For human-robot dialogue in a search-and-rescue scenario, a strong knowledge of the conditions and objects a robot will face is essential for effective interpretation of natural language instructions. In order to utilize the power of large language models without overwhelming the limited storage capacity of a robot, we propose PropBank-Powered Data Creation. PropBank-Powered Data Creation is an expert-in-the-loop data generation pipeline which creates training data for disaster-specific language models. We leverage semantic role labeling and Rich Event Ontology resources to efficiently develop seed sentences for fine-tuning a smaller, targeted model that could operate onboard a robot for disaster relief. We developed 32 sentence templates, which we used to make 2 seed datasets of 175 instructions for earthquake search and rescue and train derailment response. We further leverage our seed datasets as evaluation data to test our baseline fine-tuned models.

This paper highlights some theoretical and quantitative issues related to the representation and annotation of aspectual meaning in the IMAGACT corpus-based multimodal ontology of action. Given the multimodal nature of this ontology, in which actions are represented through both prototypical visual scenes and linguistic captions, the annotation of aspect in this resource allows us to draw some important considerations about the relation between aspectual meaning and eventualities. The annotation procedure is reported and quantitative data show that, both in the English and Italian corpora, many verbs present aspectual variation, and many eventualities can be represented by locally equivalent verbs with different aspect. The reason why verb aspectual class may vary is investigated. Our analysis makes once more evident that verbs may vary their aspectual properties with respect not only to their argument structure but, more precisely, to the inner qualities of the eventualities they express. Crucially, when eventualities are expressed by equivalent verbs with different aspectual properties, the verbs put on focus different parts of the structure of the eventuality.

In this paper, we describe NoVRol, a semantic role lexicon of Norwegian verbs. We start from the NorVal valency lexicon, which describes the syntactic frames of 7.400 verbs. We then enrich each of these frames by annotating, based on the VerbNet annotation scheme, each argument of the verb with the semantic role that it gets. We also encode the syntactic roles of the arguments based on the UD annotation scheme. Our resource will faciliate future research on Norwegian verbs, and can at a future stage be expanded to a full VerbNet

Semantic role labeling (SRL) resources, such as Proposition Bank (PropBank), provide useful input to downstream applications. In this paper we present some challenges and insights we learned while expanding the previously developed Russian PropBank. This new effort involved annotation and adjudication of all predicates within a subset of the prior work in order to provide a test corpus for future applications. We discuss a number of new issues that arose while developing our PropBank for Russian as well as our solutions. Framing issues include: distinguishing between morphological processes that warrant new frames, differentiating between modal verbs and predicate verbs, and maintaining accurate representations of a given language’s semantics. Annotation issues include disagreements derived from variability in Universal Dependency parses and semantic ambiguity within the text. Finally, we demonstrate how Russian sentence structures reveal inherent limitations to PropBank’s ability to capture semantic data. These discussions should prove useful to anyone developing a PropBank or similar SRL resources for a new language.

This study evaluates the extent to which semantic information is preserved within sentence embeddings generated from state-of-art sentence embedding models: SBERT and LaBSE. Specifically, we analyzed 13 semantic attributes in sentence embeddings. Our findings indicate that some semantic features (such as tense-related classes) can be decoded from the representation of sentence embeddings. Additionally, we discover the limitation of the current sentence embedding models: inferring meaning beyond the lexical level has proven to be difficult.

This article discusses the challenges to meaning representation of terms posed by a quantum theory of terms (QTT) that was recently reported. We first summarize this theory and then highlight the difficulties of representing quanterms, which is the name we coined for the view that the QTT has of terms as quantum systems by analogy with quantum objects in quantum mechanics. We briefly summarize the representation practices followed to date to record and represent terminology. We use findings reported in the literature to model both terms and quanterms and found that current representations of terms in specialized repositories are collapsed quanterms at the expense of other states of the original quanterm. In this work, both quanterms and collapsed quanterms are mathematically modelled following formulations used in quantum mechanics. These formulations suggest that representations of quanterms need to include information about the probabilities of quanterm states and the role they play in the entanglement of terms for phenomena such as specialized collocations.

