NEWS

---

stylo 0.7.5 (2024-04-03)

• size.penalize() renamed to samplesize.penalize(), for CRAN
  • bug in size.penalize() fixed
  • improved performance of dist.minmax()
  • oppose() update, to allow having just one text per set
  • a solid clean-up in a few functions
  • several minor improvements here and there

stylo 0.7.4 (2020-12-06)

• option for shading in rolling.classify()
  • performance.measures() greatly improved
  • supervised classifiers updated, to be compliant with cross-validation
  • SVM output fixed
  • bugs in rolling.classify() fixed
  • bugs in load.corpus() causing codepage mismatches fixed
  • general code cleanup

stylo 0.7.3 (2020-08-11)

• perfom.svm() improved to work with R >4.0.0
  • oppose() not restricted anymore to have at least 2 texts per set
  • better color management in rolling.classify()
  • CPU performance improvements

stylo 0.7.2 (2020-04-20)

• fixes required by CRAN to meet R >3.6.3 requirements
  • CPU performance improvements
  • improvements in performance.measures()
  • confusion matrices fixed
  • oppose() update, to allow having just one text per set

stylo 0.7.1 (2019-11-04)

• improvements in crossv(): confusion matrix fully operational
  • new funcion performance.measures(), providing recall, precision, f1, etc.
  • performance measures made available via classify()
  • new function size.penalize() to assess minimal sample size
  • extension to the generic plot() function, to plot size.penalize() results

stylo 0.7.0

• Unicode (UTF-8) made the default encoding, also for Windows

stylo 0.6.9 (2019-01-21)

• check.encoding() and change.encoding() introduced
  • GUI allows for changing the working directory with one click
  • metadata handling through a dedicated variable
  • {Steffen Pielström joins!}

stylo 0.6.8 (2018-10-08)

• support for JCK (Japanese-Chinese-Korean) significantly improved
  • a fix for exporting networks to Gephi ver. 0.9.2
  • support for rmarkdown: stylo(), classify(), oppose()

stylo 0.6.7 (2018-05-26)

• supports the following taggers: TaKIPI (for Polish), Alpino (Dutch)
  • the Imposters method reimplemented, via the new function imposters()
  • fine tuning the parameters of the Imposters method via imposters.optimize()

stylo 0.6.6 (2018-04-13)

• Cosine Delta implemented and aviable via GUI
  • Min-Max distance implemented
  • Entropy distance implemented

stylo 0.6.5 (2017-11-05)

• support for interactive network visualisations via stylo.network()
  • corrected Spanish pronouns
  • fixes in documentation
  • countless minor fixes

stylo 0.6.4 (2016-10-23)

• citation hint updated; to see the changes type: citation("stylo")
  • the impostors method almost implemented, see help(perform.impostors)
  • confusion table for supervised classification via classify()
  • a separate funtion for cross-validation, see help(crossv)
  • a significant change in SVM wrapper: the procedure automatically gets rid of the variables with all 0s in the training set
  • the file inst/CITATION updated to meet recent CRAN requirements
  • man files for perform.delta, perform.svm etc. updated: new executable examples added, so that one can perform a supervised test without any corpus
  • perform.knn(), perform.svm() etc. improved, in order to handle custom vectors of classes provided by a user
  • an improved output of the oppose() function

stylo 0.6.3 (2015-12-20)

• significant performance improvement in make.table.of.frequencies()
  • PCA values (rotation, explained variance, etc.) saved in final results

stylo 0.6.2 (2015-11-11)

• the package 'stringi' involved to optimize n-gram computing
  • three datasets added to the package
    • data(novels), a collection of 9 novels by the Bronte sisters and Jane Austen (full text)
    • data(galbraith), a table of frequencies of 26 novels by 5 authors, including Galbraith's "Cacoo's Calling"
    • data(lee), a table of frequencies of 28 American novels by 8 authors, including the new novel by Harper Lee
  • new version of make.table.of.frequencies(), which speeds up the tasks radically
  • delete.markup(), delete.stop.words(), make.samples(), make.frequency.list(), txt.to.features(), txt.to.words.ext() remodelled so that can be applied to single texts and/or to corpora
  • countless improvements in most of the functions

stylo 0.6.1 (2015-09-28)

