Introduction

Long-standing strife plaguing the Gaza strip has been a major source of humanitarian concern, prompting decades of discursive dissection through various lenses of political science, history, and international law. The region’s conflict stems from deep-rooted tensions over the past century, competing national claims, colonial legacies, and cycles of displacement and occupation, positioned within the broader context of international geopolitics [1] and a protracted struggle over land, sovereignty and identity [2], alongside historical justice [3]. This paper contributes to existing literature from a communications perspective, adding to a robust pool of academic papers that have examined the conflict through the lens of social media. We provide a macro-overview of the most recent wave of the Israel-Hamas armed conflict, engendered by a surprise militant attack launched on 7 October 2023, resulting in the deaths of a reported 1,195 people [4], with further impacts compounded in destroyed infrastructure, severe shortage of essentials and an inability to flee to safety [5].

Few big data studies have covered this most recent onslaught of violence in the region; notably, an investigation into forum posts has yielded a marked increase in controversial comments, intensified discourse, and engagement metrics during this period [6]. Retrospective studies on online discourses covering past key events all point to the significance of social media in mobilizing international audiences and inciting digital action. For instance, various investigations into the 2012-period of the conflict find that shifts in public support on twitter reduce conflict intensity [7]; with image-based visual propaganda frames (resistance, unity, civilian casualties) co-opted to spur emotional engagement [8]. Analyses of the 2014-period of the conflict found that Israel news media primarily framed social media as spaces of hate speech and distribution of rumors; rarely portrayed as platforms to orchestrate collective action [9]; tweets from the Israeli MFA during the 2014 Gaza War co-opted linguistic frames to legitimize Israel’s policies (e.g., images to support strategic narratives of shared identities) [10]. On the flipside, other researchers found that social media itself has become the news [11]: when a celebrity expresses online sympathy with the Palestinian civilians in Gaza [12], or when the Palestinian militant organizations falsify the Israeli army story about taking civilians as human shields [13], or when the supporters of the Palestinians attack Israeli websites and media forums [14], turning the online space into a center for message transmission and the winning of support, sympathy, and public opinion domestically and internationally [15]. In that vein, analyses of the 2022-period of the conflict found that pro-Palestine accounts often heavily tweet about Israeli aggressions, constructing a positive representation of the Palestinian Self and a negative representation of the Israeli Other; a mediated discourse conveyed to international and external audiences [16].

Against this backdrop underscoring the role of social media in mediating and mediatizing conflict, we aim to analyze the initial key narratives that captured public attention during the latest Israel-Hamas armed conflict. We interrogate a corpus dataset of all English-language media content, obtained using academic-level API access across social media platforms (Twitter, Reddit, Tiktok), interrogating content containing search term keywords associated with the conflict (Palestine, Palestinian(s), Israel(i)(s), Gaza, Hamas) from October 2023 to January 2024 (n = 1,079,676,984) to ask two main research questions. First, what events were the biggest generators of buzz, proxied by mention count spikes? Second, what content trends can be evinced, using (i) quantitative tracking of sentiment valence, top keywords and emojis, and (ii) qualitative themes found across the biggest generators of media engagement topics, proxied by engagement metrics (likes, reposts)? Our first aim was achieved by running a prominent peaks model, identifying peaks falling in the upper-quartile significance threshold of prominence (P > 1,500,000), and identifying events associated with these spikes in mention count metrics. Our second aim conducted content analysis in quantitative metrics of valence, top keywords and emojis, and qualitatively identified a subset of highest-engaged posts, delving into the themes that generated the most discussion online, providing a bird’s eye view of the stories that captured the most online attention during this period.

