This paper presents DEC Facile, a tool specifically designed for the creation of computational Explanatory Combinatorial Dictionaries (ECDs). Powered by LexO-server and Lemon modules, DEC Facile integrates Linked Data principles for efficient lexical resource management. The tool's userfriendly interface supports tasks like dictionary search, sense editing, and lexical function linking, making it a valuable resource for lexicography, terminography, and computational linguistics.
4th International Conference on “Multilingual digital terminology today. Design, representation formats and management systems” (MDTT) 2025, June 19-20, 2025, Thessaloniki, Greece. ∗ Corresponding author.
mailto:abdelouahab.masbah@ilc.cnr.it (A. Masbah); andrea.bellandi@ilc.cnr.it (A. Bellandi); silvia.piccini@ilc.cnr.it (S. Piccini) 0009-0009-9692-0533 (A. Masbah); 0000-0002-1900-5616 (A. Bellandi); 0000-0002-2584-0191 (S. Piccini) © 2025 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
Sens-Texte (OLST) at the University of Montréal, several ECDs have been developed, including DiCoEnviro (environmental terminology), DiCoInfo (computer science terminology), and DiCoLexitrans (legal terminology).
Creating ECDs is a complex and time-consuming process. Previous attempts to develop editors for ECDs have been made [e. g., 14, 15], with the primary goal of assisting users in this challenging task. DEC Facile continues this effort while introducing a significant innovation. Specifically, it leverages semantic web technologies [16] and the Linked Open Data (LOD) paradigm [17] to facilitate the creation of ECDs that are natively structured for interoperability with other lexical resources and language technologies.
By adhering to the FAIR (Findable, Accessible, Interoperable, and Reusable) principles [18], DEC Facile ensures that the lexical data it produces can be integrated into existing linguistic infrastructures. For instance, resources generated with DEC Facile can be linked to external lexical databases such as BabelNet2, WordNet3, and Wikidata4, thanks to its compatibility with the OntoLex-Lemon framework. This enables cross-resource querying and enhances the usability of ECDs in multilingual applications. Additionally, as we will see in detail in section 3, DEC Facile provides structured exports in RDF and JSON-LD formats, ensuring that its outputs can be used in natural language processing (NLP) pipelines, machine translation engines, and computational terminology management systems.
DEC Facile has been developed as part of a nationally funded research project titled “Old Words for a New World. Translating Christianity to Baltic Pagans”, which includes the creation of an ECD focused on the religious lexicon extracted from the earliest Baltic catechisms as one of its primary objectives5.
In the following sections, we will present the data model underlying DEC Facile (§2) and provide a brief overview of the tool’s backend and frontend components (§3). Finally, we will draw conclusions (§4).
The theoretical framework of the ECL, central to the development of DEC Facile, represents a cornerstone of modern lexicography and terminology. ECL’s structured approach to modelling lexical relations, combinatorial properties, and semantic nuances offers a powerful tool not only for dictionary creation but also for broader applications such as computational linguistics, terminography, and specialized language analysis.
At its core, an ECD entry is organized into three distinct sections, as illustrated in Figure 1 by the Lithuanian entry peržegnoti “to bless; to cross oneself (reflexive form)”. The first section (a) presents the definition(s) of the lexeme L, formulated using a predefined metalanguage, where the semantic actants [SemAs] introduced by L are explicitly identified as variables (X, Y, Z etc.). 2 https://babelnet.org/ 3 https://wordnet.princeton.edu/ 4 https://www.wikidata.org/wiki/Wikidata:Main_Page 5 The project, funded in 2023 and coordinated by Professor Pietro U. Dini, brings together historical linguistics, comparative religious studies, computational lexicography, and knowledge representation to achieve three interconnected objectives. First, it conducts a systematic contrastive analysis of the translation strategies used for religious vocabulary in the earliest Baltic (Old Lithuanian, Old Latvian and Old Prussian) Catechisms. Second, it develops a diachronic computational lexicon, formally describing key religious terminology according to the principles of Explanatory and Combinatorial Lexicology. Third, it examines the reception of fundamental Christian concepts in the Baltic region, tracing points of convergence and divergence with other European linguistic and cultural traditions, including Celtic and Slavic. These efforts come together in a formal ontology, where Pagan and Christian concepts are systematically represented. The Catechisms will be stored in a digital repository, where each religious term will be linked to its corresponding lexicon entry and ontology concept, ensuring semantic accessibility through advanced query mechanisms. For further details, see: https://www.ilc.cnr.it/en/progetti/ownw/. The second section (b) describes the morphosyntactic realization of the SemAs, represented as a rectangular matrix, where columns correspond to the SemAs, and rows indicate the possible morphosyntactic realizations. Finally, the third section (c), which is pivotal in Mel’čuk’s theory, focuses on lexical functions, which define the syntagmatic and paradigmatic relations that the lexeme L has with other dictionary entries.