• UTF-8 issue in txt.to.words.ext() fixed, according to the CRAN's request

stylo 0.6.0 (2015-08-18)

• support for Georgian
  • plot size in rolling.classify() improved
  • distance measure engine thoroughly restructured
  • custom distance measures allowed
  • cosine distance introduced
  • new functions: dist.cosine(), dist.delta(), dist.argamon(), dist.eder(), dist.simple()
  • extracting POS tags via the function parse.pos.tags()

stylo 0.5.9-3

• support for Coptic
  • customizable graphs size in rolling.classify()
  • custom graph filename
  • integration with CLARIN-PL stylometric infrastructure

stylo 0.5.9 (2015-01-30)

• non-ASCII chars in the source code neutralized (required by CRAN)
  • random sampling substantially improved

stylo 0.5.8-3

• bug fixes: options for assign.plot.colors()

stylo 0.5.8-2 (2014-10-19)

• bug fixes: 'start.at' parameter in stylo()

stylo 0.5.8-1 (2014-10-01)

• bug fixes (mostly: colors on dendrograms)

stylo 0.5.8 (2014-09-03)

• new sequential methods available: rolling SVM, rolling NSC, and rolling Delta
  • bug in load.corpus.and.parse() fixed
  • bug in rolling.delta() fixed
  • network related bug in stylo() neutralized
  • classification procedures as separate functions: perform.delta(), perform.svm(), perform.knn(), perform.naivebayes(), perform.nsc()
  • classification output enhanced
  • doc files for new functions added

stylo 0.5.7 (2014-08-13)

• culling implemented as a separate function
  • custom stop words deletion: delete.stop.words()
  • a thoroughly re-written oppose() to use the same tokenizing, corpus loading, sampling etc. functions as stylo() and classify()
  • zeta.chisquare(), zeta.craig(), and zeta.eder() derrived as separate functions
  • gui.oppose() derrived as a separate function
  • distinctive words visualization in oppose() improved
  • draw.polygons derrived as a separate function (hidden to the end user, though)
  • cross-validation in classify() improved
  • fixed bug in cross-validation for naivebayes
  • a very unpleasant bug in oppose() fixed: the preferred and avoided words were calculated using the I set only
  • help files significatnly improved

stylo 0.5.6 (2014-05-21)

• support for Unicode on Windows
  • support for a few non Latin scripts
  • experimental support for CJK (Chinese-Japanese-Korean)
  • the function txt.to.words() remodelled
  • loading corpus files improved
  • printing variables on screen improved
  • better class inheritance
  • an issue with hclust and "ward", "ward.D" fixed
  • man files extended and updated

stylo 0.5.5 (2014-04-03)

• cross-validation in classify()
  • lots of bugs fixed

stylo 0.5.4 (2014-02-26)

• tSNE implemented
  • preserve.case option
  • more flexible function for splitting input text

stylo 0.5.3 (2014-01-02)

• custom regular expressions to tokenize input texts
  • support for external corpora or frequencies
  • support for external set of features (e.g. frequent words)
  • class "stylo.results" for formatting final results
  • class "stylo.corpus" for formatting loaded corpora
  • class "stylo.data" for formatting tables and vectors
  • PCA coordinates piped to final results
  • optional choice between relative/raw frequencies
  • xml support improved (bug fixed)
  • codepage bug in oppose() fixed

stylo 0.5.2 (2013-09-08)

• CRAN-related issue with .Rbuildignore fixed
  • network analysis support significantly improved
  • improvements in man pages

stylo 0.5.1 (2013-08-08)

• bug fixes, minor improvements
  • different options for k-NN and SVM
  • submitted to CRAN for the first time (!)

stylo 0.5.0-58

• batch mode improved
  • several clustering algorithms available

stylo 0.5.0-50

• man pages revised and improved

stylo 0.5.0-49

• poster presentation at DH2013 (Lincoln, NE)
  • minor improvements

stylo 0.5.0-48

• namespace issues solved
  • documentation corrected (typos)

stylo 0.5.0-45

• arguments can be passed from command-line
  • man pages cleaned and extended
  • global variables abandoned
  • innumerable minor improvements

stylo 0.5.0-43

• thousands of changes and improvements
  • documentation improved and augmented
  • stylo R package (un)officially released

stylo 0.5.0-30

• changes in names of some functions
  • code cleaning, improvements, improvements, ...

