Data Mining Unit 5 Notes | Web, Spatial, Temporal and Text Mining

1. Introduction to Web Mining 14 Marks

Web mining is the application of data-mining techniques to discover useful information and patterns from web documents, hyperlinks, server logs and user interactions.

Need for Web Mining

The web contains an extremely large amount of information.
Web data is heterogeneous and continuously changing.
Users need relevant information quickly.
Businesses want to understand customer behaviour.
Search engines need to rank and organize web pages.

Major Types

Web Mining ├── Web Content Mining ├── Web Structure Mining └── Web Usage Mining

General Web-Mining Process

Web Data Collection | Preprocessing | Pattern Discovery | Pattern Analysis | Useful Web Knowledge

Applications

Search engines
Recommendation systems
Personalized websites
Customer behaviour analysis
Fraud and bot detection

2. Characteristics of Web Data 14 Marks

Large-scale: billions of pages and interactions.
Heterogeneous: text, images, audio, video and structured data.
Semi-structured: HTML and XML have tags but flexible organization.
Dynamic: pages and links change frequently.
Distributed: information is stored across many servers.
Noisy: pages may contain advertisements, duplicate content and irrelevant data.
Linked: hyperlinks create a graph structure.
Usage-oriented: logs record user navigation and behaviour.

Traditional database-mining methods cannot always be directly applied because web data is unstructured, distributed and rapidly changing.

3. Web Content Mining 14 Marks

Web content mining extracts useful information from the actual contents of web pages.

Content Sources

Text documents
HTML and XML pages
Images
Audio and video
Tables and metadata
Product descriptions and reviews

Approaches

Agent-Based Approach

Intelligent agents automatically search, filter and organize web information.

Database Approach

Web content is transformed into structured or semi-structured form for querying and analysis.

Techniques

Information extraction
Natural language processing
Classification and clustering
Topic modelling
Sentiment analysis
Multimedia feature extraction

An e-commerce system can mine product descriptions and customer reviews to identify product features and user sentiment.

4. Web Structure Mining 14 Marks

Web structure mining discovers knowledge from the hyperlink structure connecting web pages.

Web as a Directed Graph

Page A ─────→ Page B │ │ ↓ ↓ Page C ←───── Page D

Web pages are graph nodes.
Hyperlinks are directed edges.
Incoming links may indicate authority.
Outgoing links may indicate useful hub pages.

PageRank

PageRank assigns importance to a page based on the number and quality of incoming links.

PR(A) = (1 − d) + d Σ [PR(Tᵢ) / C(Tᵢ)]

HITS Algorithm

Authority: a page containing valuable information.
Hub: a page linking to useful authority pages.

Applications

Search-result ranking
Community discovery
Important-page identification
Link-spam detection

5. Web Usage Mining 14 Marks

Web usage mining discovers user-access and navigation patterns from web server logs, cookies, browser records and clickstream data.

Data Sources

Web server logs
Proxy server logs
Browser logs
Cookies
Session data
Clickstream records

Sample Log Fields

IP Address Date and Time Requested URL HTTP Status Referrer Page Browser / User Agent Transferred Bytes

Discovered Patterns

Frequently visited pages
Common navigation paths
Average session duration
Entry and exit pages
Abandoned transactions
User segments

6. Web Usage Mining Process 14 Marks

1. Data Collection

Collect logs, cookies and clickstream information.

2. Data Cleaning

Remove requests for images, style sheets, scripts and bot traffic when not required.

3. User Identification

Identify individual users using login details, cookies or combinations of IP and browser data.

4. Session Identification

Divide a user's activity into separate visits or sessions.

5. Path Completion

Infer missing page requests caused by browser or proxy caching.

6. Pattern Discovery

Apply association, clustering, classification and sequential-pattern mining.

7. Pattern Analysis

Filter unimportant patterns and present useful findings.

7. Types of Web Mining – Comparison 14 Marks

Type	Data Source	Main Purpose	Example
Content Mining	Page content	Extract information	Review sentiment
Structure Mining	Hyperlinks	Study page relationships	Page ranking
Usage Mining	Logs and clickstreams	Study user behaviour	Personalized recommendation

8. Spatial Data Mining 14 Marks

Spatial data mining discovers useful patterns and relationships from data containing geographic locations, shapes, distances and spatial relationships.

Examples of Spatial Data

Maps and geographic information systems
Satellite images
Location coordinates
Road and traffic networks
Weather regions
Land-use information

Spatial Relationships

Near
Far
Inside
Overlaps
Adjacent
Intersects
North or south of

Challenges

Complex spatial objects
Spatial dependency
Very large geographic datasets
Multiple resolutions
Uncertainty in measurements

9. Spatial Mining Tasks and Techniques 14 Marks

Spatial Classification

Assigns spatial objects to predefined classes.

