume of misinformation and counter-misinformation tweets [Lanier et al. 2022].

The specific words that Twitter users use in their tweets will be related to the text classification, distinguishing between misinformation and counter-misinformation [Alsmadi et al. 2022].

Users who frequently post misinformation are more likely to have created their accounts shortly before they began spreading this information, suggesting that these accounts may have been created specifically for the purpose of disseminating misinformation [Ferrara et al. 2020].

Time-based Data Analysis: The timeline chart illustrates the number of tweets labeled as "misinformation", "counter-misinformation", and "neutral" from early 2020 to the end of 2022.

In the timeline graph, we can see a significant rise in misinformation tweets, peaking towards mid-2021. This surge may correspond to significant global events, policy changes, or misleading advertising campaigns during this period. Decline after peak: Following the peak in late 2021, there's a gradual decrease in the number of misinformation tweets, although smaller spikes continue to occur occasionally. Additionally, counter-misinformation tweets show a steady increase along the timeline, starting from a lower base, indicating a growing effort to address false information. In contrast to misinformation, the volume of counter-misinformation tweets remains relatively consistent without significant peaks, suggesting an ongoing effort rather than sharp increases.

Word clouds between the two classifications. There are clear characteristic differences in content between tweets that support, oppose, and present conspiracy theories about COVID-19, which can be identified and analyzed. We conducted a word cloud analysis to illustrate the most common terms used in misinformation tweets and counter-misinformation tweets.

Results: The word cloud for misinformation includes words that spread doubt and negative feelings about vaccines and COVID-19, such as "fake", "fraud", and "ineffective".

In contrast, the word cloud for counter-misinformation includes words used to debunk false claims and promote factual information, such as "work", "vaccinated", and "news". The main difference lies in the intention: misinformation aims to sow distrust and fear, while counter-misinformation focuses on providing accurate and supportive content.

The analysis of words in the cloud clearly distinguishes the language used in misinformation tweets and counter-misinformation tweets, supporting the hypothesis.

Box Plot graph representing the time differences between Twitter user registrations according to their classification.

We collected information on the 40 users from the 2 classifications we are examining (20 for each classification) who tweeted the highest number of tweets. We then checked when they registered on Twitter, using additional data we had with user details. We then plotted the data on a Box Plot graph to show the range of user registrations on Twitter by date (represented on the Y-axis by numbers, where higher numbers mean more recent dates and lower numbers mean earlier dates). In this graph, we can see a very large upper part for "misinformation" users, which may indicate that some of the users were created after the start of the COVID-19 period. After this graph, we examined in detail how many users out of the 20 were created after 2020. In "misinformation", the amount was 25%, and in "counter-misinformation", the amount was 10%.