cs.CLNov 19, 2026

FoodReact-AR: Learning Audience Reactions that Zero-Shot LLMs Miss in Arabic Food-Insecurity Discourse

Authors: Wajdi Zaghouani, Mabrouka Bessghaier, Salma Azouzi, Shimaa Amer Ibrahim

Organizations: Northwestern University in Qatar, Doha, Qatar · Independent Researcher, Tunisia

Abstract

Facebook reactions record how an audience received a post, a signal absent from the post's text. We treat the six differentiated reactions on 74,241 Arabic food-insecurity posts from public Facebook pages (2020-2024) as compositions, apply a centered log-ratio transformation, and cluster them into four audience profiles: Solidarity (Love-dominant), Dismissal (Haha-dominant), Grief (Sad-dominant), and a flat Ambivalent profile. Dismissal peaks in 2022 (22.3%), the year of the wheat price shock, and grief in 2024 (14.0%), during the Gaza food emergency; a page-clustered GEE confirms both shifts (odds ratios 2.18 and 1.34), and they survive minimum-reaction filtering, alternative zero replacements and clustering methods, and bootstrap resampling. We then test whether the profiles can be predicted without reactions. A zero-shot Arabic LLM recovers Grief for 83.8% of posts but Dismissal for only 3.4%, and in a matched ablation, it shows the reaction percentages do not help. A simple classifier trained on the reaction-derived labels and tested on unseen pages recovers dismissal for 60.0% (macro-F1 0.56 against 0.43), mainly from mentions of officials and institutions. Derisive audience responses are thus learnable from text, yet a zero-shot LLM misses them. We release the corpus, LLM, and human annotation layers, and an evaluation protocol as FoodReact-AR.

Explore similar work

May 21, 2026

ArabDiscrim: A Decade-Long Arabic Facebook Corpus on Racism and Discrimination

We present ArabDiscrim, a decade-long lexical resource and corpus of 293K public Arabic Facebook posts (2014--2024) discussing racism and discrimination. Unlike existing Twitter-centric datasets, ArabDiscrim integrates platform-native engagement signals, including reactions, shares, comments, and page metadata, enabling joint analysis of language and audience response. The resource includes 200 curated terms (100 racism-related and 100 discrimination-related) with morphological regex families (13+ inflections per lemma), and 20 discrimination axes capturing identity-based grounds for unequal treatment. It also provides explicit attribution patterns. Released under a restricted research-use license for ethical compliance with platform terms, ArabDiscrim supports weak supervision, axis-aware sampling, and platform ecology research. By bridging lexical depth and ecological validity, it establishes a foundation for fairness-oriented, platform-aware Arabic NLP.
Nov 5, 2026

AraEngage: Interpretable Linguistic Correlates of Engagement in Arabic Facebook News

Engagement features developed for English are routinely applied to Arabic, but unvocalized Arabic changes what a lexical feature measures. We introduce AraEngage, a query-defined sample of 10,000 public Arabic Facebook news posts from 2,451 pages in 57 countries (2023-2024) that pairs engagement counts with interpretable linguistic features. We use it to examine which interpretable Arabic indicators are associated with follower-normalized engagement, and how far those associations can be trusted given publisher context, the definition of engagement, and lexical ambiguity. Religious expressions, message length, sentence length, and lexical density are the leading correlates in a Random Forest model (R2R^2 = 0.38 on a post-level split) interpreted with SHAP. Unlike English news, positive sentiment is associated with higher engagement in this Arabic news sample. Message length shows a length paradox: unadjusted raw engagement is highest for long posts, whereas the model's attributions for follower-normalized engagement favor short posts. A linguistic audit identifies why standard lexicon features are fragile in unvocalized Arabic: imperative lists contain homographs such as the form read as "express!" that is also the frequent preposition "via", hedge lists conflate the possibility marker qad + imperfect ("may") with the emphatic qad + perfect ("has indeed"), and religious lists mix devotional formulas with place names in political news. These are plausible sources of measurement error that may attenuate the near-zero associations of imperatives and hedges. All findings are observational associations with operationalized indicators. We release the derived linguistic features, engagement counts, refined lexicons, and analysis code; raw posts are not redistributed, in line with CrowdTangle terms.
Nov 11, 2026

AraCrypto-ML: A Multi-Label Corpus of Communicative Functions in Arabic Facebook Cryptocurrency Discourse

Arabic cryptocurrency discourse on social media is communicatively heterogeneous: a single post may report prices, explain protocols, promote investment, warn of scams, or invoke Islamic jurisprudence. Yet no annotated resource captures this functional diversity. We introduce AraCrypto-ML, a 10,000-post multi-label corpus of Arabic Facebook cryptocurrency discourse (2014-2024) from 2,898 public pages using a set of related keywords. Collected posts are annotated into six categories: Market Commentary, Educational, Promotional, Skeptical-Warning, Religious Framing, and Other/Unclear, using Fanar-2-27B under a prompt-only multi-label protocol. We validate 1,000 posts against adjudicated gold labels (macro kappa = 0.74) and apply Rogan-Gladen calibration to correct labelling bias. The two annotators agree with each other at macro kappa = 0.62 before adjudication, below the model's agreement with either of them (0.70 and 0.69). Reaction profiles differ strongly by function: Love is about 70% of reactions per post on Educational and Promotional posts but about 33% on Skeptical-Warning, which draws the largest Haha share (about 39%).