Tracking the 2024 US Presidential Election Chatter on TikTok: A Public Multimodal Dataset DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation emai; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/18/06
Gabriela Pinto 1 , Charles Bickham 1 , Tanishq Salkar 1 , Joyston Menezes 1 , Luca Luceri 1 , Emilio Ferrara 1 · 2024
Read Tracking the 2024 US Presidential Election Chatter on TikTok: A Public Multimodal Dataset DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation emai; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/18/06 by Gabriela Pinto 1 , Charles Bickham 1 , Tanishq Salkar 1 , Joyston Menezes 1 , Luca Luceri 1 , Emilio Ferrara 1 (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-15583.txt.
Introduction
Gabriela Pinto and five co-authors release a public dataset of TikTok videos about the 2024 United States presidential election, in a 2024 paper. The dataset holds 3.14 million videos posted between November 2023 and mid-October 2024, with video identifiers and transcripts. Data came through the TikTok research interface using a wide set of election keywords and hashtags. Outside tools filled gaps left by that interface.
With the data, researchers can map networks of hashtags that cluster by political leaning. They can also track how top hashtags rose and fell across the campaign. Summary statistics show the scale of the collection. The resource supports studies of how a video platform shapes election talk, and it is open for other teams to use.
Catalog record
| CREATOR | Gabriela Pinto 1 , Charles Bickham 1 , Tanishq Salkar 1 , Joyston Menezes 1 , Luca Luceri 1 , Emilio Ferrara 1 |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-15583
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Tracking the 2024 US Presidential Election Chatter on TikTok: A Public Multimodal Dataset DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation emai; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/18/06 Gabriela Pinto 1 , Charles Bickham 1 , Tanishq Salkar 1 , Joyston Menezes 1 , Luca Luceri 1 , Emilio Ferrara 1 Abstract . This paper presents the TikTok 2024 U.S. Presidential Election Dataset, a large-scale, resource designed to advance research into political communication and social media dynamics. The dataset comprises 3.14 million videos published on TikTok between November 1, 2023, and October 16, 2024, encompassing video ids and transcripts. Data collection was conducted using the TikTok Research API with a comprehensive set of election-related keywords and hashtags, supplemented by third-party tools to address API limitations and expand content coverage, enabling analysis of hashtag co-occurrence networks that reveal politically aligned hashtags based on ideological affiliations, the evolution of top hashtags over time, and summary statistics that highlight the dataset’s scale and richness. This dataset offers insights into TikTok’s role in shaping electoral discourse by providing a multimodal view of election-related content. It enables researchers to explore critical topics such as coordinated messaging, misinformation spread, audience engagement, and linguistic trends. The TikTok 2024 U.S. Presidential Election Dataset is publicly available and aims to contribute to the broader understanding of social media’s impact on democracy and public opinion. 1. Introduction Social media has profoundly reshaped electoral politics, emerging as a critical medium for disseminating election-related information and influencing public opinion. This shift has garnered significant attention from researchers ( Jungherr 2016 ; Kratzke 2017 ; Deb et al. 2019 ; Luceri et al. 2019 ; Abilov et al. 2021 ) . Although platforms like Twitter have long been a cornerstone in the study of global geopolitical events ( Davis et al. 2016 ; Chen et al. 2022 ; Chen and Ferrara 2023 ) , TikTok offers a distinctive perspective on political discourse. Its short-form video format, combined with interactive features like comments, hashtags, and trends, enables rapid content dissemination and highly visual engagement, making it uniquely suited for studying political communication. With an estimated global user base exceeding one billion, 1 1 1 https://whatsthebigdata.com/tiktok-statistics/ TikTok has quickly become a platform where political campaigns engage and inform audiences, particularly adolescents ( Montag et al. 2021 ) . The 2024 U.S. Presidential Election highlighted TikTok’s growing influence in electoral politics, as major candidates actively engaged with the platform to connect with voters. For instance, Donald Trump joined TikTok in June 2024, rapidly gaining over 6 million followers, 2 2 2 https://cnn.com/2024/06/02/politics/donald-trump-joins-tiktok/index.html while Joe Biden actively posted over 200 videos starting in February 2024, amassing more than 373,000 followers. 