UNIT 3

Definition of Network Structure

a network structure refers to the organization and pattern of relationships (edges) between
entities (nodes) such as individuals, organizations, or systems in a network. Each node represents
a data point (e.g., a user), and each edge shows a connection or interaction (e.g., a friendship,
follow, message, or like).
Types of Network Structures:
1. Centralized Network
2. Decentralized Network
3. Distributed Network
4. Hierarchical Network

1. Centralized Network Structure

Definition:
In a centralized network, all data flows through a central node (hub). Other nodes are connected
to the central node, but not directly to each other.

Advantages:
 Easy to manage and control
 Centralized decision-making and monitoring
 Efficient data collection from all nodes

Disadvantages:
 Single point of failure — if the central node fails, the network fails
 Can be a bottleneck under heavy load

Example in Social Media Analytics:

A customer support system where all complaints or queries from users are directed to a single
social media manager account. This account is the central hub for all communications.

2. Decentralized Network Structure

Definition:
There are multiple central nodes in a decentralized network. Each central node manages a subset
of nodes, and there’s partial connectivity among hubs.

Advantages:
 More fault-tolerant than centralized networks
 Scalable and easier to manage compared to distributed networks
 Reduces bottlenecks

Disadvantages:
 Complex management compared to centralized
 Potential inconsistency in data across nodes
Example in Social Media Analytics:
Large brands with regional social media teams (e.g., different Twitter handles for USA,
Europe, Asia) analyzing regional audience interactions separately.

3. Distributed Network Structure

Definition:
Every node in the network is connected to every other node either directly or indirectly. There is
no central point.

Advantages:
 Very resilient — no single point of failure
 Equal distribution of load and data
 Highly scalable

Disadvantages:
 Complex to set up and maintain
 High communication overhead

Example in Social Media Analytics:

Blockchain-based social networks or peer-to-peer data collection platforms, where every
user's actions and data contribute equally to analytics and insights.

4. Hierarchical Network Structure

Definition:
Network is arranged in a tree-like structure where nodes are connected in a parent-child
relationship, often used in organizational setups.

Advantages:
 Clear data flow paths and authority structure
 Efficient management of different levels of data

Disadvantages:
 Not fault-tolerant — failure in higher-level nodes can affect lower levels
 Less flexible

Example in Social Media Analytics:

A corporate social media analysis setup where local branches report to regional offices, which
in turn report to a national headquarters — each level analyzing data relevant to their scope.
Equivalence
Equivalence refers to the similarity in roles or positions of nodes in a network. It shows how
similar two users are in terms of their connections.
 Structural Equivalence: Same neighbors
 Regular Equivalence: Similar types of relationships, not necessarily with the same
nodes

Advantages:
1. Identifies user roles (influencers, hubs, etc.)
2. Helps in user classification
3. Supports targeted content delivery
4. Detects duplicate accounts
5. Aids in role-based recommendation systems
6. Useful for studying competitive brands or products
7. Helps model peer influence
8. Enhances community detection
9. Assists in simplifying large networks
10. Enables user behavior prediction

Disadvantages:
1. High computational complexity
2. Requires detailed network data
3. Sensitive to missing/incomplete data
4. Fails with dynamic networks
5. May miss subtle behavioral differences
6. Assumes uniform importance of all links
7. Cannot handle weighted edges well
8. Scalability issues with large networks
9. Difficult to interpret in heterogeneous networks
10. Misleading if based on weak interactions

Example:
On Twitter, two influencers with similar followers and tweet interactions may be structurally
equivalent, helping analysts group them for campaigns.
Homophily
Homophily is the principle that similar people tend to connect. In social media, it means
people with shared beliefs, interests, or backgrounds often interact more.

Advantages:
1. Enables community detection
2. Helps in content personalization
3. Predicts user behavior
4. Improves friend/follow recommendations
5. Useful in sentiment clustering
6. Supports niche targeting in ads
7. Simplifies model design
8. Facilitates trend forecasting
9. Useful in political/ideological mapping
10. Highlights content virality patterns

Disadvantages:
1. Encourages echo chambers
2. Reduces content diversity
3. Promotes misinformation loops
4. Leads to bias in AI algorithms
5. Limits exposure to new ideas
6. Makes user segmentation harder across diverse groups
7. Overlooks weak ties with influence
8. Can lead to stereotyping in data
9. May not hold true for all platforms
10. Hard to measure accurately

Example:
Facebook groups around a shared hobby or belief (e.g., vegan cooking) where users mainly
interact within the group — a clear case of interest-based homophily.
Clustering
Clustering is grouping a set of nodes in a network so that nodes within a group are more
connected to each other than to others. Common for detecting communities.

