Course notes on Social Computing
L1 - Introduction
• Definitions of social computing:
⁃ Any type of computing application in which software serves as an
intermediary or a focus for a social relation
⁃ Use of social software, a growing trend of ICT usage of tools that
support social interaction and communication
⁃ A social structure in which technology puts power in individuals and
communities, not institutions
⁃ Computational facilitation of social studies and human social dynamics
as well as the design and use of ICT technologies that consider social
context
• Computational analysis of human behavior:
⁃ Sensing:
⁃ Real-world behavior: sensors (cameras, mics, IoT, phones)
⁃ Digital-world behavior: internet, games, social media
⁃ Applications: personalization & recommendation, security and
surveillance, healthcare and robotics, ambient intelligence,
entertainment, public good
⁃ Human behavior analysis:
⁃ Theory-driven: psychology, cognitive science and sociology
⁃ Data-driven: PR, ML, computational perception
⁃ Individual behaviors: speech recognition, action recognition, facial
expressions, mental health diagnostics, sentiment analysis, deception
detection, apparent personality
⁃ Dyadic/group behaviors: turn taking, mimicry and rapport, non-verbal
social signs, games and playful interactions, child-parent interactions
⁃ Social computing: migration and mobility, urban computing, epidemics
models, social indicators
• Social computing research problems: social information (societies’ features,
such as social relations, institutional structure, roles, power, influence, and
control) and social knowledge (agents’ cognitive and social states (for
example, actors’ motivations, intentions, and attitudes)) provide a basis for
inferring, planning, and coordinating social activities
• Important notes:
⁃ Using image, video, audio, text as part of the social science research
challenge
⁃ Datasets: ‘ready-made’ vs. ‘custom made’
⁃ Differences between data science & social sciences in terms of goals,
interpretation, emphasis, questions, concerns
• Examples:
⁃ Predicting flu outbreaks with google trends
⁃ Discovering factors influencing civil wars
• Prediction versus explanation:
⁃ Focus on specific features or effects:
⁃ No interventions or distributional changes:
⁃ Quadrant 1 = descriptive modeling: describe situations in
the past or present (but neither causal nor predictive)
, ⁃ Under interventions or distributional changes:
⁃ Quadrant 2 = explanatory modeling: estimate effects of
changing a situation (but many effects are small)
⁃ Focus on predicting outcomes:
⁃ No interventions or distributional changes:
⁃ Quadrant 3 = predictive modeling: forecast outcomes for
similar situations in the future (but can break under
changes)
⁃ Under interventions or distributional changes:
⁃ Quadrant 4 = integrative modeling: predict outcomes and
estimate effects in as yet unseen situations
Lecture 2: Ethics
• Data-driven governing: datasets collected via public and private sectors
⁃ Utility companies (use of electricity, gas and water)
⁃ Transport providers (location/movement, travel flow)
⁃ Mobile phone operators (location/movement, app use and behaviour)
⁃ Travel and accommodation websites (reviews, location/movement and
consumption)
⁃ Social media sites (opinions, photos, personal information and location/
movement)
⁃ Crowdsourcing and citizen science (maps, local knowledge, weather)
⁃ Government bodies and public administration (services, performance
and surveys)
⁃ Financial institutions and retail chains (consumption and location)
⁃ Private surveillance and security firms (location and behavior)
⁃ Emergency services (security, crime, policing and response)
⁃ Home appliances and entertainment systems (behavior and
consumption)
• Privacy: to selectively reveal oneself to the world, basic human right
⁃ A measure of the access others have to you through information,
attention, and physical proximity
⁃ The ability to control the acquisition or release of information about
oneself
⁃ A right to privacy is neither a right to secrecy or a right to control but a
right to appropriate flow of personal information
⁃ How to protect privacy?
⁃ Informed consents and data protection plans
⁃ Identity privacy (protect personal and confidential data)
⁃ Bodily privacy (protect the integrity of the physical person)
⁃ Territorial privacy (protect personal space, objects and property)
⁃ Locational and movement privacy (protect against the tracking
of spatial behavior)
⁃ Communications privacy (protect against the surveillance of
conversations and correspondence)
⁃ Transactions privacy (protect against monitoring of queries/
searches, purchases, and other exchanges)
⁃ GDPR:
⁃ ‘Personal data’ means any information relating to an
, identified or identifiable natural person (‘data subject’),
which refers to one who can be identified, directly or
indirectly, in particular by reference to an identifier such
as a name, an id number, location data, an online ID or to
one or more factors specific to the physical, physiological,
genetic, mental, economic, cultural or social identity of
that natural person
• PII: personally identifiable information
• Dark patterns: manipulative and deceptive UX designs and practices in online
commercial settings influencing the decisions users take
• Datafication and privacy:
⁃ Datafication, dataveillance and geosurveillance
⁃ Inferencing and predictive privacy harms
⁃ Anonymization and re-identification
⁃ Notice and consent is absent
• Data Ethics:
⁃ Ethics of data: collection, storage, and usage of large scale data, the
biases inherent in the dataset itself, consent of data owners, risks and
benefits arising from the analysis of the data, and privacy
⁃ Ethics of algorithms: implementation of data processing approaches
that may bring their own biases into the equation, auditing and
transparency of such algorithms, trust, and responsibility
⁃ Ethics of practice: results of data analysis, their effects on real-world
decisions, and actual deployment of algorithms, focusing on power,
authority, policy, user rights, and secondary use of data
• Principles of data ethics:
⁃ Ownership: individuals own their data and their info; they have control
and right to decide with whom to share it with and levels of access to it
⁃ Transparency: acquire data with consent from whoever collecting it
from, letting them know where, for how long and how it will be stored
⁃ Privacy: data should be secured safely to ensure the prevention of data
breaches/leaks of PIIs
⁃ Intention: data should be used for its intended purpose and nothing
more; endure it’s being used and shared fairly with authorization
⁃ Ethics of data collection: consent, data protection, identity
⁃ Ethics of algorithm development: fairness, transparency, bias
⁃ Ethics of practice and deployment: accountability, power
• Ethics plan, account for:
⁃ Definition of personal data, subjects
⁃ Privacy by design and default
⁃ Consent, legitimate and fair processing
⁃ Encryption and storage (data leakage), destruction, archiving
⁃ Right of access, correction, erasure, objection
⁃ Responsibility and accountability (especially when working with 3rd
parties)
• Three approaches to design ethical guidelines:
⁃ Rules-based approach
⁃ Ad hoc approach
⁃ Principles-based approach
