Objectives
The tutorial aims to introduce the architecture, use and enhancement of the LiLa Knowledge Base of interlinked linguistic resources for Latin, developed in the context of the LiLa: Linking Latin project. In particular, the tutorial will present how the Linked-Data model adopted by LiLa is used to connect distributed lexical and textual resources, to ensure their interoperability. We show how, via lemmatisation, texts become part of this network of resources. We provide participants with a theoretical introduction to the architecture of LiLa, as well as hands-on support in their interaction with the LiLa Knowledge Base.
The proposed tutorial falls within the field of Linguistic Linked Open Data (LLOD: Cimiano et al. 2020). While we focus on Latin, the methods discussed are language independent and thus have a much wider application, proving useful for similar initiatives on other languages.
Starting from the experience of LiLa, we introduce the audience to some of the most relevant topics in current digital textual studies and linguistic resources, including:
-
Language (Meta)data. Tools and techniques of linguistic annotation. Specific concepts and technologies discussed: tokenisation, part-of-speech tagging, lemmatisation, and application of a Natural Language Processing (NLP) pipeline (UDPipe) with specific models trained on a large Latin corpus.
-
Knowledge Graphs. Tools and techniques to interoperably link linguistically-annotated datasets. Specific concepts and technologies discussed: interoperability standards, ontologies, RDF, SPARQL, Linguistic Linked Data.
-
Use Cases in Language, Data and Knowledge. Possible uses for philology and linguistics, namely searches combining etymology, polarity, frequency, (graphical) variation, meaning, and more.
Detailed description
We propose a 1-day tutorial, divided in two sections.
Morning (conceptual foundations)
In this session the LiLa project will serve as a use-case to illustrate a typical LLOD workflow, that is:
- inguistic annotation/processing (Gleim et al. 2019; Sprugnoli et al. 2020)
- data modelling (McCrae et al. 2017)
- linking (Declerck et al. 2012).
Afternoon (practical activity)
Participants will be divided into groups, supervised by one or more presenters/assistants. We provide them with a selection of Latin raw texts from various types and genres, and guide them through the stages leading up to the connection of texts to the LiLa Knowledge Base. Specifically, we focus on how to perform automatic tokenisation, part-of-speech tagging and lemmatisation, run a custom tool to automatically transform lemmatised output into the RDF, link it to the LiLa Knowledge Base and, finally, query the Knowledge Base with SPARQL (DuCharme 2013).
In closing, all groups will come together to present their results and draw conclusions.
Data and tools necessary to participate in the tutorial will be provided ahead of the event.
Tentative schedule
09:00-09:45 – General introduction (Passarotti)
09:45-10:30 – Data model (Mambrini)
10:30-10:45 – Coffee Break
10:45-11:30 – Processing (Cecchini) and currently linked resources (Litta, Sprugnoli)
11:30-12:00 – Questions and group formation
12:00-13:30 – Lunch
13:30-16:00 – Hands-on work
16:00-17:00 – Group presentations and conclusions
Instructors
Presenters and assistants (all affiliated with Università Cattolica del Sacro Cuore):
- Marco Passarotti (
[marco.passarotti@unicatt.it](mailto:marco.passarotti@unicatt.it)). Associate Professor of
Computational Linguistics and Director of the CIRCSE Research Centre. Since
2006 he has headed the Index Thomisticus Treebank project. He is the
Principal Investigator of the LiLa: Linking Latin ERC Consolidator Grant
(2018–2023).
- Flavio Massimiliano Cecchini (
[flavio.cecchini@unicatt.it](mailto:flavio.cecchini@unicatt.it)) graduated in Mathematics from the
Università degli Studi in Milan and then focused his interest towards NLP
and Computational Linguistics; he obtained a PhD in Computer Science with a
thesis about Word Sense Discrimination. Currently, Flavio works as a
postdoctoral researcher on the LiLa: Linking Latin project.
- Francesco Mambrini (
[francesco.mambrini@unicatt.it](mailto:francesco.mambrini@unicatt.it)). PhD in Classical Philology
(University of Trento / EHESS). Researcher for the LiLa: Linking Latin
project. Previously, he worked at the Deutsches Archäologisches Institut,
Berlin, and at the University of Leipzig. His research interests focus on
the language and style of Ancient Greek and Latin literary texts.
Assistants:
- Eleonora Litta (
[EleonoraMaria.Litta@unicatt.it](mailto:EleonoraMaria.Litta@unicatt.it)). Eleonora has worked on the
development of digital language resources, text analysis, corpus and
computational linguistics. She was awarded a Marie Skłodowska-Curie Research
fellowship to develop WFL, a word formation lexicon for Latin, at the
Università Cattolica del Sacro Cuore. She is currently working as part of
the LiLa project on a new Latin valency lexicon and WordNet.
