Text Analytics, Web Usage Mining & Sentiment Analysis
Idea in plain words
Web Analytics taxonomy: Web Content Mining (text/media on pages), Web Structure Mining (hyperlink topologies, PageRank), Web Usage Mining (server logs, clickstreams).
Exam tip
High-yield exam topic (High).
Explanation
Over 80% of enterprise information resides in unstructured text documents, customer emails, and web server logs. Web usage mining parses common log format (CLF) server entries to discover conversion bottlenecks and path drop-offs in e-commerce purchase flows. Concurrently, sentiment analysis interrogates social media mentions and customer support tickets, calculating polarity and subjectivity scores to detect product dissatisfaction before it causes measurable customer churn.
Key points
- Clickstream Analysis: Tracking user navigation paths, bounce rates, session durations, and cart abandonment funnels.
- Text Analytics: Structuring unstructured corporate documents (emails, support tickets, reviews) via NLP pipelines.
- Sentiment Analysis (Opinion Mining): Classifying customer text reviews as positive, neutral, or negative using lexicon-based or supervised ML classifiers.
- Search Engine Optimization (SEO) analytics: Monitoring organic search rankings, keyword CTR, and crawl error telemetry.
Example
Example
An airline analyzing customer tweets using sentiment analysis: detecting sudden spikes in negative sentiment containing the word 'delay' to rapidly deploy customer care agents.
Code examples
import re
from collections import Counter
# Log parsing snippet for Web Usage Mining
log_line = '192.168.1.50 - - [28/Sep/2026:10:15:30 +0545] "GET /checkout HTTP/1.1" 200 4520'
log_pattern = r'(\S+) \S+ \S+ \[(.*?)\] "(\S+) (\S+) \S+" (\d{3}) (\d+)'
match = re.match(log_pattern, log_line)
if match:
ip, timestamp, method, endpoint, status, bytes_sent = match.groups()
print(f"IP: {ip} visited {endpoint} at {timestamp} with status {status}")
Exam questions
- [8 Marks] Differentiate between Web Content Mining, Web Structure Mining, and Web Usage Mining with practical examples.
- [6 Marks] Explain the methodology and practical applications of Sentiment Analysis (Opinion Mining) for corporate brand monitoring.