Review of: Romain Hilgert (ed.): Michel Rodange, Renert: De Fuuss am Frack an a Maansgréisst. Komplett Editioun mat historeschen a politeschen Explicatioune, Lëtzebuerg: Éditions Guy Binsfeld, 2020

in Hemecht: Zeitschrift für Luxemburger Geschichte (2021), 73(3), 377-378

An Annotation Framework for Luxembourgish Sentiment Analysis

in Besacier, Laurent; Sakti, Sakriani; Soria, Claudia (Eds.) et al Proceedings of the LREC 2020 1st Joint SLTU and CCURL Workshop (SLTU-CCURL 2020) (2020, May)

The aim of this paper is to present a framework developed for crowdsourcing sentiment annotation for the low-resource language Luxembourgish. Our tool is easily accessible through a web interface and ... [more ▼]

The aim of this paper is to present a framework developed for crowdsourcing sentiment annotation for the low-resource language Luxembourgish. Our tool is easily accessible through a web interface and facilitates sentence-level annotation of several annotators in parallel. In the heart of our framework is an XML database, which serves as central part linking several components. The corpus in the database consists of news articles and user comments. One of the components is LuNa, a tool for linguistic preprocessing of the data set. It tokenizes the text, splits it into sentences and assigns POS-tags to the tokens. After that, the preprocessed text is stored in XML format into the database. The Sentiment Annotation Tool, which is a browser-based tool, then enables the annotation of split sentences from the database. The Sentiment Engine, a separate module, is trained with this material in order to annotate the whole data set and analyze the sentiment of the comments over time and in relationship to the news articles. The gained knowledge can again be used to improve the sentiment classification on the one hand and on the other hand to understand the sentiment phenomenon from the linguistic point of view. [less ▲]

A Temporal Warehouse for Modern Luxembourgish Text Collections

Presentation (2019)

Mind and Language. AI in an Example of Similar Patterns of Luxembourgish Language

in International Conference on Artificial Intelligence Humanities, Book of Abstracts (2018, August 16)

Compiling Tools and Resources for Studying of Luxemburgish Language and beyond

Scientific Conference (2016, June)

Das luxemburgischsprachige Oeuvre von Michel Rodange (1827-1876). Editionsphilologische und korpuslinguistische Analyse

Book published by Universität Trier (2015)

The current work focuses on theoretical and practical aspects of the analysis and publication of important literary texts using the methods of digital humanities, as well as corpus and computational ... [more ▼]

The current work focuses on theoretical and practical aspects of the analysis and publication of important literary texts using the methods of digital humanities, as well as corpus and computational linguistics. The oeuvre of Michel Rodange in the Luxembourgish language provides the basic material for this study. This includes the works ‘Renert oder de Fuuss am Frack an a Maansgréisst‘ - ca. 40.000 Tokens, ‘Dem Léiweckerche säi Lidd‘ – ca. 5.000 Tokens, ‘Dem Grof Sigfrid seng Goldkuemer‘ – ca. 15.000 Tokens and two poems – ca. 500 Tokens. On the empirical level the work involves compiling a corpus with text critical and linguistic annotations and its presentation as a web portal. A very interesting interdependency arises at this point between theory and practice. It is possible to use the created annotations in order to investigate the oeuvre from a philological point of view on the one hand; on the other hand one can use this knowledge in the development of tools, which can create similar or new annotations and improve their correctness. The text critical annotations consist of reading variants, corrections and a word glossary, whereas the linguistic annotations are of an orthographical, morphological (also including word classes and lemmata) and phraseological nature. The annotations are codes in XML. The first step in working with one’s own corpus is its digitalization. In the case of manuscripts this occurred by means of manual transcription, but the prints were digitized with the help of OCR-software. At this stage the texts were already well organized and enriched with the initial annotations and metadata. At first sight less important but still useful information such as page numbers, line breaks, chapters etc. were marked. However in compiling a corpus from an important literary work the conservation of text critical annotations is of special significance. In order to create such annotations the theoretical principles of textual criticism first needed to be discussed. Then the most important phenomena, such as text fragments, corrections, deletions etc., were documented. For poorly readable areas new interpretations were proposed and those from other editions were compared and discussed. The Text Encoding Initiative (TEI) offers a number of XML-elements to organize such annotations. It was possible to limit much manual work by using tools such as TUSTEP or oXygen as well as scripting languages, e.g. Perl. These provide a powerful technique of “regular expressions”, which can automate the search and replacement work at an abstract level. The next and very useful step is tokenization. At this level text critical annotations almost overlap with linguistic annotations. A fuzzy border was defined for tokens and they were marked with their own XML-element. It is essential for further investigation to consider and study the language of the author. For this purpose special attention was paid to poetic styles in the Luxembourgish language of the 19th century, literary genres, and the orthography and spelling of the author. On the empirical level, the analysis of the corpus from a linguistic point of view with the help of corpus linguistics methods, e.g. extraction of multi word units, forms the scientific basis for later digital presentation of the oeuvre. For this purpose a number of tools for analyzing the oeuvre quantitatively as well as qualitatively were designed and developed. The output of these tools is discussed in the current work. Here theories from classical linguistics e.g. morphology, phraseology and corpus linguistics e.g. POS-Taggers, concordances and collocational analysis, were discussed. All in all the following programs were implemented and described: Tokenizer, Frequency List, POS-Trainer, POS-Tagger, Lemmatizer and further programs for word formation and phraseological analysis. The program for morphological analysis and Lemmatizer are basically rule-based, whereas most of the other programs work by means of statistics. Hidden Markov Models, which derive from probabilistic theory, underlie the assignment of part of speech tags to words. For phraseological analysis many current statistical models such as z-score, t-score, mutual information, chi-square-test and fisher’s exact test were implemented and tested. Chapter 3.4 is dedicated to the output of the program for morphological analysis and discusses word formation in the oeuvre of Michel Rodange. Chapter 3.6 focuses on the interpretation of the output for collocational- phraseological analysis. As a result of this investigation it became apparent that many automatically identified phraseological units in Michel Rodange’s oeuvre are not only a part of Luxembourgish language and culture, but also found in other European cultures. [less ▲]

Die Erstellung sprach- und literaturwissenschaftlicher Tools für das Historisch-kritische Michel-Rodange-Portal

in Gilles, Peter; Wagner, Mélanie (Eds.) Linguistische und soziolinguistische Bausteine der Luxemburgistik (2011)