stylo 0.5.0-23

• first prototype of an R package

stylo 0.5.0-1

• first attempt to port the stylo script into R package

stylo 0.4.9-2

• code OS-independent
  • minor cleaning

stylo 0.4.9-1

• experimental support for network analysis (output to Gephi)
  • bugs fixed

stylo 0.4.9

• added option to dump samples for closer post-analysis inspection

stylo 0.4.8

• customizable plot area, font size, etc.
  • thoroughly rewritten code for margins assignment
  • scatterplots represented either by points, or by labels, or by both (customizable label offset)
  • saving the words (features) actually used
  • saving the table of actually used frequencies

stylo 0.4.7

• new output/input extensions: optional custom list of files to be analyzed, saving distance table(s) to external files
  • support for TXM Textometrie Project
  • color cluster analysis graphs (at last!)

stylo 0.4.6

• code revised, cleaned, bugs fixed

stylo 0.4.5-4

• added 2 new PCA visualization flavors

stylo 0.4.5-3

• new GUI written

stylo 0.4.5-2

• added functionality for normal sampling

stylo 0.4.5-1

• support for Dutch added
  • {Mike Kestemont joins!}

stylo 0.4.5

• option for choosing corpus files
  • code cleaned; bugs fixed

stylo 0.4.4

• the core code rewritten
  • I/II set division abandoned
  • GUI remodeled
  • GUI tooltips added
  • different input formats supported (xml etc.)
  • config options loaded from external file
  • the code forked into (1) the Stylo script, supporting explanatory analyses (MDS, Cons. Trees, ...), (2) the Classify script for machine-learning methods (Delta, SVM, NSC, Bayes)

stylo 0.4.3

• feature selection (word and character n-grams)

stylo 0.4.2

• three ways of splitting words in English
  • bugs fixed
  • GUI code rearranged and simplified

stylo 0.4.1

• better output
  • better text files uploading
  • new options for culling and ranking of candidates

stylo 0.4.0

• the official world-premiere, at DH2011 (Stanford, CA)

stylo 0.3.9

• the code simplified; minor cleaning

• uploading wordlist from external source
  • thousands of improvements
  • the code simplified

stylo 0.3.8

• skip top frequency words option added

stylo 0.3.7

• better graphs
  • attempt at better graph layout

stylo 0.3.6

• more graphic options
  • dozens of improvements

stylo 0.3.5

• module for color graphs
  • module for PCA

stylo 0.3.4

• module for uploading corpus files improved

stylo 0.3.3

• the core code simplified and improved (faster!)

stylo 0.3.2

• reordered GUI
  • minor cleaning

stylo 0.3.1

• the z-scores module improved

stylo 0.3.0

• better counter of "good guesses"
  • option for randomly generated samples
  • minor improvements

stylo 0.2.99

• platform-independent outputfile saving

stylo 0.2.98

• GUI thoroughly integrated with initial variables

stylo 0.2.10

• corrected MFW display in graph
  • more analysis description in outputfile

stylo 0.2.9

• auto graphs for MSD and CA

stylo 0.2.8

• remodeled GUI

• GUI: radiobuttons, checkbuttons

stylo 0.2.7

• language-determined pronoun selection

stylo 0.2.6

• dialog box (GUI)
  • {Jan Rybicki joins!}

stylo 0.2.5

• module for different distance measures
  • thousands of improvements (I/O, interface, etc.)

stylo 0.2.2

• numerous little improvements
  • deleting pronouns

stylo 0.2.1

• module for culling
  • module for bootstrapping

stylo 0.2.0

• module for uploading plain text files

stylo 0.1.9

• innumerable improvements
  • the code simplified
  • {this version was completed on a train from Leipzig to Krakow (a looong trip...), after a very successful R course taught by Stefen Gries at ESU "C&T", Leipzig, Germany (26-31/08/2009)}

stylo 0.1.4

• loop for different MFW settings

stylo 0.0.1

• some bash and awk scripts translated into R