In this paper, we introduce a novel meaning representation, which is based on AMR but extends it towards a visual ontological representation. We visualize concepts by representative images, and roles by emojis. All concepts are identified either by PropBank rolesets, Wikipedia page titles, WordNet synsets, or Wikidata lexeme senses. We have developed a Web-based annotation environment enabled by augmented browsing and interactive diagramming. As first application, we have implemented a multilingual annotation solution by using English as anchor language and comparing it with French and Japanese language versions. Therefore, we have extended our representation by a translation deviation annotation to document the differences between the language versions. The intended user groups are, besides professional translators and interpreters, students of translation, language, and literary studies. We describe a first use case in which we use novels by French authors and compare them with their English and Japanese translations. The main motivation for choosing Japanese is the soaring popularity of Japanese courses at our university and the particular challenges involved with trying to master this language.

In this paper, we present the first version of YARN, a new semantic representation formalism. We propose this new formalism to unify the advantages of logic-based formalisms while retaining direct interpretation, making it widely usable. YARN is rooted in the encoding of different semantic phenomena as separate layers. We begin by presenting a formal definition of the mathematical structure that constitutes YARN. We then illustrate with concrete examples how this structure can be used in the context of semantic representation for encoding multiple phenomena (such as modality, negation and quantification) as layers built on top of a central predicate-argument structure. The benefit of YARN is that it allows for the independent annotation and analysis of different phenomena as they are easy to “switch off”. Furthermore, we have explored YARN’s ability to encode simple interactions between phenomena. We wrap up the work presented by a discussion of some of the interesting observations made during the development of YARN so far and outline our extensive future plans for this formalism.

We present a contrastive study of argument sharing across three graph-based meaning representation frameworks, where semantically shared arguments manifest as reentrant graph nodes. For a state-of-the-art graph parser, we observe how parser performance – in terms of output quality – covaries with overall graph complexity, on the one hand, and presence of different types of reentrancies, on the other hand. We identify common linguistic phenomena that give rise to shared arguments, and therefore node reentrancies, through a small-case and partially automated annotation study and parallel error anaylsis of actual parser outputs. Our results provide new insights into the distribution of different types of reentrancies in meaning representation graphs for three distinct frameworks, as well as on the effects that these structures have on parser performance, thus suggesting both novel cross-framework generalisations as well as avenues for focussed parser development.

For many years, there has been attempts to compare predicate-argument labeling schemas between formalism, typically under the dependency assumptions (even if the annotation by these schemas could have been performed on either constituent-based specifications or dependency ones). Given the growing number of resources that link various lexical resources to one another, as well as thanks to parallel annotated corpora (with or without annotation), it is now possible to do more in-depth studies of those correspondences. We present here a high-coverage pilot study of mapping the labeling system used in PropBank (for English) to Czech, which has so far used mainly valency lexicons (in several closely related forms) for annotation projects, under a different level of specification and different theoretical assumptions. The purpose of this study is both theoretical (comparing the argument labeling schemes) and practical (to be able to annotate Czech under the standard UMR specifications).

This paper presents an adaptation of the Abstract Meaning Representation (AMR) framework for European Portuguese. This adaptation, referred to as Lexicalized Meaning Representation (LMR), was deemed necessary to address specific challenges posed by the grammar of the language, as well as various linguistic issues raised by the current version of AMR annotation guidelines. Some of these aspects stemmed from the use of a notation similar to AMR to represent real texts from the legal domain, enabling its use in Natural Language Processing (NLP) applications. In this context, several aspects of AMR were significantly simplified (e.g., the representation of multi-word expressions, named entities, and temporal expressions), while others were introduced, with efforts made to maintain the representation scheme as compatible as possible with standard AMR notation.

