Social Networks (IT-703 (C)) - Unit 1 Short Notes

UNIT 1: Social Networks

1. Foundations of the Social Web

Emergence of the Social Web

The Social Web refers to the evolution of the World Wide Web from a static information repository (Web 1.0) to a dynamic, participatory platform (Web 2.0) where users create, share, and collaborate on content. It is characterized by:

• User-Generated Content: Platforms like blogs, wikis, and social media.

• Participatory Architecture: APIs and mashups enabling integration.

• Social Interaction: Building and maintaining online relationships.

• Collective Intelligence: Harnessing the wisdom of crowds (e.g., tagging, voting).

[!TIP] Exam Focus: Be prepared to contrast Web 1.0 ("read-only") with Web 2.0 ("read-write") and list key enabling technologies (AJAX, APIs, RSS).

Types of Web-Based Networks

Importance of Social Network Analysis (SNA)

SNA is the process of investigating social structures through the use of network and graph theory. Its importance lies in:

• Understanding Structure: Reveals patterns of connections, not just individual attributes.

• Identifying Key Actors: Finds influential individuals (via centrality).

• Detecting Communities: Uncovers subgroups or clusters within the network.

• Analyzing Information Flow: Models how information, ideas, or behaviors spread.

• Applications: Marketing, public health, cybersecurity, organizational management.

2. Semantic Web Foundations for Social Networks

Resource Description Framework (RDF) & RDF Schema (RDFS)

• RDF: A standard for data interchange on the Web. It represents information as subject-predicate-object triples (statements).

  • Example: (Alice) -- (knows) --> (Bob).

  • Data model is a directed labeled graph.

• RDFS: An extension of RDF that provides basic ontology modeling primitives.

  • Key Elements: rdfs:Class (to define types), rdfs:subClassOf (hierarchy), rdfs:subPropertyOf (property hierarchy), rdfs:domain and rdfs:range (constraints on properties).

Web Ontology Language (OWL)

OWL is a more expressive language than RDFS for defining ontologies—formal, explicit specifications of shared conceptualizations.

• Features & Significance:

  • Rich Expressiveness: Supports complex class relationships (equivalence, disjointness), cardinality constraints, property characteristics (transitive, symmetric, functional).

  • Automated Reasoning: Enables logic-based inference (e.g., if A is a parent of B and B is a parent of C, then A is a grandparent of C).

  • Interoperability: Provides a common vocabulary for machines to understand data semantics.

  • Key Constructs: owl:Class, owl:ObjectProperty, owl:DatatypeProperty, owl:equivalentClass, owl:inverseOf.

FOAF (Friend of a Friend)

FOAF is an RDF-based ontology specifically for describing people, their activities, and their relationships to other people and objects.

• Significance: Provides a machine-readable, decentralized standard for social profiles.

• Core Classes & Properties:

  • foaf:Person (or foaf:Agent)

  • foaf:name, foaf:mbox (email), foaf:homepage

  • foaf:knows (relationship between people)

  • foaf:member (group membership)

• How it Works: A user publishes their FOAF profile (an RDF file) on their website. Other FOAF agents (software) can parse these files to discover connections and build a decentralized social graph without a central database.

3. Social Network Analysis (SNA) Techniques

Basic Network Metrics

Types of Centrality

Identifies the most important nodes in a network.

1. Degree Centrality: Number of direct connections (edges) a node has.

   • $$\displaystyle C_D(v) = deg(v) $$ (Undirected). High degree = local hub/influencer.

2. Betweenness Centrality: Measures the extent a node lies on shortest paths between other nodes. A broker or bridge.

   • $$\displaystyle C_B(v) = \sum_{s \neq v \neq t} \frac{\sigma_{st}(v)}{\sigma_{st}} $$, where $$\displaystyle \sigma_{st} $$ is total shortest paths from s to t, $$\displaystyle \sigma_{st}(v) $$ is those passing through v.

3. Closeness Centrality: Inverse of the average shortest path from a node to all others. Measures reachability.

   • $$\displaystyle C_C(v) = \frac{1}{\sum_{u \neq v} d(v,u)} $$. High closeness = can quickly influence the whole network.

4. Eigenvector Centrality: Measures influence based on the influence of a node's neighbors. "Your importance depends on your important friends."

   • $$\displaystyle Ax = \lambda x $$, where $A$ is adjacency matrix, $x$ is eigenvector, $\lambda$ is eigenvalue. The principal eigenvector gives centrality scores.

[!TIP] Common Pitfall: Don't confuse Betweenness (control over flow) with Closeness (ease of reaching others). Betweenness is about paths through, Closeness is about paths from.

Clustering

• Clustering Coefficient (Local/Global): As defined above, measures triangle formation.

• Community Detection: Algorithms to find subgroups with dense internal connections and sparser external connections.

  • Methods: Modularity optimization (e.g., Louvain), hierarchical clustering, label propagation.

Matrix Representation

• Adjacency Matrix (A): $n \times n$ matrix where $$\displaystyle A_{ij} = 1 $$ if edge $(i,j)$ exists, else 0. Symmetric for undirected graphs.

• Incidence Matrix (B): $n \times m$ matrix (nodes x edges). $$\displaystyle B_{ij} = 1 $$ if node $i$ is incident to edge $j$.