This theoretical structure is operationalized and implemented in a computationally actionable format through the OntoLex-Lemon model, developed by the W3C Ontology-Lexicon community group. The model is a suite of RDF vocabularies (referred to as modules) specifically designed to represent and interlink lexicons, terminologies, lexical-semantic resources and, more broadly, language resources in the Linguistic Linked Open Data (LLOD) cloud [19], in alignment with semantic web best practices and LOD principle. Building on prior lexical modeling efforts— including LingInfo [20], LexOnto [21], LIR [22], and LexInfo [23]—the model was first developed within the Monnet project, an initiative aimed at supporting multilingual, ontology-based knowledge representation. Since its initial release, OntoLex-Lemon has undergone continuous refinements and extensions to enhance its expressiveness and interoperability, leading to its widespread adoption in projects focused on multilingual lexicography, terminology management, and computational linguistics (e.g., [24, 25, 26]).
The OntoLex-Lemon modules cover various aspects of linguistic representation, including morphology, syntax-semantics mapping, variation, translation, and linguistic metadata.
In DEC Facile, the data model is primarily structured around three key modules: the lexicography module (Lexicog), the lexical functions module (Lexfom), and the syntax-semantics module (SynSem).
Designed to represent existing dictionaries and lexicographic resources as Linked Data, the lexicography module7 extends the OntoLex core module (often referred to as OntoLex), enhancing its functionality to efficiently manage the structures and annotations commonly found 6 The Government Pattern in Section (b) and the Lexical Functions in Section (c) refer to sense I.1 7 https://www.w3.org/2019/09/lexicog/. in lexicographic practices. Referring to Figure 2(a), a lexical entry (LexicalEntry) is defined as an entry of a lexicon (Lexicon) characterized by a set of senses (LexicalSenses) and a set of written forms (not depicted in Figure 2(a)). A dictionary (LexicographicResource), in turn, is composed of a set of dictionary entries (Entry), each corresponding to the description of a lexical entry (Figure 2(b)). In DEC Facile, the lexicography module is particularly useful for modelling the ordering of senses, as an ECD requires lexical senses to be hierarchically organized, when applicable, according to three levels that reflect the semantic distance between lexemes: large distances, indicated by Roman numerals (I, II, III, etc.); medium distances, indicated by Arabic numerals (1, 2, 3, etc.); and small distances, indicated by lowercase letters (a, b, c). In our example entry peržegnoti, as shown in Figure 1, only large distances can be found. To represent this ordering of senses in Dec facile, the LexicographicComponent class is used, which is a structural element that reflects the arrangement of descriptions provided in a lexicographic resource according to various aspects such as order, hierarchy, grouping, etc.
Figure 3 illustrates a fragment of the encoding for the Lithuanian lexical entry peržegnoti. In lines 01 - 04, peržegnoti is defined as a verb with two lexical senses, whose definitions are provided in lines 20-23. Note that the two references to the senses of other dictionary entries, included in the definition of the first sense of peržegnoti, are formally encoded by the rdfs:seeAlso property (line 24). Lines 5-6 organize the lexical entry as a lexicographic component of the corresponding dictionary entry. Lines 7-19 define the ordering and the labelling (e.g., “I1”, “II1”) of the lexical senses. Each sense is linked to a lexicographic component via the describes property.