Advantages:
1. Reveals hidden communities
2. Aids in market segmentation
3. Improves user targeting
4. Reduces dimensionality in data
5. Helps in fraud detection
6. Supports viral marketing strategies
7. Enables topic modeling
8. Allows detection of interaction hubs
9. Makes visualization easier
10. Supports platform optimization

Disadvantages:
1. May produce overlapping clusters
2. Sensitive to algorithm choice
3. High resource consumption on large networks
4. Not all networks are naturally clusterable
5. Can misclassify peripheral nodes
6. Might ignore inter-cluster ties
7. Requires parameter tuning
8. Can generate meaningless clusters without validation
9. Data noise affects accuracy
10. Interpretation of clusters may be subjective

Example:
Analyzing Instagram data to detect fashion communities by clustering users based on hashtags
and follows.
Snowball Sampling
Snowball sampling is a method where existing participants recruit further participants. It’s
common in studying hidden or hard-to-reach groups on social media.

Advantages:
1. Easy to implement
2. Cost-effective
3. Accesses hidden populations
4. Useful for network mapping
5. Builds trust in closed communities
6. Efficient for initial exploratory research
7. Generates real-world contact networks
8. Provides insight into group influence
9. Requires fewer resources than full surveys
10. Works in the absence of a complete sampling frame

Disadvantages:
1. Not random; sampling bias
2. Overrepresents highly connected users
3. Results not generalizable
4. Depends on referral quality
5. Can miss isolated users
6. Prone to data duplication
7. Ethical concerns if privacy isn't managed
8. Difficult to stop once started
9. Cannot calculate sampling error
10. Results sensitive to initial seed node

Example:
Tracking hate speech users by starting with a known account and expanding through mentions
and followers.
Contact Tracing
In social media, contact tracing maps the flow of information (likes, shares, comments) to
track how a piece of content spreads through a network.

Advantages:
1. Detects origin of viral content
2. Helps identify super-spreaders
3. Useful in misinformation tracking
4. Aids in content strategy planning
5. Models influence paths
6. Supports public health messaging
7. Maps user engagement chains
8. Assists in crisis management
9. Provides time-based diffusion insights
10. Enables targeted countermeasures

Disadvantages:
1. Requires detailed interaction logs
2. Privacy concerns
3. Computationally intensive
4. Complex in dynamic networks
5. Hard to trace across platforms
6. Obscured by retweet bots
7. Ethical concerns in user surveillance
8. Data noise from inactive users
9. Can be misused for profiling
10. Depends heavily on platform access

Example:
Mapping the spread of a fake news story on Facebook by tracing who shared it and when, to
identify early spreaders.
Random Walks
A random walk is a method of moving through a network by selecting a random neighbor at
each step. It’s used for sampling, influence estimation, and ranking.

Advantages:
1. Scalable to large graphs
2. Requires less memory
3. Good for local exploration
4. Useful in PageRank and influence scoring
5. Can reveal hidden structures
6. Works well in incomplete networks
7. Effective for sampling
8. Resistant to noise
9. Simple to implement
10. Adaptable for various models

Disadvantages:
1. May miss rare nodes or communities
2. Biased toward high-degree nodes
3. Unpredictable traversal
4. Doesn’t guarantee coverage
5. May take long to reach certain areas
6. Depends on walk length
7. Sensitive to network sparsity
8. Not suitable for fine-grained analysis
9. May require multiple runs
10. Difficult to tune stopping criteria

Example:
Twitter uses random walk-based algorithms to recommend new accounts to follow based on
indirect connections.
Ego-Centered Networks

An ego-centered network (also known as an egocentric network) is a type of network structure

that focuses on one central node (the “ego”) and all the nodes directly connected to it (called
“alters”), including the relationships among those alters.
 Ego: The main individual or user being studied (e.g., you).
 Alters: The people directly connected to the ego (e.g., your friends).
 Ego Network: Includes ego, alters, and all interactions among them.

A typical ego-centered network contains:

 Node 1 (Ego) — e.g., a Twitter user.
 Nodes 2–n (Alters) — their followers/friends/interactors.
 Edges — direct links between ego and alters, and among alters.

