This paper describes the recently developed Latvian WordNet and the main linguistic principles used in its development. The inventory of words and senses is based on the Te̅zaurs.lv online dictionary, restructuring the senses of the most frequently used words based on corpus evidence. The semantic linking methodology adapts Princeton WordNet principles to fit the Latvian language usage and existing linguistic tradition. The semantic links include hyponymy, meronymy, antonymy, similarity, conceptual connection and gradation. We also measure inter-annotator agreement for different types of semantic links. The dataset consists of 7609 words linked in 6515 synsets. 1266 of these words are considered fully completed as they have all the outgoing semantic links annotated, corpus examples assigned for each sense, as well as links to the English Princeton WordNet formed. The data is available to the public on Te̅zaurs.lv as an addition to the general dictionary data, and is also published as a downloadable dataset.
: Latvian WordNet is a resource where word senses are connected based on their semantic relationships. The manual construction of a high-quality core Latvian WordNet is currently underway. However, text processing tasks require broad coverage, therefore, this work aims to extend the wordnet by automatically linking additional word senses in the Latvian online dictionary T¯ezaurs.lv and aligning them to the English-language Princeton WordNet (PWN). Our method only needs translation data, sense definitions and usage examples to compare it to PWN using pretrained word embeddings and sBERT. As a result, 57 927 interlanguage links were found that can potentially be added to Latvian WordNet, with an accuracy of 80% for nouns, 56% for verbs, 67% for adjectives and 66% for adverbs.
Arabic language lacks semantic datasets and sense inventories. The most common semantically-labeled dataset for Arabic is the ArabGlossBERT, a relatively small dataset that consists of 167K context-gloss pairs (about 60K positive and 107K negative pairs), collected from Arabic dictionaries. This paper presents an enrichment to the ArabGlossBERT dataset, by augmenting it using (Arabic-English-Arabic) machine back-translation. Augmentation increased the dataset size to 352K pairs (149K positive and 203K negative pairs). We measure the impact of augmentation using different data configurations to fine-tune BERT on target sense verification (TSV) task. Overall, the accuracy ranges between 78% to 84% for different data configurations. Although our approach performed at par with the baseline, we did observe some improvements for some POS tags in some experiments. Furthermore, our fine-tuned models are trained on a larger dataset covering larger vocabulary and contexts. We provide an in-depth analysis of the accuracy for each part-of-speech (POS).
In this paper we describe a new methodology to expand the Abui Wordnet through data collected using the Rapid Word Collection (RWC) method – based on SIL’s Semantic Domains. Using a multilingual sense-intersection algorithm, we created a ranked list of concept suggestions for each domain, and then used the ranked list as a filter to link the Abui RWC data to wordnet. This used translations from both SIL’s Semantic Domain’s structure and example words, both available through SIL’s Fieldworks software and the RWC project. We release both the new mapping of the SIL Semantic Domains to wordnet and an expansion of the Abui Wordnet.
Recent advances in Word Sense Disambiguation suggest neural language models can be successfully improved by incorporating knowledge base structure. Such class of models are called hybrid solutions. We propose a method of improving hybrid WSD models by harnessing data augmentation techniques and bilingual training. The data augmentation consist of structure augmentation using interlingual connections between wordnets and text data augmentation based on multilingual glosses and usage examples. We utilise language-agnostic neural model trained both with SemCor and Princeton WordNet gloss and example corpora, as well as with Polish WordNet glosses and usage examples. This augmentation technique proves to make well-known hybrid WSD architecture to be competitive, when compared to current State-of-the-Art models, even more complex.
GermaNet is a large lexical-semantic net that relates German nouns, verbs, and adjectives semantically. The word net has been manually constructed over the last 25 years and hence presents a high-quality, valuable resource for German. While GermaNet is maintained in a Postgres database, all its content can be exported as an XML-based serialisation. Recently, this XML representation has been converted into RDF, largely by staying close to GermaNet’s principle of arrangement where lexunits that share the same meaning are grouped together into so-called synsets. With each lexical unit and synset now globally addressable via a unique resource identifier, it has become much easier to link together GermaNet entries with other lexical and semantic resources. In terms of semantic interoperability, however, the RDF variant of GermaNet leaves much to be desired. In this paper, we describe yet another conversion from GermaNet’s XML representation to RDF. The new conversion makes use of the OntoLex-Lemon ontology, and therefore, presents a decisive step toward a GermaNet representation with a much higher level of semantic interoperability, and which makes it possible to use GermaNet with other wordnets that already support this conceptualisation of lexica.
