Session Recordings
Week 1 — Foundations, Methods & Theory · RMIT
Day 1 — Introduction to Computational Social Science · Monday 22 June
Day 2 — Data Donation, Publishing & Research Infrastructure · Tuesday 23 June
Data Donations and Participant-Centric Research
Speakers: Kellie Vella (AIO), Lauren Hayden
Chapters (12)
- 00:00 Introductions and the AIO
- 01:14 What is data donation?
- 02:58 DDPs and screen capture
- 05:00 Inside an Instagram DDP
- 08:26 Screen capture and participant burden
- 10:30 A user-centric view of platforms
- 15:27 Combining methods
- 19:23 Activity: finding DDPs
- 24:14 The Schema Explorer tool
- 27:06 Ethics and participant-centric research
- 35:45 Building data literacy
- 41:05 Q&A and your own DDP
Demystifying Publishing in Computational Social Science
Speakers: Olga Boichak (USyd), Kateryna Kasianenko
Chapters (14)
- 00:00 Framing: the messiness of publishing
- 00:59 Who we are: four years, SICSS Sydney 2022
- 02:03 Mixed methods and the NAFO study
- 05:35 Reverse-engineering outputs; XKCD paper types
- 08:33 Papers vs articles; interdisciplinary venues
- 14:38 Conferences and AoIR
- 17:36 Research design as a post-mortem
- 19:17 Story 1: NAFO avatars and identity
- 28:36 Story 2: Non-human humanitarians (Kakhovka)
- 40:39 Story 3: Participatory war literature review
- 44:19 Designing a CSS project and validity
- 48:03 Choosing a venue: argument, rejection, costs
- 57:22 Authorship, conventions, reviewer responses
- 1:02:52 Q&A: reviews, single vs co-authorship, mentors
Nectar: Australian Research Infrastructure for Computational Analysis
Speaker: Sonia Ramza (ARDC)
Chapters (9)
- 00:00 Introducing Nectar Cloud
- 01:55 Cloud vs HPC
- 03:27 Allocations and the free project trial
- 05:49 Managed services: desktops, Binder, Jupyter
- 08:40 Real-world research use cases
- 11:14 Eligibility, support and follow-up
- 15:44 Q&A: NCI, AI-ready tools, GPUs and VRAM
- 22:04 Live demo: dashboard and flavours
- 27:31 Q&A: ethics, Binder and data sovereignty
Day 3 — Data Collection and Working Across Disciplines · Wednesday 24 June
Does Computational Social Science Lack Theory?
Speaker: Ehsan Dehghan (QUT)
<a class="res-pdf" href=https://internetobservatory.org.au/training_materials/sicss/Does_computational_social_science_lack_theory-Ehsan_Deghan.pdf" target="_blank" rel="noopener noreferrer">Training in PDF</a>
Chapters (13)
- 00:00 Opening: a discourse scholar's view
- 01:56 The wrong question: where theory hides
- 04:03 A short history of the field
- 14:17 Five resistances to theory
- 21:57 Three layers: phenomenon, measurement, method
- 24:07 Activity: what's wrong with this abstract?
- 33:51 Group work: theory in your own pipeline
- 50:13 Building theory with CSS: finding a spine
- 55:01 Discourse theory, ANT, and flat ontology
- 1:03:16 The Reddit case study
- 1:08:54 Dangers: speed, hype, empiricism of presence
- 1:17:13 Takeaways: be friends with theory
- 1:19:07 Q&A: stretching, power, and Žižek
The AIReD Platform for Australia-wide Social Media Discovery and Usage
Speaker: Richard Sinnott (University of Melbourne)
Chapters (12)
- 00:00 A systems-builder, not a theorist
- 05:46 What AIReD is: Australia-wide social media
- 10:16 What a single post leaks about you
- 14:51 Scale, heterogeneity, and the ethics problem
- 17:12 Drowning in data; discovery as the answer
- 20:09 Funding, platforms, and losing Twitter
- 26:03 Topic modelling and rehydrating posts
- 32:24 Live demo: the AIReD dashboard
- 47:53 GDELT, multilingual search, and the chat interface
- 56:23 Use cases and teaching with the data
- 1:06:21 Official data and running infrastructure
- 1:14:35 Q&A: discovery, not answers
Collecting and Analysing Data Download Packages
Speakers: Dan Tran (AIO), Michael Esteban
Chapters (11)
- 00:00 Recap and the plan for today
- 01:31 What a DDP is, and the GDPR right to export
- 04:33 Requesting a DDP, and designing for attrition
- 06:09 Advantages and limitations of DDPs
- 08:48 Exercise: your phone's most sensitive data
- 09:36 Kellie Vella: designing a study with DDPs
- 14:07 Demo: setting up a DDP project with consent controls
- 22:40 Dan Tran: making sense of JSON
- 28:09 Opening a DDP: zip files, text editors, and Chrome
- 35:27 The AIO DDP viewer
- 40:56 The lab workspace and AI-assisted analysis
Working with Text Using Computational Techniques
Speakers: Kim Doyle, Daniel Russo-Batterham (MDAP)
Chapters (11)
- 00:00 Introduction
- 02:42 Session goal: LLM-assisted scraping
- 09:07 Claude Code plan mode demo
- 14:20 Activity: writing a clear scraping brief
- 16:35 Report-back: refusals, ethics, robots.txt
- 26:54 Activity: spot the errors in scraped data
- 34:10 The light and dark sides of vibe coding
- 42:24 Transformers and notebooks explained
- 47:45 Sentiment analysis with RoBERTa
- 54:34 Topic labelling the episodes
- 1:02:44 Audience Q&A
Day 4 — Tools and Approaches to Data Analysis · Thursday 25 June
Screen Capture for Data Collection
Speakers: Dan Tran, Daniel Angus
Collecting images, text and ads from participants' own screens with the AIO Mobile Screen Capture tools — privacy engineering, ethics, a hands-on app install, and the analysis pipeline. Includes the technical deep-dive and debrief.