- Giovanni Moretti (
[giovanni.moretti@unicatt.it](mailto:giovanni.moretti@unicatt.it)). Software developer. Since 2007
Giovanni has been working in many Human Language Technology projects
developing scripts and web interfaces for data acquisition, annotation,
analysis and data visualisation. Currently, he is working in the LiLa:
Liking Latin project strengthening his expertise in Linked Data technology.
- Matteo Pellegrini (
[matteo.pellegrini@unicatt.it](mailto:matteo.pellegrini@unicatt.it)). Postdoctoral researcher working on
the LiLa: Linking Latin project. His research is mainly focused on
inflectional morphology, with a specific interest in automatic processing of
Latin morphology.
- Paolo Ruffolo (
[paolo.ruffolo@posteo.eu](mailto:paolo.ruffolo@posteo.eu)). Computer science engineer. Worked as
software developer and tester, database designer and administrator, and data
analyst both for research institutions and private companies. Currently
working on the LiLa: Liking Latin project.
- Rachele Sprugnoli (
[rachele.sprugnoli@unicatt.it](mailto:rachele.sprugnoli@unicatt.it)). Postdoctoral researcher working on
the LiLa: Linking Latin project. Her research is mainly focused on text
annotation, evaluation of NLP tools and on how computational methods and
text analysis technologies can be applied to the treatment of cultural
content.
Target audience and expected attendance
This tutorial is intended for those who wish to explore solutions to publish texts using LOD. Although we focus on our experience with Latin, we welcome any participant interested in the theme of textual resources and LOD; prior knowledge of Latin and/or LOD is not required.
We expect around 20-30 participants from different backgrounds: computational linguists, theoretical linguists, classicists, philologists.
Proposed budget
As the tutorial will be run by the members of the LiLa team, no costs to pay for the instructors is foreseen.
Neither publication of proceedings nor invitation of keynote speakers is planned. The website of the tutorial (if required to be external to that of EADH 2021) will be hosted on that of LiLa, free of charge.
Should the tutorial take place in loco, the costs for the coffee breaks will be covered by the tutorial fees.
Requirements and format
Should the tutorial take place in loco, we would require a projector and a whiteboard with markers.
Should it take place online, we would adopt Zoom as our preferred platform for its “Breakout Rooms” functionality.
Bibliography
Cimiano, Philipp / Chiarcos,
Christian / McCrae, John P. / Gracia, Jorge (2020): Linguistic Linked Data:
Representation, Generation and Applications. Cham: Springer.
Declerck, Tierry / Lendvai,
Piroska / Mörth, Karlheinz / Budin, Gerhard / Váradi, Tamás (2012):
"Towards linked language data for digital humanities", in: Chiarcos,
Christian / Nordhoff, Sebastian / Hellmann, Sebastian (eds.): Linked Data in Linguistics. Representing and
Connecting Language Data and Language Metadata. Berlin / Heidelberg:
Springer 109–116 DOI: <https://doi.org/10.1007/978-3-642-28249-2_11>.
DuCharme, Bob (2013): Learning
Sparql. Querying and Updating with Sparql 1.1. Sebastopol, CA:
O’Reilly.
Gleim, Rüdiger / Eger, Steffen /
Mehler, Alexander / Uslu,
Tolga / Hemati, Wahed / Lücking, Andy / Henlein, Alexander / Kahlsdorf, Sven / Hoenen, Armin
(2019): "A practitioner’s view: a survey and comparison of lemmatization and
morphological tagging in German and Latin", in: Journal of
Language Modelling 7, 1: 152 DOI:
<https://doi.org/10.15398/jlm.v7i1.205>.
McCrae, John P. / Bosque-Gil,
Julia / Gracia, Jorge / Buitelaar, Paul / Cimiano, Philipp (2017):
"The Ontolex-Lemon model: development and applications", in: Kosem, Iztok /
Tiberius, Carole / Jakubíček, Miloš / Kallas, Jelena / Krek, Simon / Baisa,
Vít (eds.): Electronic lexicography in the 21st century.
Proceedings of eLex 2017 conference. Brno: Lexical Computing 19–21.
Sprugnoli, Rachele / Passarotti,
Marco / Cecchini, Flavio M. / Pellegrini, Matteo (2020): "Overview of the EvaLatin 2020
Evaluation Campaign", in: Sprugnoli, Rachele / Passarotti, Marco (eds.): Proceedings of LT4HALA 2020. 1st Workshop on Language
Technologies for Historical and Ancient Languages. Marseille: ELRA 105–110.