We evaluate the ability of large language models (LLMs) to provide PropBank semantic role label annotations across different realizations of the same verbs in transitive, intransitive, and middle voice constructions. In order to assess the meta-linguistic capabilities of LLMs as well as their ability to glean such capabilities through in-context learning, we evaluate the models in a zero-shot setting, in a setting where it is given three examples of another verb used in transitive, intransitive, and middle voice constructions, and finally in a setting where it is given the examples as well as the correct sense and roleset information. We find that zero-shot knowledge of PropBank annotation is almost nonexistent. The largest model evaluated, GPT-4, achieves the best performance in the setting where it is given both examples and the correct roleset in the prompt, demonstrating that larger models can ascertain some meta-linguistic capabilities through in-context learning. However, even in this setting, which is simpler than the task of a human in PropBank annotation, the model achieves only 48% accuracy in marking numbered arguments correctly. To ensure transparency and reproducibility, we publicly release our dataset and model responses.

This work proposes expanding the thematic role selectional preferences used in the lexical resource VerbNet as a way to increase the available semantic information in the resource, induce semantically-based subclasses for the more generic VerbNet classes, and create new links across classes. The addition of verb-specific features in the latest version of VerbNet provides a means for adding more specific selectional preferences based on the meaning of a class’s individual member verbs. These features could refine both the instantiated class roles and the new implicit roles introduced in VerbNet version 4. We suggest 49 classes that would benefit from 111 verb-specific selectional preferences and explain how they would enhance VerbNet’s semantic representations.

We explore using LLMs, GPT-4 specifically, to generate draft sentence-level Chinese Uniform Meaning Representations (UMRs) that human annotators can revise to speed up the UMR annotation process. In this study, we use few-shot learning and Think-Aloud prompting to guide GPT-4 to generate sentence-level graphs of UMR. Our experimental results show that compared with annotating UMRs from scratch, using LLMs as a preprocessing step reduces the annotation time by two thirds on average. This indicates that there is great potential for integrating LLMs into the pipeline for complicated semantic annotation tasks.

Despite Uniform Meaning Representation’s (UMR) potential for cross-lingual semantics, limited annotated data has hindered its adoption. There are large datasets of English AMRs (Abstract Meaning Representations), but the process of converting AMR graphs to UMR graphs is non-trivial. In this paper we address a complex piece of that conversion process, namely cases where one AMR role can be mapped to multiple UMR roles through a non-deterministic process. We propose a neuro-symbolic method for role conversion, integrating animacy parsing and logic rules to guide a neural network, and minimizing human intervention. On test data, the model achieves promising accuracy, highlighting its potential to accelerate AMR-to-UMR conversion. Future work includes expanding animacy parsing, incorporating human feedback, and applying the method to broader aspects of conversion. This research demonstrates the benefits of combining symbolic and neural approaches for complex semantic tasks.

This paper evaluates how well English Abstract Meaning Representation parsers process an important and frequent kind of Long-Distance Dependency construction, namely, relative clauses (RCs). On two syntactically parsed datasets, we evaluate five AMR parsers at recovering the semantic reentrancies triggered by different syntactic subtypes of relative clauses. Our findings reveal a general difficulty among parsers at predicting such reentrancies, with recall below 64% on the EWT corpus. The sequence-to-sequence models (regardless of whether structural biases were included in training) outperform the compositional model. An analysis by relative clause subtype shows that passive subject RCs are the easiest, and oblique and reduced RCs the most challenging, for AMR parsers.

The Parallel Meaning Bank (PMB) serves as a corpus for semantic processing with a focus on semantic parsing and text generation. Currently, we witness an excellent performance of neural parsers and generators on the PMB. This might suggest that such semantic processing tasks have by and large been solved. We argue that this is not the case and that performance scores from the past on the PMB are inflated by non-optimal data splits and test sets that are too easy. In response, we introduce several changes. First, instead of the prior random split, we propose a more systematic splitting approach to improve the reliability of the standard test data. Second, except for the standard test set, we also propose two challenge sets: one with longer texts including discourse structure, and one that addresses compositional generalization. We evaluate five neural models for semantic parsing and meaning-to-text generation. Our results show that model performance declines (in some cases dramatically) on the challenge sets, revealing the limitations of neural models when confronting such challenges.