Modelling lexicographic resources using CIDOC CRM, FRBRoo and Ontolex Lemon

Fahad Shahbaz Khan, Ana SalgadoZenodo 2021 · DOI

TL;DR

This paper proposes a novel approach to model and publish lexicographic resources, like dictionaries, as linked data. By integrating CIDOC-CRM, FRBRoo, and Ontolex-Lemon, they create a richer, more interconnected representation that captures both the linguistic content and the historical/bibliographic aspects of these resources. This work is crucial for making historical dictionaries more accessible and queryable in the digital age.

What Problem Does This Paper Address?

Lexicographic resources, especially historical dictionaries that have been digitized (retro-digitised), present a complex modeling challenge. These resources are not just collections of words and definitions; they are also cultural artifacts with their own publishing histories, editorial decisions, and even typographical layouts. Existing models, like TEI-XML, offer different "views" (typographical, editorial, lexical) but are not ideal for the interconnectedness and semantic richness required for linked data on the Semantic Web. The paper aims to bridge this gap by creating a unified model that can represent these multiple facets of lexicographic resources in a machine-actionable way, facilitating interoperability and deeper analysis.

Methods

The core of the paper's contribution lies in proposing a new conceptual modeling approach. They achieve this by:

  • Integrating Existing Ontologies: The approach leverages the CIDOC-CRM (for cultural heritage), FRBRoo (for bibliographic entities), and Ontolex-Lemon (for lexical data).
  • Defining New Classes and Properties: To bridge the gaps between these existing ontologies and to specifically address the nuances of lexicographic resources, they introduce new classes like Lexicographic Work and Lexicographic Expression. These new classes are designed to be subclasses of both FRBRoo and Ontolex-Lemon classes, allowing for inheritance of properties and a more unified representation.
  • Modeling Multiple Views: The approach explicitly aims to model the different "views" of a dictionary (typographical, editorial, lexical) and relate them to each other and to the underlying linguistic content. This is achieved by mapping these views to FRBRoo concepts like Work, Expression, and Manifestation.
  • Leveraging Provenance and Citation Ontologies: They also propose using ontologies like Prov-O and CiTO to capture provenance information and the relationships between citations and attestations within dictionaries.

Datasets

The paper does not explicitly list datasets used for training or evaluation in the traditional machine learning sense. Instead, it uses examples from existing lexicographic resources to illustrate its modeling approach. Specifically, it discusses:

  • The Liddell-Scott-Jones Ancient Greek-English Lexicon (LSJ).
  • The Italian verb riprovare as discussed in the Il vocabolario Treccani and Il Grande Dizionario della Lingua Italiana (GDLL).

The paper's focus is on the modeling of these resources, not on training models on them.

Key Insights and Ideas

  • Dictionaries are More Than Just Words: A key insight is the recognition that dictionaries are dual entities: they are both linguistic resources (containing lexical data) and cultural artifacts (with a history, publication details, and specific presentation). The proposed model effectively captures this duality.
  • Bridging the Gap Between Bibliographic and Lexical Data: The paper highlights the challenge of linking bibliographic information (like editions, authors, publication dates) with the actual lexical content (entries, senses, forms). Their approach provides a robust way to do this using FRBRoo and Ontolex-Lemon.
  • The Importance of Citations and Attestations: The paper emphasizes the need to model citations and attestations not just as simple links, but as complex relationships that can be erroneous or have different levels of veracity. This is crucial for understanding the historical development and reliability of lexicographic claims.
  • Unified Representation for Interoperability: By creating a common linked data model, the authors pave the way for greater interoperability between different lexicographic resources and other cultural heritage data.

Key Results

The paper does not present quantitative results in terms of performance metrics. Instead, its key results are conceptual and methodological:

  • A Novel Conceptual Model: The primary result is the proposed ontology-based model that integrates CIDOC-CRM, FRBRoo, and Ontolex-Lemon to represent lexicographic resources at multiple descriptive levels.
  • Demonstration of the Model: The paper demonstrates the applicability of its approach through detailed modeling of the riprovare example, showing how it can capture complex relationships between dictionary entries, textual attestations, and provenance information.
  • Foundation for Future Work: The paper lays the groundwork for future research and development in making lexicographic resources more accessible and analyzable as linked data.

Limitations and Future Work

  • Scope of New Classes: While the paper introduces Lexicographic Work and Lexicographic Expression, it acknowledges that further development might be needed for other specific aspects of lexicographic resources.
  • Ongoing Development of Modules: The paper mentions the use of the Frequency Attestation and Corpus Information (FrAC) module, which was still under development at the time of writing. The finalization and integration of such modules will be important.
  • Large-Scale Application: The authors are currently involved in a project to digitize and make available three editions of the Diccionario da Lingua Portugueza. This project will serve as a large-scale testbed for their proposed approach, allowing them to apply and refine it on real-world data.
  • Representing Linguistic Hypotheses: They are also working on representing linguistic hypotheses, such as etymologies, using CRMSci, indicating a direction for extending their modeling capabilities.

Summary updated

Abstract

This presentation describes a new approach to the modelling and publication of lexicographic resources, including retro-digitised dictionaries, as linked data. This approach is based on the use of the CIDOCCRM aligned FRBRoo ontology together with the Ontolex-Lemon vocabulary and its follow up lexicographic module, lexicog.
BibTeX
@inproceedings{khan2021modelling,
  author = {Fahad Shahbaz Khan and Ana Salgado},
  title = {{Modelling lexicographic resources using CIDOC CRM, FRBRoo and Ontolex Lemon}},
  booktitle = {Zenodo},
  year = {2021},
  doi = {10.5281/zenodo.8162949},
}
Figure