Comparison and evaluation of graph-based representations of sentence meaning is a challenge because competing representations of the same sentence may have different number of nodes, and it is not obvious which nodes should be compared to each other. Existing approaches favor node mapping that maximizes F_1 score over node relations and attributes, regardless whether the similarity is intentional or accidental; consequently, the identified mismatches in values of node attributes are not useful for any detailed error analysis. We propose a node-matching algorithm that allows comparison of multiple Uniform Meaning Representations (UMR) of one sentence and that takes advantage of node-word alignments, inherently available in UMR. We compare it with previously used approaches, in particular smatch (the de-facto standard in AMR evaluation), and argue that sensitivity to word alignment makes the comparison of meaning representations more intuitive and interpretable, while avoiding the NP-hard search problem inherent in smatch. A script implementing the method is freely available.

Uniform Meaning Representation (UMR) is a semantic framework designed to represent the meaning of texts in a structured and interpretable manner. In this paper, we evaluate the results of the automatic conversion of existing resources to UMR, focusing on Czech (PDT-C treebank) and Latin (LDT treebank). We present both quantitative and qualitative evaluations based on a comparison between manually and automatically generated UMR structures for a sample of Czech and Latin sentences. The findings indicate comparable results of the automatic conversion for both languages. The key challenges prove to be the higher level of semantic abstraction required by UMR and the fact that UMR allows for capturing semantic structure in multiple ways, potentially with varying levels of granularity.

We present LatinPipe, the winning submission to the EvaLatin 2024 Dependency Parsing shared task. Our system consists of a fine-tuned concatenation of base and large pre-trained LMs, with a dot-product attention head for parsing and softmax classification heads for morphology to jointly learn both dependency parsing and morphological analysis. It is trained by sampling from seven publicly available Latin corpora, utilizing additional harmonization of annotations to achieve a more unified annotation style. Before fine-tuning, we train the system for a few initial epochs with frozen weights. We also add additional local relative contextualization by stacking the BiLSTM layers on top of the Transformer(s). Finally, we ensemble output probability distributions from seven randomly instantiated networks for the final submission. The code is available at https://github.com/ufal/evalatin2024-latinpipe.

This paper focuses on the process of harmonising the five Latin treebanks available in Universal Dependencies with respect to morphological annotation. We propose a workflow that allows to first spot inconsistencies and missing information, in order to detect to what extent the annotations differ, and then correct the retrieved bugs, with the goal of equalising the annotation of morphological features in the treebanks and producing more consistent linguistic data. Subsequently, we present some experiments carried out with UDPipe and Stanza in order to assess the impact of such harmonisation on parsing accuracy.

This paper presents the harmonisation process carried out on the five treebanks available for Latin in Universal Dependencies, with the aim of eliminating the discrepancies in their annotation styles. Indeed, this is the first issue to be addressed when parsing Latin, as significant drops in parsing accuracy on different Latin treebanks have been repeatedly observed. Latin syntactic variability surely accounts for this, but parsing results are as well affected by divergent annotation choices. By analysing where annotations differ, we propose a Python-based alignment of the five UD treebanks. Consequently, the impact of annotation choices on accuracy scores is assessed by performing parsing experiments with UDPipe and Stanza.

This paper is framed in the context of the SSHOC project and aims at exploring how Language Technologies can help in promoting and facilitating multilingualism in the Social Sciences and Humanities (SSH). Although most SSH researchers produce culturally and societally relevant work in their local languages, metadata and vocabularies used in the SSH domain to describe and index research data are currently mostly in English. We thus investigate Natural Language Processing and Machine Translation approaches in view of providing resources and tools to foster multilingual access and discovery to SSH content across different languages. As case studies, we create and deliver as freely, openly available data a set of multilingual metadata concepts and an automatically extracted multilingual Data Stewardship terminology. The two case studies allow as well to evaluate performances of state-of-the-art tools and to derive a set of recommendations as to how best apply them. Although not adapted to the specific domain, the employed tools prove to be a valid asset to translation tasks. Nonetheless, validation of results by domain experts proficient in the language is an unavoidable phase of the whole workflow.

This paper investigates the recent advances in parsing the Index Thomisticus Treebank, which encompasses Medieval Latin texts by Thomas Aquinas. The research focuses on two types of variables. On the one hand, it examines the impact that a larger dataset has on the results of parsing; on the other hand, performances of new parsers are analysed with respect to less recent tools. Term of comparison to determine the effective parsing advances are the results in parsing the Index Thomisticus Treebank described in a previous work. First, the best performing parser among those concerned in that study is tested on a larger dataset than the one originally used. Then, some parser combinations that were developed in the same study are evaluated as well, assessing that more training data result in more accurate performances. Finally, to examine the impact that newly available tools have on parsing results, we train, test, and evaluate two neural parsers chosen among those best performing in the CoNLL 2018 Shared Task. Our experiments reach the highest accuracy rates achieved so far in automatic syntactic parsing of the Index Thomisticus Treebank and of Latin overall.

These texts are the result of Museology and collection studies course held by Marinella Pigozzi (a.a. 2017-18). They are about artworks from different collections, and proof the diversity of history of collecting approaches and sources. Texts by: Alex Barbanti, Giulia Calì, Ilenia Carozza, Ilaria Chia, Laura Coppa, Annapaola Di Maio, Cristina Elia, Chiara Forconi, Federica Gamba, Alessandra Lisbona, Marida Mariano, Serena Mauro, Francesca Passerini, Leonardo Regano, Guicciardo Sassoli de’ Bianchi Strozzi, Ilaria Siboni, Sara Zugni.

The article presents the process of linking the Dictionary of Medieval Latin in the Czech Lands to the LiLa Knowledge Base, which adopts the Linked Data paradigm to make linguistic resources for Latin interoperable. First, we provide an overview of the architecture of the LiLa Knowledge Base and of the Dictionary; then, we detail the stages of the process of linking the Dictionary to the collection of Latin lemmas that represents the core of LiLa. Once completed the linking process, we demonstrate to what extent the publication of the Dictionary as Linked Data proves beneficial by presenting the linking to LiLa of a new text from the same area and period covered by the Dictionary. In conclusion, a few queries illustrate how interoperability allows for full exploitation of a set of interlinked Latin resources.1