Spatial Clustering

Groups nearby or spatially similar objects.

Spatial Association Rules

Areas near major roads → high commercial development.

Spatial Trend Analysis

Discovers changes along a spatial direction or distance.

Spatial Outlier Detection

Finds objects differing from neighbouring objects.

Co-Location Pattern Mining

Discovers features frequently located together.

Hospitals, pharmacies and diagnostic centres may frequently occur in nearby locations.

Applications

Urban planning
Agriculture
Environmental monitoring
Disaster management
Traffic management
Location-based services

10. Temporal Data Mining 14 Marks

Temporal data mining discovers patterns, sequences, changes and relationships from data having a time dimension.

Types of Temporal Data

Time-series data
Event sequences
Transaction sequences
Interval data
Streaming data

Examples

Daily stock prices
Hourly temperature readings
Website click sequences
Medical-event histories
Machine-sensor streams

Important Temporal Concepts

Time order
Duration
Periodicity
Trend
Seasonality
Time lag

11. Temporal Mining Techniques 14 Marks

Time-Series Analysis

Studies trends, cycles and seasonal behaviour in numerical observations over time.

Sequential Pattern Mining

Discovers frequently occurring ordered events.

View Product → Add to Cart → Purchase

Temporal Association Rules

Event X → Event Y within time interval Δt

Periodicity Analysis

Discovers patterns repeating at regular intervals.

Trend Analysis

Identifies long-term increase, decrease or stable behaviour.

Change Detection

Detects sudden or gradual changes in a data stream.

Applications

Sales forecasting
Weather prediction
Financial analysis
Healthcare monitoring
Equipment failure prediction

12. Text Mining 14 Marks

Text mining extracts useful information, concepts, relationships and patterns from large collections of unstructured text documents.

Text Data Sources

News articles
Emails
Reports
Social-media posts
Customer reviews
Research papers
Support tickets

Major Tasks

Text classification
Document clustering
Information extraction
Topic discovery
Sentiment analysis
Keyword extraction
Document summarization

Text Mining vs Data Mining

Data Mining	Text Mining
Mainly structured data	Mainly unstructured text
Uses attributes and records	Uses words, phrases and documents
Less language processing	Requires NLP and text preprocessing

13. Text Mining Process 14 Marks

1. Document Collection

Collect documents from files, websites, databases or APIs.

2. Tokenization

Break text into words, terms or tokens.

3. Stop-Word Removal

Remove common words such as the, is and of when they do not add useful meaning.

4. Stemming or Lemmatization

Reduce words to a common root or dictionary form.

5. Feature Representation

Convert documents into numerical vectors.

6. Pattern Discovery

Apply classification, clustering, topic modelling or sentiment analysis.

Term Frequency

TF(t,d) = Frequency of term t in document d

TF-IDF

TF-IDF(t,d) = TF(t,d) × log(N / DF(t))

14. Information Retrieval and Text Mining 14 Marks

Information retrieval finds documents relevant to a user's query, while text mining discovers hidden patterns and knowledge across document collections.

Information Retrieval Components

Document collection
Indexing
User query
Similarity matching
Ranking

Precision

Precision = Relevant Retrieved Documents / Total Retrieved Documents

Recall

Recall = Relevant Retrieved Documents / Total Relevant Documents

Vector Space Model

Documents and queries are represented as vectors, and similarity may be measured using cosine similarity.

Cosine Similarity = (A · B) / (||A|| × ||B||)

15. Security Issues in Data Mining 14 Marks

Security issues concern protecting mining systems, datasets, algorithms and discovered knowledge from unauthorized access, modification or misuse.

Major Security Threats

Unauthorized data access
Data theft
Data tampering
Malicious insiders
Model theft
Inference attacks
Poisoned training data
System and network attacks

Security Requirements

Confidentiality: prevent unauthorized disclosure.
Integrity: prevent unauthorized modification.
Availability: keep systems accessible.
Authentication: verify user identity.
Authorization: control user permissions.

16. Privacy Issues in Data Mining 14 Marks

Privacy issues arise when personal or sensitive information is collected, combined, analyzed or revealed without proper knowledge, consent or protection.

Major Privacy Risks

Collection without informed consent
Re-identification of anonymous users
Profiling and continuous tracking
Secondary use of data
Disclosure of sensitive attributes
Combining datasets to infer private facts

Examples

Inferring health conditions from purchase history
Tracking browsing behaviour across websites
Discovering identity from supposedly anonymous records

Privacy Principles

Data minimization
Purpose limitation
User consent
Transparency
Limited retention
User access and correction

17. Ethical Issues in Data Mining 14 Marks

Ethical issues concern whether the collection, analysis and use of data is fair, transparent, responsible and respectful of human rights.

