Social Media & Web Analytics (AD-703 (D)) - Unit 3 Short Notes

How unit 3 is examined

This unit covers web metrics, reports and Google Analytics, KPIs, network measures and random graphs, social context, analytics tools and NLP for short text; no topic was asked recently, so each is short but complete.

Common Metrics: Hits, Page Views, Visits, Unique Page Views, Bounce, Bounce Rate & its Improvement, Average Time on Site

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Common web metrics are the basic counts and ratios that measure how much traffic a site receives and how visitors behave on it.</mark>

Key points.

1. A hit is any single file request to the server (page, image, script), so it overstates real traffic; a page view counts one page load, and a unique page view counts a page once per visit.
2. A visit is one session of activity by a visitor; a bounce is a visit that views only one page and leaves.
3. Bounce rate = (single-page visits / total entry visits) x 100; improve it with faster loading, relevant content, clear calls to action and better navigation.
4. Average time on site = total duration of all visits / number of visits.

Real Time Report, Traffic Source Report, Custom Campaigns, Content Report, Google Analytics

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Google Analytics is a free tool that collects visitor data with a JavaScript tag and presents it in reports on audience, traffic sources and content.</mark>

Key points.

1. The real-time report shows users active on the site right now, with their pages, locations and sources.
2. The traffic source report shows where visitors came from: organic search, direct, referral, social or paid.
3. Custom campaigns tag URLs with UTM parameters (source, medium, campaign) so each marketing effort is tracked separately.
4. The content report shows page views, time on page and bounce rate per page, revealing which content works.

Key-Performance Indicator: Need, Characteristics, Perspective and Uses

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>A KPI is a measurable value that shows how effectively a business is achieving a key objective.</mark>

Key points.

1. KPIs are needed because raw metrics do not show progress; KPIs tie numbers to goals and support decisions.
2. Good KPIs are specific, measurable, relevant to goals, timely and actionable, and they are usually ratios or rates.
3. Perspective: a KPI is chosen per business goal (sales, leads, engagement, support), e.g. conversion rate for e-commerce.
4. Uses: tracking progress, benchmarking, spotting problems early and justifying spend.

Graphs and Matrices: Basic Measures for Individuals and Networks

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. ==A social network is a graph of nodes (people) and edges (ties), stored as an adjacency matrix where entry $a_{ij}=1$ if $i$ and $j$ are linked.==

Key points.

1. Degree of a node is its number of ties; degree centrality = $\deg(v)/(n-1)$.
2. Closeness centrality measures how near a node is to all others; betweenness measures how often it lies on shortest paths between others.
3. Network density = $2m/(n(n-1))$ for $m$ edges and $n$ nodes.
4. Clustering coefficient shows how many of a node's neighbours are also linked to each other.

Random Graphs & Network Evolution

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>An Erdos-Renyi random graph $G(n,p)$ has $n$ nodes in which each possible edge exists independently with probability $p$.</mark>

Key points.

1. Expected number of edges is $p\,n(n-1)/2$ and expected degree is $p(n-1)$.
2. Degrees follow a binomial (about Poisson) distribution, so random graphs have no hubs.
3. Real networks evolve by growth and preferential attachment, where new nodes link to popular nodes, giving power-law degrees.
4. Real networks show small-world behaviour: short paths and high clustering.

Social Context: Affiliation & Identity

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Social context is the set of groups, affiliations and identities that shape who people connect with and how they behave online.</mark>

Key points.

1. Affiliation networks link people to groups or events they share, such as pages, clubs or hashtags.
2. Identity is how a person presents and is categorised, through profile, interests and group membership.
3. Homophily is the tendency of similar people to connect, so shared affiliation predicts ties.
4. Influence and selection both create similar neighbours: people befriend similar people and also become like their friends.

Web analytics Tools: A/B testing, Online Surveys, Web Crawling and Indexing

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Web analytics tools are methods that test, ask or collect data to understand and improve a website.</mark>

Key points.

1. A/B testing shows two page versions to random visitor halves and picks the one with the better conversion rate.
2. Online surveys ask visitors directly, giving qualitative reasons that numbers cannot show.
3. Web crawling is automated fetching of pages by following links.
4. Indexing stores the crawled content in a searchable structure so a search engine can retrieve pages by keyword.

Natural Language Processing Techniques for Micro-Text Analysis

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Micro-text analysis applies NLP to very short, noisy texts such as tweets and comments to extract meaning, topics and sentiment.</mark>

Key points.

1. Preprocessing cleans the text by tokenisation, lowercasing, stop-word removal and stemming, and handles hashtags, emoticons and slang.
2. Sentiment analysis classifies text as positive, negative or neutral using lexicons or machine learning.
3. Named entity recognition and topic modelling find people, brands and themes in posts.
4. Short length and sparse context make accuracy lower than for long documents.

Last-minute revision

• Hit = any file request; page view = one page load; unique page view = page counted once per visit.
• Bounce rate = single-page visits / entry visits x 100.
• Average time on site = total visit time / number of visits.
• GA reports: real time, traffic source, content; UTM tags track custom campaigns.
• KPI = measurable value tied to a business goal.
• Degree centrality = deg/(n-1); density = 2m/(n(n-1)).
• Random graph $G(n,p)$: expected edges $p\,n(n-1)/2$.
• Preferential attachment gives power-law degrees; homophily means like connects with like.
• A/B testing compares two versions by conversion rate.
• Crawl, then index, then search.
• Micro-text NLP: preprocess, then sentiment, entities and topics.

Memory hooks

• Hits are noisy, page views are cleaner, visits are sessions.
• KPI = Key number tied to a Purpose and an Indicator of progress.
• Erdos-Renyi is random with no hubs; rich-get-richer creates hubs.
• Crawl finds, index files, search fetches.

Coverage checklist

• Common Metrics: Hits, Page Views, Visits, Unique Page Views, Bounce, Bounce Rate & its Improvement, Average Time on Site (no past questions).
• Real Time Report, Traffic Source Report, Custom Campaigns, Content Report, Google Analytics (no past questions).
• Key-Performance Indicator: Need, Characteristics, Perspective and Uses (no past questions).
• Graphs and Matrices- Basic Measures for Individuals and Networks (no past questions).
• Random Graphs & Network Evolution (no past questions).
• Social Context: Affiliation & Identity (no past questions).
• Web analytics Tools: A/B testing, Online Surveys, Web Crawling and Indexing (no past questions).
• Natural Language Processing Techniques for Micro-Text Analysis (no past questions).