Applications in Social Media and Data Analysis:

1. Influencer Analysis
Determine how influential a user is based on the structure of their ego network.
2. Content Recommendation
Recommend posts or connections based on the ego’s interactions.
3. Behavior Prediction
Analyze past behavior of alters to predict ego’s future actions.
4. Friend/Follower Suggestion
Suggest new contacts by analyzing the alters of similar egos.
5. Information Diffusion
Track how far and fast a user’s post reaches through their ego network.
6. Privacy and Security
Detect spam or bots by studying abnormal ego-centered patterns.
7. Community Engagement
Understand how connected an individual is within a social community.

Advantages:
1. Focused on individual behavior — highly personalized analysis.
2. Efficient for small-scale social media studies.
3. Useful in micro-influence marketing.
4. Helps identify key supporters or detractors.
5. Easy to visualize and interpret.
6. Enables localized interventions (e.g., targeted ads).
7. Supports qualitative analysis of relationships.
8. Effective for tracing misinformation from a user.
9. Requires less data than global networks.
10. Can reveal tight-knit communities or echo chambers.
Disadvantages:
1. Doesn’t capture broader network effects.
2. Misses weak ties outside the ego’s reach.
3. Not suitable for analyzing global trends.
4. May ignore indirect influence from second-degree nodes.
5. Biased by ego’s activity level.
6. Alter connections may be incomplete.
7. Dynamics are hard to track in real time.
8. Susceptible to platform privacy limitations.
9. Ineffective for viral content spread beyond ego’s network.
10. Results are not generalizable to the full population.

Example in Social Media Analytics:

Suppose a Facebook user (the ego) frequently interacts with 15 friends (alters). By examining:
 Who comments on the ego’s posts
 Which friends interact with each other
 How information flows between them
Analysts can determine:
 The ego’s centrality
 Influence level
 Content reach potential
 Behavioral patterns of the immediate social circle
Dominance Hierarchies

A dominance hierarchy refers to a social structure where individuals are ranked relative to each
other in terms of influence, authority, or control. In social media, this means some users have
more power or visibility in a network, influencing the behavior and opinions of others.
 Often visualized as a pyramid or tree structure, with dominant users at the top and
followers or less influential users below.

Characteristics:
1. Ranking: Users are ranked by metrics like followers, engagement, retweets, or mentions.
2. Asymmetry: Influence is not mutual; one user often influences more than they are
influenced.
3. Stability: Hierarchies tend to be stable over time unless disrupted by major events (viral
posts, scandals, etc.).
4. Control: Top-tier users (influencers, celebrities) can shape trends, opinions, or even
market behavior.

Applications in Social Media Analytics:

1. Influencer Marketing
Identify users at the top of the hierarchy for brand promotion.
2. Trend Prediction
Monitor high-ranking users to predict viral trends or hashtags.
3. Misinformation Detection
Track dominant sources spreading fake news or biased content.
4. Community Structure Analysis
Determine power dynamics within online groups or fan communities.
5. User Engagement Strategy
Tailor content to target users based on their position in the hierarchy.
6. Sentiment Impact Study
Study how dominant users shift public opinion.
7. Crisis Management
Engage top influencers to counteract negative campaigns.

Advantages:
1. Reveals power dynamics in social media.
2. Helps in targeted marketing and messaging.
3. Improves influencer discovery.
4. Aids in detecting information gatekeepers.
5. Useful for behavioral analysis in online groups.
6. Allows micro-targeting based on user rank.
7. Enhances community detection by rank.
8. Tracks real-time influence shifts.
9. Supports reputation scoring.
10. Can help prioritize content moderation efforts.
Disadvantages:
1. May reinforce social inequality or echo chambers.
2. Difficult to measure across platforms uniformly.
3. Influencer metrics can be manipulated (e.g., fake followers).
4. Changes in hierarchy can be hard to detect in real time.
5. Over-focus on top users may ignore niche voices.
6. Cannot capture informal or hidden influence.
7. Privacy concerns in tracking user behavior and rank.
8. Biased by platform algorithms (e.g., who gets recommended).
9. Difficult to establish objective dominance criteria.
10. Hierarchies may vary across topics (someone dominant in tech might not be in fashion).