Major Ethical Concerns

Bias and discrimination
Unfair automated decisions
Lack of transparency
Manipulative personalization
Excessive surveillance
Misleading correlations
Use of data beyond its original purpose
Lack of human accountability

Algorithmic Bias

A mining model may reproduce unfair patterns present in historical data.

Transparency

Users should understand what data is collected and how important decisions are made.

Accountability

Organizations must remain responsible for the impact of data-driven decisions.

Ethical data mining should be fair, explainable, privacy-aware, secure and used only for legitimate purposes.

18. Security and Privacy Protection Techniques 14 Marks

Access Control

Allow only authorized users to access sensitive data.

Encryption

Protect data during storage and transmission.

Anonymization

Remove or modify direct identifiers such as name and phone number.

Pseudonymization

Replace identifiers with artificial values while keeping mapping information separately.

Data Masking

Hide sensitive parts of data.

k-Anonymity

Each record should be indistinguishable from at least k−1 other records for selected identifying attributes.

Differential Privacy

Adds carefully controlled randomness so that an individual's participation has limited influence on released results.

Audit Logs

Record access and changes for accountability.

Privacy-Preserving Data Mining

Discovers patterns without revealing unnecessary sensitive information.

19. Applications of Advanced Data Mining 14 Marks

Web Mining Applications

Search ranking
Recommendation systems
Website personalization
Customer journey analysis

Spatial Mining Applications

Traffic analysis
Urban planning
Disaster prediction
Agricultural monitoring

Temporal Mining Applications

Demand forecasting
Predictive maintenance
Financial trend analysis
Medical monitoring

Text Mining Applications

Sentiment analysis
Spam filtering
Document categorization
Customer-support analysis

Important Comparisons

Web Content vs Structure vs Usage Mining

Feature	Content	Structure	Usage
Focus	Page information	Hyperlinks	User behaviour
Data	Text and media	Link graph	Logs and sessions
Result	Extracted knowledge	Page importance	Navigation patterns

Spatial vs Temporal Mining

Spatial Mining	Temporal Mining
Focuses on location and geometry	Focuses on time and sequence
Uses distance and topology	Uses order, duration and periodicity
Example: traffic hotspot	Example: monthly traffic trend

Security vs Privacy vs Ethics

Security	Privacy	Ethics
Protects systems and data	Protects personal information	Ensures responsible use
Access, encryption, integrity	Consent and limited disclosure	Fairness and accountability

Unit 5 Quick Revision

Web mining discovers patterns from web content, links and usage data.
Content mining analyzes page information.
Structure mining analyzes hyperlinks.
Usage mining analyzes logs and clickstreams.
Spatial mining handles location, distance and topology.
Temporal mining handles time order, trends and sequences.
Text mining converts unstructured text into useful knowledge.
TF-IDF represents term importance in documents.
Security protects confidentiality, integrity and availability.
Privacy protects personal and sensitive information.
Ethics requires fairness, transparency and accountability.
Anonymization, encryption and access control reduce risks.

Important RGPV Exam Questions

Long Answer Questions

Define web mining and explain its major types.
Explain web content mining with techniques and applications.
Explain web structure mining and the concepts of PageRank and HITS.
Explain web usage mining and its complete process.
Compare web content, structure and usage mining.
Define spatial mining and explain its major tasks.
Explain spatial association, clustering and trend analysis.
Define temporal mining and explain its techniques.
Explain sequential and temporal pattern mining.
Define text mining and explain its complete process.
Explain TF-IDF and document representation.
Differentiate information retrieval and text mining.
Explain security issues in data mining.
Discuss privacy issues and privacy-preserving techniques.
Explain ethical issues in data mining.

Short Answer Questions

Define web mining.
What is web content mining?
What is a web server log?
Define spatial data.
What is temporal data mining?
Define tokenization.
What is TF-IDF?
Define anonymization.
What is algorithmic bias?
What is differential privacy?

Exam Strategy: Draw the web-mining classification diagram, web-usage process, spatial examples, temporal sequence and text-mining pipeline. Write security, privacy and ethical issues under separate headings.

Download Study Resources

Unit 5 PDF

Printable detailed notes will be uploaded soon.

Coming Soon

Important Questions

Most expected Unit 5 questions coming soon.

Coming Soon

PYQ Analysis

Repeated web and text-mining questions coming soon.

Coming Soon

Frequently Asked Questions

Web content mining, web structure mining and web usage mining.

It mainly comes from server logs, browser records, cookies, sessions and clickstream data.

Spatial mining focuses on location and geographic relationships, whereas temporal mining focuses on time, order and sequence.

Raw text is unstructured, so tokenization, stop-word removal and stemming convert it into a form suitable for mining.

No. Security protects data and systems from unauthorized actions, while privacy concerns the proper collection and use of personal information.

Related Study Pages