Chapters (12)
- 00:00 Introduction and speaker handover
- 02:01 Why mobile screen capture?
- 05:27 Building an Android capture app
- 09:59 The challenge: a stream of images
- 12:58 Cloud vs on-device processing
- 22:21 Detecting ads with YOLO and OCR
- 29:12 Hands-on: installing the Mobile Ad Toolkit
- 32:28 Debrief: inside the capture pipeline
- 38:10 Matching ads to the ad library
- 44:06 Researcher dashboard demo
- 47:14 Participant dashboards and data quality
- 56:41 Audience questions and discussion
Using LLMs to Create Data Analysis Pipelines for Text-as-Data Research
Speaker: Seraphine F. Maerz
Chapters (14)
- 00:00 Introduction
- 01:54 Setup check: R, RStudio, API keys
- 05:52 LLMs as a research methodology
- 11:22 Open vs closed models and sensitive data
- 17:44 The quallmer five-step workflow
- 21:44 Worked example: 5,000 political speeches
- 33:20 Questions before hands-on
- 40:56 Hands-on: installing quallmer
- 57:34 Defining code books and schemas
- 1:02:18 Running your first LLM coding
- 1:03:16 Local models with Ollama
- 1:13:18 Replication and model comparison
- 1:19:06 Validation: Krippendorff and gold standards
- 1:22:42 Wrap-up, exercises and open-model outlook
RAG Systems in Research
Speakers: Futoon Abushaqra, Sachin Pathiyan Cherumanal
Chapters (9)
- 00:00 Session goals
- 02:30 The traditional research pipeline
- 04:53 Why not just use ChatGPT?
- 08:15 Effort vs control: where RAG sits
- 11:17 How a RAG pipeline works
- 14:42 Case study: RAG failure modes
- 19:11 Hallucination and missing context
- 24:51 Takeaways: can you trust RAG?
- 28:32 Group activity briefing
Image Analysis for Qualitative and Quantitative Research
Speakers: Kunal Chand, Lauren Hayden
Chapters (11)
- 00:00 Kunal Chand: computational image toolkits
- 01:14 The image-sorting activity
- 02:19 What happens when we throw a computer at it?