Example in Social Media Analytics:

On Twitter, a dominance hierarchy might show:
 @ElonMusk at the top with millions of followers and high retweet rates.
 Tech influencers below him, reacting to or amplifying his tweets.
 General users at the bottom, mostly consuming content and following trends.
This hierarchy allows analysts to:
 Predict which tweets will go viral.
 Determine who initiates conversations and who spreads them.
 Design targeted campaigns that leverage the hierarchy.
Third-Party Records
Third-party records refer to external data sources that are not directly collected from the
social media platform itself, but still provide valuable contextual, demographic, behavioral,
or transactional data related to users or content.
These records are maintained by independent organizations, apps, data brokers, or tools that
collect, aggregate, or enhance social media data.

Types of Third-Party Records:

1. Demographic Databases (e.g., census data, surveys)
2. Consumer Purchase Data (e.g., retail, loyalty card programs)
3. Browser/Device Tracking (via cookies, apps, SDKs)
4. Public Records (e.g., voter lists, property ownership)
5. Data Brokers (e.g., Acxiom, Oracle Data Cloud)
6. CRM/Marketing Tools (e.g., HubSpot, Salesforce)
7. Web Scraping Tools (from blogs, forums, websites)
8. Mobile App Usage Data
9. Third-party Analytics Tools (e.g., Hootsuite, Brandwatch)
10. Search Behavior Data (from ad networks or SEO tools)

Importance in Social Media Analytics:

Social media platforms like Facebook, Twitter, or Instagram provide only partial views of user
behavior. Third-party records help enrich this data for better profiling, targeting, and
insight generation.

Applications in Social Media and Data Analysis:

1. Audience Profiling
Combine social media likes with offline purchase behavior to build robust customer
profiles.
2. Sentiment Enhancement
Merge social opinions with historical behaviors to better understand intent.
3. Market Segmentation
Use geographic or economic records to segment social media users into target groups.
4. Campaign Performance Tracking
Link social engagement with actual purchase data from third-party sources.
5. Ad Targeting and Retargeting
Use cookie-based third-party data to target users across platforms.
6. Brand Reputation Analysis
Enhance social media listening with records from reviews and forums.
7. Crisis Detection and Management
Merge social panic or protest data with third-party local records (e.g., crime, weather).
8. Political Forecasting
Combine voter records with political tweets and campaign interactions.
9. Fake Account Detection
Cross-check user behavior with known bot lists or third-party blacklists.
10. Influencer Evaluation
Validate social influencers’ claims using external engagement or sales data.
Advantages:
1. Enriches Social Media Data with external context.
2. Enables advanced behavioral targeting.
3. Helps in cross-platform analysis.
4. Supports offline-to-online attribution.
5. Validates user demographics and interests.
6. Improves machine learning models with extra features.
7. Helps detect anomalies or frauds.
8. Reduces bias from social-only datasets.
9. Enables multi-dimensional audience segmentation.
10. Enhances predictive analytics and decision-making.

Disadvantages:
1. Privacy concerns with data sharing and tracking.
2. Ethical issues in using consumer data without consent.
3. May violate platform policies (e.g., Facebook bans some data merges).
4. Data accuracy may vary across sources.
5. Can lead to profiling bias or discrimination.
6. Integration complexity — different formats, standards.
7. May involve high costs (licensed or brokered data).
8. GDPR and CCPA compliance issues.
9. Data may become outdated quickly.
10. Risk of re-identification from anonymous datasets.

Example in Social Media Analytics:

A brand wants to run a hyper-targeted campaign for a new fitness product. They:
 Use Instagram data to find users posting about workouts.
 Combine it with third-party purchase data showing people who recently bought fitness
equipment.
 Further filter using location data from mobile apps.
The result: A high-ROI marketing campaign targeting real fitness enthusiasts with both intent
and buying power.
Affiliation Networks
An affiliation network is a bipartite network where two different types of nodes exist, and
links only form between types, not within.
 Common structure: Individuals ↔ Events, or Users ↔ Groups
 It represents membership, participation, or association with a common activity or
object.

Applications:
1. Community Detection: Identify groups of users participating in similar events or topics.
2. Influence Analysis: See which users span multiple interest groups.
3. Recommendation Systems: Suggest groups, hashtags, or pages based on shared
affiliations.
4. Trend Prediction: Spot emerging interests through common affiliations.
5. Brand Targeting: Find where audiences overlap between different brands or influencers.

