Changelog:
New Analyses
SVO Triplets (Subject-Verb-Object)
- New analysis for extracting subject-verb-object triplets from the
dependency tree
- C++ backend (extract_svo.cpp) for high performance
- Dedicated UI under Documents with Run/Export/Report, results table,
Info &
References, TALL AI
- Exported R function: extract_svo_triplets()
Syntactic Complexity
- New document-level syntactic complexity analysis based on dependency
parsing
- C++ backend (syntactic_complexity.cpp) for computing syntactic
metrics
- Dedicated UI under Documents with Run/Export/Report, results table,
Info &
References, TALL AI
- Exported R function: compute_syntactic_complexity()
Emotion Analysis (NRC EmoLex)
- New emotion analysis based on the NRC Word-Emotion Association
Lexicon
- Detection of 8 emotions: anger, anticipation, disgust, fear, joy,
sadness,
surprise, trust
- Core functions: emotionAnalysis(), emotionBarChart(),
emotionWordPlot(),
emotionHeatmap()
- loadEmotionLexicon() for loading NRC data (English: mapped via
Italian
lexicon’s English Word column)
- Dedicated UI with 6 tabs: Emotion Distribution, Top Words by
Emotion, Document Heatmap, Table, Info & References, TALL AI
- PNG export and Excel report integrated
- Gemini AI integration across all 5 switch points
Noun Phrase Extraction
- Noun phrase extraction via dependency tree
- C++ backend (extract_np.cpp)
- Exported R function: extract_noun_phrases()
Topic Modeling (major enhancement)
- CTM (Correlated Topic Model) and STM (Structural Topic Model) added
as new methods
- STM prevalence covariates with effect plots and regression
coefficients
- Model diagnostics tab: coherence, exclusivity, log-likelihood
- Multi-metric comparison plot and consensus K recommendation
- Estimated K automatically transferred to the model estimation
panel
Image Export System (complete rewrite)
- Rewrite of plot2png() with DPI-aware rendering (biblioshiny
approach)
- Graph export settings: export/report DPI, height, aspect ratio
- Persistent settings, temporary file cleanup
- JavaScript canvas capture for visNetwork
Homepage & References
- Complete homepage redesign in biblioshiny style
- Updated citation to the SoftwareX paper (Aria et al., 2026)
- Added open-access paper and supplementary material links
Settings & UI
- Added icons to all tab panels across the app
- Added colored boxes next to hex color codes in Thematic Map
tables
- Fixed Reinert/CA dendrogram “Words in Context” error (uninitialized
variable)
- Fixed Import tab: table disappearing after Remove, “Back to original
text”, dfTag sync
- Fixed Topic Modeling estimation error without options and
conditional arrow buttons
- Fixed report items overlapping with box title
- Fixed Similarity page margins and Embedding distribution
spacing
- Fixed Supervised Classification header formatting
- Moved Info & References to last tab in Thematic Map and
Similarity
- Removed “in TALL” from all Info & References section
titles
- Updated KWIC Info styling and added Dependency reference in Word
Network info
- Added morphological features stats to Overview report export
- Added PoS selection guidance in Overview Frequency options
- Updated Custom PoS List template and docs to clarify any custom tag
is allowed
- Improved working folder selector with visual feedback
- Reorganized settings layout (2x2 grid)
- Time variable aggregation for Date types
(day/week/month/quarter/year)
Dependency Parsing Features
- Dependency-based word network as an alternative to
co-occurrence
- Configurable dependency relation filters
(all/noun_mod/SVO/custom)
Architectural Refactoring
- Split tallFunctions.R into 10 thematic modules: tallSentiment.R,
tallNetwork.R, tallTopicModel.R, tallEmbeddings.R,
tallOverview.R, tallUtils.R, tallVisualization.R, tallTextIO.R,
tallNLP.R,
tallReport.R, tallLanguages.R
- Extracted static CSS into www/tall-static.css
- Added www/tall-handlers.js for JavaScript handlers
- Modernized icons (migrated to FontAwesome 6)
- Cleaned up NAMESPACE (removed unnecessary exports)
- Added test suite: test-calculate_ngram_is.R,
test-process_multiwords.R,
test-reinert.R, test-txt_recode_fast.R
- Added stm dependency to DESCRIPTION
Code Quality
- Propagated random seed to all analyses (topic models, networks,
wordclouds, sampling)
- Default community detection switched from walktrap to louvain
- Removed deprecated Grako module
- Renamed “Co-Word Analysis” to “Word Network Analysis”
- Updated all Info & References sections
- Removed legacy code from tallEmbeddings.R and words.R (~1,160 lines
removed)
Bug Fixes
- Fixed KWIC on filtered collections
- Fixed overview with filtered corpus
- Fixed group handling in filters_groups.R
- Minor fixes in import, edit, settings, collocation