In order to model lexical functions among lexical senses, we used the LexFom [27] module. Based on the OntoLex elements, this module provides the necessary classes to implement lexical functions according to Mel’čuk’s theory. These include distinctions such as whether a lexical function is simple or complex, syntagmatic or paradigmatic, as well as identifying the constituents of a lexical function, and other related aspects. Referring to Figure 3, line 25, the lexical function S0 indicates the structural derivative formed through nominalization: peržegnojimas “blessing”. Lines 26 and 27 introduce the lexical functions S1 and S2, which represent meaningful derivatives reflecting specific syntactic roles. S1 corresponds to the typical designation of the Deep-Syntactic Actant I (e.g., Dievas “God”), broadly corresponding to the agent (or subject, from a syntactic perspective). S2, on the other hand, corresponds to the typical designation of the Deep-Syntactic Actant II (e.g., vyras “man” or moteris “woman”), typically corresponding to the patient (the syntactic object) introduced by the verb peržegnoti. 3. The tool Dec Facile In line with modern web application development practices, the architecture of the DEC Facile system is divided into a frontend (FE) and a backend (BE). The FE interface operates directly in the user's browser and is developed using Javascript, HTML, and CSS. Conversely, the BE component is implemented in Java and connects to a triple store to provide the services required by the FE.
BE is powered by LexO-server8 [28], a free and open-source backend built on the semantic repository GraphDB. It is implemented as a collection of REST (Representational State Transfer) services based on the HTTP protocol and exchanges data in JSON format. The services adhere to the OpenAPI specification, which facilitates the description, production, consumption, and visualization of RESTful APIs in a machine-readable format. LexO-server originates from LexOlite [29], a full-stack tool designed for editing OntoLex-Lemon resources. LexO-server manages both lexical and dictionary layers and has been enhanced to handle lexical functions.
As for the FE, the DEC Facile tool provides a streamlined interface for exploring and editing lexical and semantic information across multiple languages. As illustrated in Figure 4, the interface consists of several panels and functionalities, each designed to support the detailed construction and enrichment of an ECD9.
Search Panel. This panel enables users to search for specific dictionary entries in multiple languages. The language selection is conveniently located next to the dictionary entry field, marked by the “@” icon, allowing users to toggle between languages (e.g., “lt” for Lithuanian). Results display dictionary entries alongside their part-of-speech (POS), forms, and associated 8 The code is free and open source, available at the following link: https://github.com/andreabellandi/LexO-backend. 9 The following link provides access to the DecFacile interface code: https://github.com/Abdoumasbah/DecFacile. senses. These senses are hierarchically structured according to previously outlined principles, ensuring clarity and alignment with the linguistic framework.
Definition and Examples Panel. When a sense is selected, its corresponding definition and contextual examples are displayed in the right-hand panel. Users have the flexibility to modify the definition or add examples if they are missing in the source resource. This feature not only ensures adaptability but also allows for the enrichment of lexical data directly within the tool, supporting custom refinements and updates.
Lexical Functions Panel. This feature facilitates the linking of a selected sense from a dictionary entry to another sense through a specific lexical function (here e.g. S0). In addition, users can click on the ad hoc tab (labeled LF) on the right-hand side to access a comprehensive list of all lexical functions managed by the LexFom module. Selecting a lexical function displays a detailed box with its definition and an illustrative example. This functionality is particularly useful for users unfamiliar with the linguistic theory underlying lexical functions, offering accessible explanations and practical examples. Consequently, it simplifies the construction of EDCs by providing both theoretical context and practical guidance.
Government Pattern Panel. Currently under development, this panel will enable users to specify the syntactic regime of a sense by defining the morphological realizations of the SemAs introduced by that sense.