Advantages:
1. Captures group-based dynamics.
2. Useful in modeling co-participation.
3. Great for studying social influence via shared interests.
4. Allows projection into single-mode networks (e.g., user–user based on co-affiliation).
5. Helps in personalization and segmentation.
6. Efficient for detecting hidden connections.
7. Supports event-based marketing strategies.
8. Enhances collaborative filtering models.
9. Bridges content and user behavior.
10. Offers scalable structures for large datasets.
Disadvantages:
1. Bipartite complexity requires special algorithms.
2. May overlook direct interactions (like messages).
3. Data sparsity if affiliations are niche.
4. Over-representation of popular groups.
5. May require normalization to avoid bias.
6. Projected networks may lose information.
7. Ambiguity in defining the affiliation threshold.
8. Privacy concerns when mapping user interests.
9. Cannot capture time-sensitive interactions easily.
10. Less effective for purely conversational data.

Citation Networks
A citation network is a directed graph where nodes represent documents or users, and edges
represent a citation or reference — meaning one node refers to or acknowledges another.
 In social media, citations can be mentions, tags, shares, or replies.

Applications:
1. Influencer Tracking: Who gets cited or referenced the most.
2. Information Flow Mapping: Track how ideas spread.
3. Trend Source Analysis: Identify who started a viral trend.
4. Sentiment Influence Study: Trace sentiment shifts via citations.
5. Academic Social Graphs: Track citation patterns among researchers on platforms like
ResearchGate.
6. Topic Evolution: Analyze how content citations change over time.
7. Credibility Scoring: Use citation counts to rank sources.
8. Spam or Bot Detection: Abnormal citation patterns may signal inauthentic behavior.
9. Cross-platform Behavior: Trace references across Twitter, blogs, YouTube, etc.
10. Misinformation Tracking: Trace the spread of false claims.
Advantages:
1. Captures directional influence.
2. Tracks knowledge or content diffusion.
3. Helps identify central or authoritative users.
4. Allows detailed temporal analysis.
5. Well-established graph theory for citations.
6. Used in reputation and credibility modeling.
7. Supports root cause tracing.
8. Highlights content reuse or remixing trends.
9. Adaptable across media types.
10. Aids in detecting content plagiarism or derivative works.

Disadvantages:
1. Citations may not imply endorsement.
2. Hard to track if content is cited without explicit linking.
3. Can be gamed (e.g., fake mentions).
4. Complex temporal modeling needed.
5. Not always reciprocal.
6. Doesn’t always show full context.
7. High-volume users can skew visibility.
8. May require NLP to detect implicit citations.
9. Cross-platform linking can be fragmented.
10. Not suitable for private messages or DMs.
Peer-to-Peer (P2P)
A Peer-to-Peer (P2P) network is a decentralized communication model in which each node
(peer) acts as both a client and a server. Unlike client-server architectures, where data flows
from central servers to users, P2P networks allow direct communication and data exchange
between users without a central authority.
In social media analytics, P2P concepts are useful for analyzing user-to-user interactions,
decentralized platforms, and distributed content sharing.

Characteristics of P2P Networks:

1. Decentralization – No single control point.
2. Symmetry – All peers have equal capabilities.
3. Scalability – Easily grows as more peers join.
4. Redundancy – Multiple copies of data across peers.
5. Fault Tolerance – If one peer fails, others still function.

Examples in Social Media:

 BitTorrent-based content sharing (file-sharing communities).
 Mastodon – Federated social network where each server is a peer.
 PeerTube – Decentralized video platform where users host videos.
 Crypto-based messaging apps (like Status or Session) using P2P encryption.
 Twitter interactions (in a graph-theoretical sense) can sometimes mimic P2P where
each user sends and receives without centralized approval.

Advantages of P2P in Social Media Analytics:

1. No single point of failure – Highly resilient systems.
2. Improved data privacy and ownership.
3. Democratized information flow – Less control by tech giants.
4. Natural modeling of user-to-user relationships.
5. Efficient content distribution (e.g., viral media).
6. Supports decentralization and Web3 principles.
7. Enables distributed moderation and governance.
8. Better scalability for global communication.
9. Good for studying community-driven platforms.
10. Reduces censorship – harder to control via central authority.

Disadvantages of P2P in Social Media Analytics:

1. Difficult to monitor or regulate.
2. Data fragmentation – hard to collect complete datasets.
3. Higher risk of misinformation and abuse.
4. Performance inconsistencies – some peers may be unreliable.
5. Security concerns – increased risk of peer impersonation or spoofing.
6. Lack of central moderation.
7. Harder to enforce platform policies.
8. Limited mainstream adoption (vs. client-server apps like Facebook).
9. Data synchronization challenges.
10. Complex network modeling and analysis.