Methods

Dataset

The dataset was created by collecting social media data across publicly-accessible textual content across Twitter (using its application programming interface standard search, publicly accessible via API V2 on the academic research access level), Reddit (using the Pushshift Reddit API adhering to fair use non-commercial academic research) and Tiktok (using the Tiktok academic research tools API). All English-language social media content containing one or more the following search term keywords: Palestine, Palestinian(s), Israel(i)(s), Gaza, Hamas, used in context as either a word or hashtag, and not case-sensitive, published between October 2023 to January 2024 were shortlisted (n = 1,079,676,984). We provide a summary of methods in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Methodological Summary: Data Collection and Analytic Plan.

https://doi.org/10.1371/journal.pone.0332746.g001

Research aim 1: prominent peaks modelling

First, we quantitatively identified prominent peak dates which experienced a spike in mention count—defined by the total number of original posts, articles, replies, and retweets containing the search term keywords—signaling a significant increment in interest and engagement in the topic. This analysis was conducted using peak prominence detection modelling (Python 3.7: SciPy 1.4.1). This Python package module operates on marking the higher of two bases as the peak’s lowest contour line; prominence calculated as the vertical difference between the peak’s height, and its lowest contour line. Essentially, the program checks for each given date’s search term keyword mention-count value (A), the lowest possible point prior to that date (B) and after that date (C): the average of these values (A-B; A-C) is defined as the date’s prominence score, a value assigned relative to the lowest possible point on either side of the chronological calendar.

A total of 8 statistically significant peaks were found during the study period ( = 1,861,378; σ = 4,235,894); of which, prominence scores of said peaks were calculated to range from 119,321 (lower quartile), to 449,288 (median), to 1,327,436 (upper quartile). Using a data-driven approach to determine an adequate threshold for significance, we selected peaks that appeared only in the upper quartile of prominence (i.e., P > 1,500,000) for further investigation, essentially, delving into the top 75% of discussion spike dates. This method yielded 8 peaks: 8 individual dates where prominence exceeded the upper quartile’s significance threshold.

Using these identified 8 upper-quartile peak dates, our second methodological step involved qualitatively sampling the top-engaged posts published on those days (i.e., a qualitative interrogation of their associated events and discussions, using a sample of the top-100 posts containing the highest engagement metrics), to identify the main themes that contributed to the day’s significant increase in discursive volume. A summary of results is presented in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Prominent Peaks of Israel-Hamas Conflict (2023) across 1 billion online posts.

https://doi.org/10.1371/journal.pone.0332746.g002

Research aim 2: content trends

Our second research aim utilized (i) Quantitative tracking of sentiment-tagging to elicit the valence scores associated with each month (October, November, December 2023, January 2024), categorizing the articles as either predominantly negative, neutral, or positive. This was conducted using a transformer-based sentiment model, a natural language processing methodology that was adapted from Google’s open-source pre-trained Bidirectional Encoder Representations from Transformer (BERT) model, which allows for deep bidirectional representations from unlabeled text through joint conditioning left and right context [17]. Each post’s polarity (positive, negative, or neutral sentiment, on the document-level) is generated through this classification method, yielding supervised sentiment-analysis modelled data [18]. Alongside that, we present the highest-frequency keywords and emojis utilized each month. Top keywords were identified by raw count with the manual exclusion of stop-words, defined as auxiliaries containing little-to-no semantic information and serving grammatical functions—for example, determiners (the, a, an, his, her, this, that), prepositions (from, to) and conjunctions (and, nor, yet, for, but, or, so) were omitted from qualifying for this list of highest-frequency keywords. Top emojis were identified by raw count attested in the dataset, and presented in order of highest count. These findings are presented in Fig 3 to provide a bird’s-eye overview of the unfolding conflict. (ii) Qualitatively, we further interrogated a subset of the highest-engaged (likes, reposts) social media posts across Twitter and Tiktok, delving into the main themes that generated the most online discussion and interactions. These qualitative findings are detailed in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Top Qualitative Themes.

https://doi.org/10.1371/journal.pone.0332746.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Valence distribution, top keywords and emojis over 4 months of the conflict.

