Specialisation: Social Media Analysis · Topic Modelling · Sentiment Analysis · Health Communication · Computational Communication

🔍 Open to: Postdoctoral positions · Research fellowships · Visiting researcher roles · Collaborative projects
in computational communication, NLP, social media analysis, topic modelling, and digital methods.

Sawood Anwar is a computational social scientist and early career researcher specialising in natural language processing (NLP), social media analysis, computational communication, topic modelling, and sentiment analysis. He is a PhD graduate of the University of Urbino Carlo Bo, Italy (Cycle XXXVII, 2025), supervised by Prof. Fabio Giglietto and co-supervised by Prof. Giovanni Boccia Artieri.

His doctoral research analysed Facebook Reactions as emotional indicators of public sentiment and engagement with COVID-19 news on Indian media platforms (March 2020 – March 2022), using a dataset of 68,319 Facebook posts from four major English-language Indian news outlets. The methodology combined time-series analysis, BERTopic embedding-based topic modelling, LLM-assisted cluster labelling, and lexicon-based sentiment analysis.

Sawood Anwar’s broader research agenda addresses health communication, misinformation detection, platform studies, disinformation, crisis communication, survey methods, psychometric scale validation, and machine learning for text classification. His published work has received 25+ citations (2024–2025).

My research sits at the intersection of computational communication science, natural language processing, and platform studies. I investigate how digital platforms mediate collective emotion, public discourse, and information ecosystems during societal crises — with a particular focus on platform-native engagement signals (such as Facebook Reactions) as proxies for affective public response.

Methodologically, I combine corpus-based text analysis, embedding-based topic modelling (BERTopic, STM), time-series analysis, lexicon-based sentiment analysis, and LLM-assisted semantic annotation to study large-scale social media corpora. I am also experienced in survey methods, psychometric scale validation (EFA, CFA, SEM), and experimental designs for communication research.

My work contributes to three overlapping fields: (1) computational communication science, by advancing mixed-method frameworks for platform data analysis; (2) health and crisis communication, by examining how Indian media audiences responded to COVID-19 information online; and (3) misinformation and disinformation studies, by developing reproducible ML pipelines for detecting and understanding false information.

“Facebook Reactions” as Emotional Indicators: A Multi-Method Approach to Analyzing User Engagement with COVID-19 News on Indian Media Platforms
Sawood Anwar — University of Urbino Carlo Bo, 2025
🔗 Full text — Institutional Repository ORA Uniurb

Abstract: This thesis investigates the role of Facebook Reactions as indicators of public sentiment and engagement with COVID-19 pandemic-related news in India during different stages of the pandemic (March 24, 2020 – March 31, 2022). The study employs a mixed-methods approach combining time-series analysis, embedding-based topic modelling, LLM-assisted cluster labelling, and lexicon-based sentiment analysis. The dataset comprises 68,319 Facebook posts from four major English-language Indian news outlets — The Times of India, The Hindu, Indian Express, and Hindustan Times — with a focused subset of 8,622 posts covering the early pandemic phase (March 24 – April 14, 2020). Findings reveal how discrete Facebook Reaction types (Like, Love, Haha, Wow, Sad, Angry) function as affective engagement signals tracking shifting public sentiment across dominant news themes and pandemic phases in India.

Keywords: Facebook Reactions COVID-19 India social media sentiment analysis topic modelling BERTopic time-series computational communication health communication misinformation The Times of India The Hindu Indian Express Hindustan Times

[1] Anwar, S., & Giglietto, F. (2024). Facebook Reactions as Emotional Indicators: Analyzing Public Engagement with COVID-19 Pandemic News on Indian Media Platforms During the Early Lockdown Phase. Frontiers in Sociology, 9, 1379265. 25+ citations.
🔗 DOI: 10.3389/fsoc.2024.1379265 · 🔓 Open Access · 🌐 Full Text

[2] Giglietto, F., Ghasiya, P., Sasahara, K., Anwar, S., & Mincigrucci, R. (2022). Between Localism and Politics: Mapping Coordinated Networks that Circulate Problematic Health Content in India. SSRN Working Paper 4164140.
🔗 View on Google Scholar

[3] Anwar, S. (2025). “Facebook Reactions” as Emotional Indicators: A Multi-Method Approach to Analyzing User Engagement with COVID-19 News on Indian Media Platforms. PhD Thesis, University of Urbino Carlo Bo.
🔗 Handle: ora.uniurb.it/handle/11576/2761691

📊 Live citation metrics: Google Scholar — Sawood Anwar

computational communication social media analysis natural language processing NLP topic modelling BERTopic STM sentiment analysis Facebook Reactions COVID-19 health communication misinformation disinformation detection crisis communication time-series analysis platform studies text embeddings survey methods psychometrics machine learning text classification Indian media digital methods

• Computational Communication Science — platform-native engagement signals, affective public response, digital media effects
• Natural Language Processing — topic modelling (BERTopic, STM), text embeddings, LLM-assisted annotation
• Sentiment & Affective Analysis — lexicon-based methods, reaction-type classification, emotion detection in social media
• Health & Crisis Communication — COVID-19, pandemic misinformation, Indian media platforms
• Disinformation & Platform Studies — coordinated inauthentic behaviour, cross-platform analysis (Facebook, Instagram, Reddit)
• Quantitative & Computational Methods — time-series, ML classification, survey experiments, psychometric scale validation

ℹ️ Open Science Commitment: All code and workflows in this repository are publicly available under open licences to support reproducibility and transparency in computational communication research. Primary research data is governed by Meta’s CrowdTangle and Content Library data-use agreements.