This paper presents a work in progress about creating a Guarani version of the WordNet database. Guarani is an indigenous South American language and is a low-resource language from the NLP perspective. Following the expand approach, we aim to find Guarani lemmas that correspond to the concepts defined in WordNet. We do this through three strategies that try to select the correct lemmas from Guarani-Spanish datasets. We ran them through three different bilingual dictionaries and had native speakers assess the results. This procedure found Guarani lemmas for about 6.5 thousand synsets, including 27% of the base WordNet concepts. However, more work on the quality of the selected words will be needed in order to create a final version of the dataset.
Verbs like make, have and get present challenges for applications requiring automatic word sense discrimination. These verbs are both highly frequent and polysemous, with semantically “full” readings, as in make dinner, and “light” readings, as in make a request. Lexical resources like WordNet encode dozens of senses, making discrimination difficult and inviting proposals for reducing the number of entries or grouping them into coarser-grained supersenses. We propose a data-driven, linguistically-based approach to establishing a motivated sense inventory, focusing on make to establish a proof of concept. From several large, syntactically annotated corpora, we extract nouns that are complements of the verb make, and group them into clusters based on their Word2Vec semantic vectors. We manually inspect, for each cluster, the words with vectors closest to the centroid as well as a random sample of words within the cluster. The results show that the clusters reflect an intuitively plausible sense discrimination of make. As an evaluation, we test whether words within a given cluster cooccur in coordination phrases, such as apples and oranges, as prior work has shown that such conjoined nouns are semantically related. Conversely, noun complements from different clusters are less likely to be conjoined. Thus, coordination provides a similarity metric independent of the contextual embeddings used for clustering. Our results pave the way for a WordNet sense inventory that, while not inconsistent with the present one, would reduce it significantly and hold promise for improved automatic word sense discrimination.
This paper studies the application of pretrained BERT in the acquisition of synonyms, antonyms, hypernyms and hyponyms in Portuguese. Masked patterns indicating those relations were compiled with the help of a service for validating semantic relations, and then used for prompting three pretrained BERT models, one multilingual and two for Portuguese (base and large). Predictions for the masks were evaluated in two different test sets. Results achieved by the monolingual models are interesting enough for considering these models as a source for enriching wordnets, especially when predicting hypernyms of nouns. Previously reported performances on prediction were improved with new patterns and with the large model. When it comes to selecting the related word from a set of four options, performance is even better, but not enough for outperforming the selection of the most similar word, as computed with static word embeddings.
Systematic polysemy is a well-known linguistic phenomenon where a group of lemmas follow the same polysemy pattern. However, when compiling a lexical resource like a wordnet, a problem arises regarding when to underspecify the two (or more) meanings by one (complex) sense and when to systematically split into separate senses. In this work, we present an extensive analysis of the systematic polysemy patterns in Danish, and in our preliminary study, we examine a subset of these with experiments on human intuition and contextual embeddings. The aim of this preparatory work is to enable future guidelines for each polysemy type. In the future, we hope to expand this approach and thereby hopefully obtain a sense inventory which is distributionally verified and thereby more suitable for NLP.
The Open Multilingual Wordnet (OMW) is an open source project that was launched with the goal to make it easy to use wordnets in multiple languages without having to pay expensive proprietary licensing costs. As OMW evolved, the interlingual indicator (ILI)1 was used to allow semantically equivalent synsets in different languages to be linked to each other. OdeNet2 is the German language wordnet which forms part of the OMW project. This paper analyses the shortcomings of the initial ILI classification in OdeNet and the consequent methods used to improve this classification.