3 3 3 https://cnn.com/2024/06/12/tech/tiktok-pew-research-politics-x/index.html These examples illustrate TikTok’s potential as both a campaign tool and a rich data source for understanding electoral sentiment, political narratives, and voter outreach strategies. Its unique format and audience demographic make TikTok an important platform for studying the evolving nature of political communication in the digital age. This paper introduces the TikTok 2024 U.S. Presidential Election Dataset, a collection of over 3 million videos that includes metadata and textual transcripts. The dataset captures the dynamics of TikTok’s election-related discourse, enabling researchers to explore key questions such as identifying coordinated messaging among political actors, detecting the spread of misinformation, examining audience engagement patterns, and analyzing discourse across linguistic and demographic groups. By leveraging this dataset, researchers can gain a deeper understanding of how political communication evolves in digital spaces, particularly on TikTok’s unique platform. It highlights TikTok’s growing influence in shaping public opinion, especially among younger voters—a pivotal demographic in modern election. Ultimately, this dataset seeks to contribute to broader discussions about the role of digital platforms in electoral processes and the implications for digital democracy in the age of social media. 2. Method of Data Collection TikTok API and Data Collection The TikTok Research API 4 4 4 https://developers.tiktok.com/doc/research-api-specs-query-videos/ provides access to both account-related and content-related data on TikTok. Account data includes user profiles, follower lists, liked videos, and reposts, while content data encompasses video descriptions, hashtags, comments, subtitles, and engagement metrics such as likes, shares, and comments. Accessing the API requires a bearer token, generated using client and secret keys, which must accompany API requests. After exploring third-party scrapers, the Official Research API proved to be the most reliable method for metadata collection. Video Metadata Collection. We used the TikTok Research API to collect metadata for videos, leveraging parameterized queries to target content by regions, topics, languages, or publication dates. Key parameters included the date range, defining the start and end dates for data collection, and fields, which specified video ID, description, creation time, region, engagement metrics (likes, views, comments, shares), music ID, hashtags, and username. Queries were restricted to 100 records per call, and advanced filters, using logical operators (e.g., AND/OR), allowed precise targeting of content based on specific hashtags and regions. For instance, videos tagged with #election2024 in the U.S. were queried using region_code=“US” and hashtag=“#election2024”. The metadata collection process was structured around phases defined by significant political events, with the project divided into several phases spanning specified date ranges. Phase 1 spans from November 1, 2023, to January 15, 2024. Phase 2 continues from January 16, 2024, to February 7, 2024, followed by Phase 3, which occurs between February 8, 2024, and February 27, 2024. Phase 4 is scheduled from February 28, 2024, to March 5, 2024. Subsequently, Phase 5 runs from March 6, 2024, to July 14, 2024, and Phase 6 takes place between July 15, 2024, and July 20, 2024. Phase 7 follows shortly after, from July 20, 2024, to August 5, 2024. Phase 8 is set for the period of August 6, 2024, to September 9, 2024, and Phase 9 will occur between September 10, 2024, and September 15, 2024. Phase 10 is scheduled from September 15, 2024, to November 10, 2024, and finally, Phase 11 begins on November 11, 2024, and continues to the present. Video Downloads and Transcript Generation To complement the collected metadata, we used PykTok 5 5 5 https://github.com/dfreelon/pyktok , a Python library designed to interact with TikTok’s platform, to download the corresponding video content. After downloading, we utilized OpenAI Whisper 6 6 6 https://github.com/openai/whisper to generate transcripts for each video. To ensure accuracy, we randomly selected a subset of 100 videos and verified the correctness of their generated transcripts. Note: The data collection process complies fully with TikTok’s Terms of Service, ensuring adherence to ethical standards. 3. Exploratory Analysis Using the collected data, we performed an exploratory analysis to identify key patterns and trends in TikTok’s election-related discourse. This analysis offers an initial glimpse into how users engaged with and discussed major political events and candidates during the 2024 U.S. Presidential Election. Summary of Data Table 1 summarizes the statistics for the current dataset. TikTok-generated transcripts account for only 15.4% of the total videos, reflecting a limitation of the TikTok Research API. To enrich the dataset with more linguistic data, we incorporate Whisper-generated transcripts to our dataset and related GitHub repository. Number of videos 3,144,836 Number of transcripts 485,343 Number of comments 161,892,802 Number of views 18,010,334,100 Number of likes 2,042,078,329 Number