R Python Facebook COVID-19 India NLP BERTopic Sentiment Analysis Time-Series PhD Thesis Open Science

Core repository for the doctoral dissertation by Sawood Anwar (University of Urbino Carlo Bo, 2025). Multi-method analysis of Facebook Reactions (68,319 posts, 2020–2022) as emotional indicators of public engagement with COVID-19 news across The Times of India, The Hindu, Indian Express, and Hindustan Times. Integrates time-series analysis, BERTopic-based topic modelling, LLM-assisted cluster annotation, and lexicon-based sentiment analysis.
📝 Linked publication: Frontiers in Sociology, 2024

R Time-Series Facebook Anomaly Detection Social Media COVID-19 Misinformation

Three-module reproducible R framework for longitudinal social media engagement research: (1) general-purpose time-series toolkit, (2) COVID-19 Facebook extension with reaction-type and pandemic-phase stratification, and (3) health misinformation spike detection with event annotation and heatmap visualisation.

R Structural Topic Model STM Social Media NLP Reproducible Research

Fully reproducible Structural Topic Model (STM) pipeline for social media corpora in R. Covers corpus construction, document-feature matrix (DFM) building, searchK model selection, prevalence covariate specification, estimateEffect inference, and publication-quality ggplot2 visualisation.

Python R BERTopic Sentence Transformers UMAP HDBSCAN Embedding-Based NLP LLM

BERTopic pipeline for thematic analysis of media corpora: sentence embeddings → UMAP dimensionality reduction → HDBSCAN clustering → c-TF-IDF topic representations → optional LLM-based semantic labelling. Outputs exported to R for post-hoc engagement analysis.

R Sentiment Analysis AFINN Bing NRC Social Media Lexicon

Systematic comparison of AFINN, Bing, and NRC sentiment lexicons applied to social media text. Annotated code, visual output comparisons, and an interpretive guide to lexicon selection for computational communication research.

R Facebook Instagram Health Misinformation STM Meta Platforms

Content analysis workflows for studying platform-mediated health communication and misinformation on Meta platforms (Facebook + Instagram). Combines STM topic modelling with engagement metric analysis.

R Reddit Content Coding Intercoder Reliability Political Communication Misinformation

Manual content coding framework for political communication and misinformation research on Reddit. Includes structured codebook, intercoder reliability scripts (Krippendorff’s α, Cohen’s κ), and descriptive analysis pipeline.

R Facebook Instagram Reddit Cross-Platform Comparative Research

Unified R framework harmonising engagement data from Facebook, Instagram, and Reddit into a single comparative schema for systematic cross-platform digital communication research.

Python R Machine Learning Disinformation Detection SHAP TF-IDF NLP Explainable AI

Supervised machine learning pipeline for classifying disinformation in news posts. Benchmarks Logistic Regression, Random Forest, SVM, and Naïve Bayes using TF-IDF features. Includes SHAP-based model interpretability for explainable AI in communication research.

R tidymodels textrecipes Supervised NLP News Classification Health Politics

Supervised text classification pipeline in R for labelling news articles by topic (health, politics, economy) and credibility (credible vs. misleading) using tidymodels and textrecipes.

Python R CrowdTangle Meta Content Library API Facebook Instagram Research Ethics

Documented academic data collection pipeline for Facebook and Instagram research. Covers legacy CrowdTangle CSV exports and current Meta Content Library API access, including research ethics guidelines and a unified data schema for downstream analysis.

R Survey Data Likert Scales srvyr gtsummary Qualtrics Descriptive Statistics

End-to-end survey data analysis pipeline in R: import, cleaning, Likert scale analysis, descriptive statistics with gtsummary, survey weighting with srvyr, and diverging bar chart visualisation. Compatible with Qualtrics, SurveyMonkey, SPSS, and Stata exports.

R psych lavaan EFA CFA Psychometrics Reliability SEM

Psychometric scale validation workflow using psych and lavaan. Covers item-level inspection, Cronbach’s α and McDonald’s ω reliability estimates, EFA with parallel analysis, CFA with fit indices (CFI, TLI, RMSEA, SRMR), and convergent/discriminant validity assessment.

R Survey Experiments Vignette Studies estimatr modelsummary Causal Inference

Analysis pipeline for survey experiments and vignette studies. Covers randomisation checks, manipulation checks, Average Treatment Effect (ATE) estimation with robust standard errors (estimatr), OLS regression with modelsummary publication tables, moderation/interaction effects, and coefficient plot visualisation.

Sawood Anwar · PhD · Computational Social Scientist · NLP Researcher · University of Urbino Carlo Bo · Italy · 2025
Computational Communication · Social Media Analysis · Topic Modelling · Sentiment Analysis · Facebook Reactions · COVID-19 · Indian Media