The Sanskrit WordNet is a resource currently under development, whose core was induced from a Vedic text sample semantically annotated by means of an ontology mapped on the Princeton WordNet synsets. Building on a previous case study on Ancient Greek (Zanchi et al. 2021), we show how sentence frames can be extracted from morphosyntactically parsed corpora by linking an existing dependency treebank of Vedic Sanskrit to verbal synsets in the Sanskrit WordNet. Our case study focuses on two verbs of asking, yāc- and prach-, featuring a high degree of variability in sentence frames. Treebanks enhanced with WordNet-based semantic information revealed to be of crucial help in motivating sentence frame alternations.
We present ongoing work dealing with a Linked Data compliant representation of infrastructures using wordnets for connecting multilingual Sign Language data sets. We build for this on already existing RDF and OntoLex representations of Open Multilingual Wordnet (OMW) data sets and work done by the European EASIER research project on the use of the CSV files of OMW for linking glosses and basic semantic information associated with Sign Language data sets in two languages: German and Greek. In this context, we started the transformation into RDF of a Danish data set, which links Danish Sign Language data and the wordnet for Danish, DanNet. The final objective of our work is to include Sign Language data sets (and their conceptual cross-linking via wordnets) in the Linguistic Linked Open Data cloud.
This paper presents the integration of WordNet knowledge resource into ClinIDMap tool, which aims to map identifiers between clinical ontologies and lexical resources. ClinIDMap interlinks identifiers from UMLS, SMOMED-CT, ICD-10 and the corresponding Wikidata and Wikipedia articles for concepts from the UMLS Metathesaurus. The main goal of the tool is to provide semantic interoperability across the clinical concepts from various knowledge bases. As a side effect, the mapping enriches already annotated medical corpora in multiple languages with new labels. In this new release, we add WordNet 3.0 and 3.1 synsets using the available mappings through Wikidata. Thanks to cross-lingual links in MCR we also include the corresponding synsets in other languages and also, extend further ClinIDMap with different domain information. Finally, the final resource helps in the task of enriching of already annotated clinical corpora with additional semantic annotations.
This paper shows how WordNets can be employed in tandem with morpho-syntactically annotated corpora to study poetic formulas. Pairing the lexico-semantic information of the Sanskrit WordNet with morpho-syntactic annotation from the Vedic Treebank, we perform a pilot study of formulas including SPEECH verbs in the RigVeda, the most ancient text of the. Sanskrit literature.
A model for preposition incorporation in the BulTreeBank WordNet is presented which follows the model for presenting open class words in wordnets. An adapted semantic classification of prepositions is done on the base of Bulgarian grammars and the classes are used for synset categories. The good coverage of prepositions in the wordnet will be used for the aim of neural language models creation for Bulgarian. This extension of the wordnet improves its utility for semantic annotation.
Focusing on recognition of multi-word expressions (MWEs), we address the problem of recording MWEs in WordNet. In fact, not all MWEs recorded in that lexical database could with no doubt be considered as lexicalised (e.g. elements of wordnet taxonomy, quantifier phrases, certain collocations). In this paper, we use a cross-encoder approach to improve our earlier method of distinguishing between lexicalised and non-lexicalised MWEs found in WordNet using custom-designed rule-based and statistical approaches. We achieve F1-measure for the class of lexicalised word combinations close to 80%, easily beating two baselines (random and a majority class one). Language model also proves to be better than a feature-based logistic regression model.
In this paper, we present the building of a CCGbank for Turkish by using standardised dependency corpora. We automatically induce Combinatory Categorial Grammar (CCG) categories for each word token in the Turkish dependency corpora. The CCG induction algorithm we present here is based on the dependency relations that are defined in the latest release of the Universal Dependencies (UD) framework. We aim for an algorithm that can easily be used in all the Turkish treebanks that are annotated in this framework. Therefore, we employ a lexicalist approach in order to make full use of the dependency relations while creating a semantically transparent corpus. We present the treebanks we employed in this study as well as their annotation framework. We introduce the structure of the algorithm we used along with the specific issues that are different from previous studies. Lastly, we show how the results change with this lexical approach in CCGbank for Turkish compared to the previous CCGbank studies in Turkish.
For Sign Languages (SLs), can we create a SignNet, like a WordNet for spoken languages: a network of semantic relations between constitutive elements of SLs? We first discuss approaches that link SL data to wordnets, or integrate such elements with some adaptations into the structure of WordNet. Then, we present requirements for a SignNet, which is built on SL data and then linked to WordNet.