• Graph Laplacian (L): $$\displaystyle L = D - A $$, where $D$ is the degree matrix (diagonal matrix of node degrees). Eigenvalues of $L$ are used in spectral clustering.

4. Communities and Network Dynamics

Definitions of Community

• Local Definition: A set of nodes more densely connected internally than with the rest of the network (structural).

• Global Definition: A partition of the network into groups that optimizes a quality function (e.g., Modularity).

• Vertex-Based Definition: A community is the set of nodes reachable from a starting node via internal paths (e.g., using random walks).

Evolution Metrics for Web Communities (from web archives)

To track how a community (e.g., a forum topic) changes over time from archived snapshots:

1. Size Metrics: Growth/decline in number of members/posts.

2. Structural Metrics: Changes in density, average path length, clustering coefficient.

3. Activity Metrics: Rate of new posts, user churn, burstiness.

4. Content Metrics: Topic drift (using text analysis), sentiment shift.

5. Membership Metrics: Core vs. peripheral user stability.

Network Reduction Methods

Techniques to simplify a large network while preserving essential properties.

• K-Core Decomposition: Recursively removes nodes with degree < k. The k-core is the maximal subgraph where all nodes have degree ≥ k. Identifies the dense core.

• K-Shell Decomposition: Similar to k-core but used for identifying influential spreaders in epidemic models.

• Bipartite Projection: Projects a bipartite network (e.g., users and groups) onto a unipartite network (users connected if they share a group), often with weighted edges.

• MST (Minimum Spanning Tree): Keeps only edges that connect all nodes with minimal total weight, removing cycles.

5. Applications and Enabling New Human Experiences

Social Networks as Enablers

• Connectivity: Overcoming geographical barriers.

• Information Dissemination: Real-time news, citizen journalism.

• Collective Action: Mobilizing social movements, crowdfunding.

• Identity & Reputation: Building digital personas, professional branding.

• Economic Opportunity: Gig economy, influencer marketing, e-commerce.

Reality Mining

• Definition: The collection and analysis of real-world, context-aware data (from mobile phones, sensors, logs) to understand human behavior, relationships, and patterns at a societal scale.

• In Social Context: Uses data from social networks + physical sensors (GPS, Bluetooth) to model:

  • Physical proximity and face-to-face interactions.

  • Mobility patterns.

  • Social context (location, activity).

  • Example: Predicting friendship strength from co-location data.

Context Awareness in Social Systems

• Definition: Systems that adapt their behavior based on information about the user's environment (location, time, activity, social setting, device).

• In Social Networks: Enables:

  • Location-based Services: "Find friends nearby."

  • Adaptive Privacy: Automatically adjusting profile visibility based on location (e.g., home vs. office).

  • Contextual Recommendations: Suggesting events or content relevant to current activity.

  • Ambient Awareness: Lightweight, peripheral updates about friends' activities.

6. Security, Privacy, and Attack Spectrum

Privacy Issues in Online Social Networks (OSNs)

• Data Collection & Profiling: Extensive personal data gathering for targeted advertising.

• Disclosure of Sensitive Information: Users overshare (location, birthdate, relationships).

• Re-identification: De-anonymizing "anonymous" datasets via graph structure or auxiliary data.

• Social Graph Exposure: Unintended visibility of connections (e.g., "friends of friends").

• Lack of Control: Difficulty in managing or deleting historical data across platforms.

Attack Spectrum and Countermeasures

Challenges for Decentralized Online Social Networks (DOSNs)

• Identity Management: No central authority for unique ID verification → Sybil attacks.

• Data Consistency & Synchronization: Across independent servers (pods).

• Privacy Enforcement: Implementing fine-grained access control in a distributed setting.

• Content Moderation: No central moderator; relies on community governance or local pod policies.

• Interoperability: Ensuring different DOSN protocols (e.g., ActivityPub) work seamlessly.

• User Experience: Higher complexity for average users compared to centralized services.

7. Specific Technologies and Tools

Email Groups as Social Networks

• Concept: An email list or group (e.g., Google Groups, mailing lists) functions as a one-to-many broadcast network.

• Social Structure: The group moderator and active participants form a star-like or small-world network.

• Analysis: Can be analyzed using SNA by constructing a graph where nodes are members and edges represent reply-to or mention relationships in email threads.

• Limitations: Connections are typically episodic (only during a thread) and asymmetric (replies are directed).

RSS Feeds in the Social Web Context

• Definition: RSS (Really Simple Syndication) is a web feed format for publishing frequently updated content (blogs, news, podcasts).

• Role in Social Web:

  • Content Aggregation: Users subscribe to feeds from friends, bloggers, or sites, creating a personalized content stream.

  • Decentralized Following: Pre-dates modern "follow" buttons; allows users to curate their own information diet from diverse sources without being on a single platform.

  • Publishing Mechanism: Enables bloggers/sites to push updates to subscribers, forming a publisher-subscriber network.

  • Social Discovery: Sharing RSS feed URLs is a form of social signaling about interests.

• Modern Equivalent: Largely superseded by social media algorithms and built-in follow features, but still used by power users and in podcasting.