https://doi.org/10.1371/journal.pone.0332746.g003

Results

Discussion and conclusion

We analyzed large-scale data of emergent public chatter, focusing on engagement and frequency metrics surrounding the most recent spate of conflict. Our research findings have focused on discussions around social media involvement and engagement trends, finding a worrying inclination toward news of violence (e.g., airstrikes, citizen harm) the circulation of gory footage. This trend is significant against a backdrop of growing research in the representation and circulation of violence in social media. Digital technologies have facilitated real-time sharing of unfolding of acts of terror to a global audience, resulting in concerns over the mainstreaming of horrific violence [19] as cyberspace ecologies become more complex [20]. The fact that violence topics garnered the most attention, engagement and discussions is double-edged [21], as increased awareness and exerts active political pressure [22], yet its psychological impact on unwitting viewers may lead to distress and anxiety [23]; and at times confusing, should its context of origin be unsubstantiated This presents impetus for policymaker attention, given that the proliferation such imagery represents an ambivalent phenomenon in digital culture: while emotionally-charged moralized content diffusing through social networks are potentially polarizing [24] and psychologically impactful [25], its very circulation represents a kind of informational mobility that often bypasses gatekeepers like traditional media or government institutions, to serve as unfiltered representations of lived realities. They can bear witness to atrocity, force acknowledgment of harm, and galvanize public attention and political action in ways that sanitized representations cannot. In this sense, their virality may function as a form of grassroots documentation or testimony within its discursive contexts.

Our findings also reveal that engagement and mention metrics were skewed toward celebrity opinions. This validates a growing body of research into celebrity influence on social advocacy, with findings suggesting that expected scopes of traditional entertainment have expanded to encompass broader global politics and domains of international diplomacy [26] and activism [27], leveraging their large audience bases to further social causes, with professional repercussions for stands taken [28]. The fixation on celebrity stances is emblematic of digitized conflict advocacy in the current century [29], which has political impacts in influencing perceptions, especially amongst the youth [30], through endorsements of social issues [31] and causes on social media platforms [32]. The prioritization of celebrity opinions attested in this study therefore reinforces and adds to the body of literature investigating political opinions in the mediatized context.

The negativity inherent in the dataset topic is countered by discoursal spikes and engagement metrics represented by groundswell movements (e.g., international activism like worldwide strikes, protests and marches, awareness movements, and outrage in response to current conditions) and actionable items (e.g., e-sims for Gaza, content creator strikes for Palestine, continued circulation of boycott consumer brand lists, co-option of the watermelon emoji as shorthand for support for the Palestinian cause). Significantly, the dominance of Palestinian groundswell causes and narratives focused on human rights violence and violations shapes perceptions on a global scale, attesting to long-standing unified strikes [33] and mobilization, even from international non-diasporic communities [34]. This amplification of specific narratives to evoke emotional empathy [35] and spur collective action is worth examining from a tactical [36] and message framing perspective [37].

The scope limitations of this study are acknowledged. The macro-level line of inquiry—meant to springboard a broader sociological viewpoint for the biggest topics that led to increased interaction and engagement permeating online spaces—is a limitation which may be complemented with a micro-level approach studying individual platforms, or the inclusion of other social media spaces like Facebook and Instagram, or broadcast media across news outlets and official distribution channels. Future research may involve tracking the chronological progression of this conflict, adopting the replicability of the methodology [38]. By that token, the focus on macro-level trends across 1 billion posts does not account for micro-level narratives, individual actions, and small-scale activism and storylines that cover greater depth and detail. We further acknowledge that social media affordances and ecologies vary across platforms: there exist differences in degrees of anonymity and network association [39]. Follow-up studies may consider micro-interactions on a smaller scale to decipher the nuances of conversational content on individual platforms. Furthermore, search term keyword data was more commonly attested on text-heavier platforms of Twitter and Reddit compared to Tiktok, where video creators speaking on these topics may not have used the search term keywords in caption data. Complementary studies may consider qualitatively coding short-form video content to elicit these embedded discussions.

Lastly, a limitation of this study is that the search term keywords (Palestine, Palestinian(s), Israel(i)(s), Gaza, Hamas) do not represent the full scope of the conflict, as many users have taken to deliberately misspelling these search terms, employing coded language, or using symbolic representations (e.g., the watermelon, as discussed) to evade platform censorship or moderation flagging [40]. Further, while the study’s data collection phase occurred in context of full firehose data obtained via academic-level access, posts available for analysis were strictly derived from public accounts, thus representing a subset of viewpoints across all users, to the exclusion of private accounts who may also have added to the discussion in personal capacities.

Overall, our study provides a broad-based macro-overview of the biggest topical buzz generators in the social media landscape, foregrounding key insights on emergent public chatter, adding to the repository of literature documenting the conflict and its global implications.