Analytical Use Case:

Imagine studying a viral campaign in a decentralized video platform like PeerTube:
 You analyze how a video spreads peer-to-peer across instances.
 Each peer hosts and shares content — no central server.
 You model nodes as peers, and edges as direct shares or re-uploads.
 Your social media analysis focuses on content propagation paths, resilience of views,
and cross-instance engagement.
Recommender Networks

A Recommender Network** is a type of graph-based model used in recommender systems,

where nodes represent users and items (e.g., products, posts, videos, etc.), and edges represent
interactions or preferences such as likes, views, ratings, or shares.
In social media, these networks power recommendation engines for:
 Suggested friends
 Trending topics
 Videos or posts you may like
 Ads tailored to your interests

Structure:
 Bipartite Graph: Users ↔ Items
 Edges: Indicate preferences (clicks, likes, purchases, etc.)
 Can be enhanced with social links, user similarity, or item similarity

Applications in Social Media Analytics:

1. Friend or Follower Suggestions (e.g., LinkedIn, Facebook)
2. Content Recommendations (e.g., TikTok's For You page)
3. Ad Targeting
4. Hashtag Recommendations
5. Group or Community Suggestions
6. Influencer Discovery
7. Feed Personalization
8. Sentiment-aware Product Suggestions
9. Cross-platform Content Promotion
10. Behavior Prediction

Advantages:
1. Personalized user experience.
2. Increases user engagement and retention.
3. Learns user preferences over time.
4. Scales well to millions of users/items.
5. Enables targeted marketing.
6. Drives content discovery.
7. Supports multi-modal data (text, image, video).
8. Can integrate collaborative and content-based filtering.
9. Useful in cross-selling or upselling.
10. Enhances social media monetization strategies.

Disadvantages:
1. Cold-start problem (new users/items have no data).
2. Privacy issues from behavioral tracking.
3. Filter bubble effect (users see only similar content).
4. Bias toward popular content.
5. Hard to explain recommendations.
6. Complex to build and maintain.
7. Vulnerable to spam or manipulation.
8. May reduce content diversity.
9. Needs continuous retraining with new data.
10. Risk of reinforcing existing preferences or stereotypes.

Example:
In YouTube:
 User A watches videos on cooking.
 The recommender network connects them with similar users and videos.
 Based on this, YouTube recommends cooking channels and recipe playlists using
collaborative filtering.

Biological Networks

Biological Networks** are graphs where nodes represent biological elements such as genes,
proteins, or cells, and edges represent biological interactions (e.g., protein–protein interactions,
gene regulation, etc.).
In social media analytics, these networks are not directly biological, but analogies from
biology are applied to model complex user behavior, information propagation, and
community formation.
Types of Biological Analogies in Social Media:
1. Epidemiological Models: Used to model viral content spread.
2. Neural Network Structures: For deep learning on user data.
3. Gene Regulatory Networks: Analogy for content influence networks.
4. Protein Networks: Used for interaction modeling among users.
5. Cellular Automata: To simulate behavioral evolution in communities.

Advantages:
1. Offers natural models for complexity and non-linear dynamics.
2. Enables simulation of viral content spread.
3. Helps identify critical influencers like proteins in biology.
4. Useful for predicting information epidemics.
5. Helps model resilience of communities.
6. Supports multi-layer network analysis (genetic = user traits).
7. Can model emergent behavior in online platforms.
8. Encourages interdisciplinary approaches.
9. Simulates network evolution over time.
10. Useful in misinformation detection and control.

Disadvantages:
1. Complex to model and interpret.
2. May require deep biological or mathematical knowledge.
3. Not all biological analogies map accurately to human behavior.
4. Difficult to validate models empirically.
5. Large-scale simulations can be computationally expensive.
6. Risk of overfitting or oversimplification.
7. Ambiguity in translating biological rules to social contexts.
8. Not well suited for short-term trend predictions.
9. Requires integration with other models.
10. Limited by data availability and granularity.

Example:
 Epidemic Modeling in Social Media:
A new hashtag starts trending. Analysts use SIR (Susceptible-Infected-Recovered)
models (from biology) to simulate how fast it spreads, how many users get "infected"
(start using it), and how long it stays active.