In this paper, we have presented an initial prototype of DEC Facile, a tool designed to assist users in constructing Explanatory Combinatorial Dictionaries, based on the theoretical framework of Explanatory Combinatorial Lexicology. Developed using semantic web technologies, the tool currently offers fundamental functionalities. Further development is planned, with a focus on two main areas: enhancing the prototype to support terminology construction for the PRIN project “Old Words for a New World: Translating Christianity to Baltic Pagans” and implementing the Government Pattern using the SynSem module of OntoLex-Lemon. As DEC Facile evolves, its features will be refined through ongoing collaboration with experts in lexicography and terminography, who apply this theoretical framework in their work.
The research presented in this paper was conducted within the framework of the PRIN project “Old Words for a New World: Translating Christianity to Baltic Pagans”, coordinated by Professor Pietro U. Dini and financed by the European Union – Next Generation EU, Mission 4 Component 1 CUP: B53D23014130006.
The authors have not employed any Generative AI tools. [13] J. Dancette, Les relations lexico-sémantiques dans un dictionnaire spécialisé, in: Th. Szende (Ed.), Le français dans les dictionnaires bilingues, Champion/Slatkine, Paris et Genève, 2006, pp. 144-156. [14] G. Sérasset, Un éditeur pour le dictionnaire explicatif et combinatoire du français contemporain, in : Journées lexique du PRC-CHM, Grenoble, France, 1996, pp.131-138. [15] N. Gader, V. Lux-Pogodalla, A. Polguere, Hand-Crafting a Lexical Network With a KnowledgeBased Graph Editor, in: Proceedings of the Third Workshop on Cognitive Aspects of the Lexicon (CogALex III), The COLING 2012 Organizing Committee, Mumbai, 2012, pp. 109– 125. [16] T. Berners-Lee, J.A. Hendler, O. Lassila, The Semantic Web: A new form of Web content that is meaningful to computers will unleash a revolution of new possibilities, Scientific American CCLXXXIV/5 (2001) 34-43. doi: https://doi.org/10.1038/scientificamerican0501-34. [17] Ch. Bizer, T. Heath, T. Berners-Lee, Linked data: The story so far, in: A. Sheth (Ed.), Semantic services, interoperability and web applications: emerging concepts, 1st ed., IGI Global, Hershey, Pennsylvania, 2011, pp. 205-227. [18] M. Wilkinson et al., The FAIR Guiding Principles for scientific data management and stewardship, Scientific Data 3, 2016. [19] C. Chiarcos, S. Nordhoff, S. Hellmann (Eds.), Linked Data in Linguistics, Springer, Berlin,
Heidelberg, 2012. doi: https://doi.org/10.1007/978-3-642-28249-2.
[20] P. Cimiano, P. Buitelaar, T. Declerck, A. Frank, S. Racioppa, M. Kiesel, R. Engel, M. Romanelli,
D. Sonntag, B. Loos, V. Micelli, R. Porzel, LingInfo: Design and Applications of a Model for the
Integration of Linguistic Information in Ontologies, in: Proceedings of the OntoLex 2006
Ontologies and Lexical Resources Interfacing Ontologies and Lexical Resources for Semantic
Web Technologies, Workshop at the LREC, Genoa, Italy, 27th May 2006, European Language
Resources Association (ELRA), pp. 28-32.
[21] P. Cimiano, P. Haase, M. Herold, M. Mantel, P. Buitelaar, LexOnto: A Model for Ontology
Lexicons for Ontology-based NLP, in: Proceedings of the OntoLex07 Workshop (held in
conjunction with ISWC'07), Busan, Korea, 11 November 2007.
[22] E. Montiel-Ponsoda, G. Aguado de Cea, A. Gómez-Pérez, W. Peters, W., Modelling
multilinguality in ontologies, in: Proceedings of the 21st International Conference on
Computational Linguistics (COLING), 8-22 August 2008, Manchester, UK, pp. 67-70.
[23] P. Cimiano, P. Buitelaar, J. McCrae, M. Sintek, Lexinfo: A declarative model for the
lexiconontology interface, Web Semantics: Science, Services and Agents on the World Wide Web 9
(