Use cases · TALL — worked examples from the paper

Case · 01

Thematic analysis on BBC News.

A collection of 386 short news articles from the BBC News corpus, retrieved from the BBC website (Greene & Cunningham, 2005). After loading, TALL applies standard pre‑processing — tokenisation and lemmatisation via the EN GUM model — retaining adjectives, nouns, proper nouns, and verbs.

English GUM model 386 docs Louvain

Community detection on BBC News corpus — **Figure · Network** Five dense sub‑graphs from community detection on the entertainment sub‑corpus: Awards, Music, Industry, Media, General News.

Step 1 · Import

Built‑in BBC News dataset loaded with TALL AI context automatically populated.

Step 2 · Tokenise & PoS

EN GUM model selected. Adjectives, nouns, proper nouns, verbs retained.

Step 3 · Multi‑word

RAKE extracts salient collocations; manual curation preserves domain phrases.

Step 4 · Overview

Descriptive statistics, lexical richness, and TF‑IDF for discriminating terms.

Step 5 · Words in Context

Concordance view surfaces co‑text environments for salient terms.

Step 6 · Topic modeling

LDA recovers thematic structure across the full news collection.

TALL AI · Interpretation

"The entertainment news focuses heavily on film awards and the movie industry. Music‑related news forms a significant portion of the content — organised around performance, industry and sales."

Polarity analysis on US Airline Tweets — **Figure · Polarity** Top word distributions by polarity category. Negative tweets cluster around service disruptions (*delay*, *cancel*, *miss*) while positive tweets express gratitude (*thank*, *appreciate*).

Case · 02

Sentiment on
airline tweets.

14,640 tweets directed at six major US airlines during February 2015 (Figure Eight / Kaggle). TALL detects platform‑specific entities — hashtags, mentions, emojis — and runs polarity detection via the Hu‑Liu opinion lexicon.

127

unique emojis

1.9k

hashtags

930

mentions

28%

negative sentiment

Step 1 · Import

Built‑in corpus loaded; TALL AI receives pre‑populated dataset context.

Step 2 · EWT tokenisation

EN EWT model handles social‑media text. Hashtags, mentions, emojis retained as entities.

Step 3 · Special entities

Automatic tagging of hashtags, mentions, emojis with frequency ranking.

Step 4 · Words in Context

Concordance for @AmericanAirlines reveals customer‑service discourse patterns.

Step 5 · Polarity

Hu‑Liu lexicon · donut chart with five levels from Very Negative to Very Positive.

Step 6 · TALL AI

Automated key takeaways & managerial implications.

TALL AI · Key takeaways

"The dominance of delay and cancel in negative tweets suggests airlines should focus on improving their processes for handling flight disruptions. Both positive and negative tweets highlight the importance of customer service — the presence of gratitude‑related terms in positive sentiment indicates that effective customer service responses can mitigate dissatisfaction and generate positive outcomes."

Case · 03

Science mapping from abstracts.

TALL natively ingests biblioshiny files from the bibliometrix package. Scientific abstracts flow into the same pre‑processing pipeline — multilingual tokenisation, lemmatisation, co‑word analysis — and can be layered with citation‑based metrics for hybrid semantic/bibliometric analysis.

A natural fit for systematic literature reviews, conceptual‑structure analyses, and policy‑oriented research synthesis.

biblioshiny import co‑word analysis citation filters

Figure · Bibliometrix Biblioshiny ingestion, corpus filtering, co‑word analysis, and TALL AI interpretation of the emerging conceptual structure.

Figure · Wikipedia An encyclopedic corpus processed through import, overview, multi‑word extraction, co‑word analysis, and topic modeling — ending with TALL AI's topic interpretation.

Case · 04

Topic discovery on Wikipedia.

An encyclopedic corpus pushes TALL through the full analytical narrative: from basic frequency and TF‑IDF to multi‑word extraction, co‑word networks, and finally LDA topic modeling with TALL AI producing concise narrative labels for each topic.

Ideal for digital humanities, educational‑content audits, and cross‑linguistic cultural analysis given TALL's 56‑language coverage.

topic modeling co‑occurrence multi‑word AI topic labels

§ · Where TALL is used

Across disciplines where text is evidence.

Social & political sciences

Large‑scale thematic and sentiment analyses of institutional reports, parliamentary proceedings, and online debates.

Digital humanities

Multilingual architecture for cross‑linguistic corpus studies and computational cultural analysis across 56 languages.

Bibliometrics

Integration with bibliometrix combines semantic and citation‑based analyses for science mapping and literature reviews.

Healthcare research

Analyses of patient feedback, clinical records and satisfaction surveys (Valentino et al., 2025).

Education

Currently used in postgraduate and doctoral programs at several Italian universities for teaching data‑driven text methods.

Industry & market research

Sentiment tracking and theme discovery across customer feedback, reviews, and social media at scale.

Four corpora.
One workflow.

BBC News

US Airline Tweets

bibliometrix

Wikipedia

Thematic analysis on BBC News.

Sentiment on
airline tweets.

Science mapping from abstracts.

Topic discovery on Wikipedia.

Across disciplines where text is evidence.

Your corpus is next.

Four corpora. One workflow.

BBC News

US Airline Tweets

bibliometrix

Wikipedia

Thematic analysis on BBC News.

Sentiment onairline tweets.

Science mapping from abstracts.

Topic discovery on Wikipedia.

Across disciplines where text is evidence.

Your corpus is next.

Four corpora.
One workflow.

Sentiment on
airline tweets.