- 04:33 Convolutional neural networks
- 05:48 Embedding vectors
- 08:37 Clustering algorithms
- 12:45 Tree data structures
- 14:35 Case study: alcohol advertising
- 19:11 The Image Machine
- 31:40 Live Image Machine demo
- 50:40 Getting access and audience Q&A
Day 5 — Disciplines, Careers, and Industry · Friday 26 June
Cross-Disciplinary Collaboration: Bringing Social Science and Computational Analysis Together
Panel: Oleg Zendel, Johanne Trippas, Hiruni Kegalle, Oliver Eklund
Chapters (13)
- 00:00 Setting up the discussion
- 00:35 Crossing disciplinary boundaries
- 11:13 Challenges of interdisciplinary work
- 13:33 Case study: cardiac-arrest collaboration
- 19:03 Publishing venues across fields
- 22:51 Speaking each other's language
- 28:51 Government survey war stories
- 31:53 Disagreeing productively
- 39:05 Job hunting as an interdisciplinary scholar
- 50:31 Choosing methods and finding coherence
- 1:00:39 Communicating with government audiences
- 1:07:01 Learning to disagree well
- 1:15:33 Wrap-up and thanks
Career Success
Speaker: Johanne Trippas
Chapters (15)
- 00:00 Introduction
- 07:18 The Rascal Pyramid
- 08:24 Layer 1: look after yourself
- 09:47 Layer 2: your physical setup
- 15:07 Cyber setup and backups
- 19:46 Pomodoro and time management
- 27:29 Planning without overplanning
- 41:36 Journaling to stay grounded
- 43:16 Taming the literature
- 47:31 Thesis files and writing templates
- 52:15 Word lists and writing with dyslexia
- 1:01:37 The support layer: mentors and lab mates
- 1:14:34 The achievement layer: online presence
- 1:17:45 The top: doing creative research
- 1:19:41 Audience Q&A
Grant Writing in Computational Social Science
Speaker: Daniel Angus
Chapters (8)
- 00:00 Let's talk about money
- 02:32 Grant writing as persuasive writing
- 04:06 Fellowships vs grants
- 14:22 Ten years of rejections, normalised
- 20:24 The three-bubble model
- 29:34 Positioning yourself clearly
- 36:42 Mentors, saying yes, career strategy
- 41:40 Audience Q&A
Week 2 — Collaborative Research Projects · Deakin Downtown
Day 6 · Monday 29 June
Knowledge Extraction and Systematic Data Curation with SciLire
Speakers: Jessica Irons, Stephen Wan (CSIRO)
Chapters (11)
- 00:00 Meet the CSIRO team and today's plan
- 02:47 What is SciLire?
- 08:24 The NCC consultation example
- 12:56 Live demo: creating a project
- 15:22 Data sensitivity and model choice
- 24:00 Reading the extracted table
- 25:26 Validating results against the text
- 35:18 Hands-on troubleshooting and features
- 43:12 Improving extraction with training examples
- 49:34 Exporting results and confidence scores
- 56:32 Use cases, publications and wrap-up
Day 7 · Tuesday 30 June
Music Score Analysis through Natural Language Interfaces
Speaker: Daniel Russo-Batterham (MDAP)
Chapters (14)
- 00:00 Natural-language interfaces to tools
- 02:27 Why give LLMs tools?
- 06:51 A brief history of LLM tool use
- 08:20 The Model Context Protocol
- 11:49 Running MCPs: local vs deployed
- 14:58 Worked example: the Zotero MCP
- 17:45 Safety when running MCPs
- 20:46 Activity: find an MCP for your work
- 25:33 Report-back: MCPs people found
- 29:20 Applying MCPs to music
- 36:02 Music21 and score analysis demo
- 44:46 Exploring a corpus of scores
- 46:07 Activity: connect to the music MCP
- 48:48 Wrap-up and audience Q&A
Day 8 · Wednesday 1 July
Validation in Computational Social Science
Speaker: Matteo Vergani (Deakin)
Chapters (12)
- 00:00 Introduction and session plan
- 02:07 What is a latent social construct?
- 08:28 The ground-truth problem
- 11:23 Precision, recall, accuracy, F1
- 15:43 Asymmetric error costs
- 20:49 Benchmarks and the model horse race
- 23:57 A construct-validation framework
- 40:14 Case study: measuring social cohesion
- 51:42 Convergent, discriminant, external validity
- 58:23 Group exercise: validate a construct
- 1:00:37 Report-back and discussion
- 1:09:13 What is populism? Measurement debates
How to use Wikibase for Mixed-Methods Research
Speaker: Francesco Bailo (USyd)
Chapters (11)
- 00:00 From Wikipedia to Wikidata to Wikibase
- 02:45 Why Wikibase for research teams
- 12:30 The shared-table metaphor
- 26:40 The data model: items, properties, statements
- 38:01 Ontologies and code books
- 50:22 Hands-on: wikibase.cloud
- 53:47 Querying with SPARQL
- 58:46 Demo: importing and linking data
- 1:12:18 Demo: MPs and speeches example
- 1:28:11 Audience Q&A
- 1:35:15 Ethics committees and privacy
The Australian Internet Observatory (https://doi.org/10.25956/twvn-ca19) is a co-investment partnership with RMIT University, QUT, University of Queensland, University of Melbourne, Swinburne University, Deakin University and the Australian Research Data Commons (ARDC) through the HASS and Indigenous Research Data Commons (DOI:10.3565/hjrp-b141). The ARDC is enabled by the Australian Government’s National Collaborative Research Infrastructure Strategy (NCRIS).
15:15–16:30 | Keynote | program entry
Social Bias in Computational Social Science
Speaker: Ahrabhi Kathirgamalingam (GESIS)
Training in PDF
Chapters (12)