Search

In recent years, multimodal natural language processing, aimed at learning from diverse data types, has garnered significant attention. However, there needs to be more clarity when it comes to analysing multimodal tasks in multi-lingual contexts. While prior studies on sentiment analysis of tweets have predominantly focused on the English language, this paper addresses this gap by transforming an existing textual Twitter sentiment dataset into a multimodal format through a straightforward curation process. Our work opens up new avenues for sentiment-related research within the research community. Additionally, we conduct baseline experiments utilising this augmented dataset and report the findings. Notably, our evaluations reveal that when comparing unimodal and multimodal configurations, using a sentiment-tuned large language model as a text encoder performs exceptionally well.

This article presents a sentence-level sentiment dataset for the Croatian news domain. In addition to the 3K annotated texts already present, our dataset contains 14.5K annotated sentence occurrences that have been tagged with 5 classes. We provide baseline scores in addition to the annotation process and inter-annotator agreement.

This paper introduces Cro-FiReDa, a sentiment-annotated dataset for Croatian in the domain of movie reviews. The dataset, which contains over 10,000 sentences, has been annotated at the sentence level. In addition to presenting the overall annotation process, we also present benchmark results based on the transformer-based fine-tuning approach

With the ever-growing popularity of the field of NLP, the demand for datasets in low resourced-languages follows suit. Following a previously established framework, in this paper, we present the UNER dataset, a multilingual and hierarchical parallel corpus annotated for named-entities. We describe in detail the developed procedure necessary to create this type of dataset in any language available on Wikipedia with DBpedia information. The three-step procedure extracts entities from Wikipedia articles, links them to DBpedia, and maps the DBpedia sets of classes to the UNER labels. This is followed by a post-processing procedure that significantly increases the number of identified entities in the final results. The paper concludes with a statistical and qualitative analysis of the resulting dataset., Comment: arXiv admin note: substantial text overlap with arXiv:2212.07162

This article presents the application of the Universal Named Entity framework to generate automatically annotated corpora. By using a workflow that extracts Wikipedia data and meta-data and DBpedia information, we generated an English dataset which is described and evaluated. Furthermore, we conducted a set of experiments to improve the annotations in terms of precision, recall, and F1-measure. The final dataset is available and the established workflow can be applied to any language with existing Wikipedia and DBpedia. As part of future research, we intend to continue improving the annotation process and extend it to other languages.

This paper presents a cross-lingual sentiment analysis of news articles using zero-shot and few-shot learning. The study aims to classify the Croatian news articles with positive, negative, and neutral sentiments using the Slovene dataset. The system is based on a trilingual BERT-based model trained in three languages: English, Slovene, Croatian. The paper analyses different setups using datasets in two languages and proposes a simple multi-task model to perform sentiment classification. The evaluation is performed using the few-shot and zero-shot scenarios in single-task and multi-task experiments for Croatian and Slovene.

In the paper we begin to explore relation between the question of unitarizability of classical p-adic groups, and Arthur packets., Comment: 35 pages; this is the last version of the manuscript before publication

This article presents the results of the evaluation campaign of language tools available for fifteen EU-official under-resourced languages. The evaluation was conducted within the MSC ITN CLEOPATRA action that aims at building the cross-lingual event-centric knowledge processing on top of the application of linguistic processing chains (LPCs) for at least 24 EU-official languages. In this campaign, we concentrated on three existing NLP platforms (Stanford CoreNLP, NLP Cube, UDPipe) that all provide models for under-resourced languages and in this first run we covered 15 under-resourced languages for which the models were available. We present the design of the evaluation campaign and present the results as well as discuss them. We considered the difference between reported and our tested results within a single percentage point as being within the limits of acceptable tolerance and thus consider this result as reproducible. However, for a number of languages, the results are below what was reported in the literature, and in some cases, our testing results are even better than the ones reported previously. Particularly problematic was the evaluation of NERC systems. One of the reasons is the absence of universally or cross-lingually applicable named entities classification scheme that would serve the NERC task in different languages analogous to the Universal Dependency scheme in parsing task. To build such a scheme has become one of our the future research directions.

This article presents the strategy for developing a platform containing Language Processing Chains for European Union languages, consisting of Tokenization to Parsing, also including Named Entity recognition andwith addition ofSentiment Analysis. These chains are part of the first step of an event-centric knowledge processing pipeline whose aim is to process multilingual media information about major events that can cause an impactin Europe and the rest of the world. Due to the differences in terms of availability of language resources for each language, we have built this strategy in three steps, starting with processing chains for the well-resourced languages and finishing with the development of new modules for the under-resourced ones. In order to classify all European Union official languages in terms of resources, we have analysed the size of annotated corpora as well as the existence of pre-trained models in mainstream Language Processing tools, and we have combined this information with the proposed classification published at META-NETwhitepaper series.

We introduce the Universal Named-Entity Recognition (UNER)framework, a 4-level classification hierarchy, and the methodology that isbeing adopted to create the first multilingual UNER corpus: the SETimesparallel corpus annotated for named-entities. First, the English SETimescorpus will be annotated using existing tools and knowledge bases. Afterevaluating the resulting annotations through crowdsourcing campaigns,they will be propagated automatically to other languages within the SE-Times corpora. Finally, as an extrinsic evaluation, the UNER multilin-gual dataset will be used to train and test available NER tools. As part offuture research directions, we aim to increase the number of languages inthe UNER corpus and to investigate possible ways of integrating UNERwith available knowledge graphs to improve named-entity recognition.

Let G be the F-points of a classical group defined over a p-adic field F of characteristic 0. We classify the irreducible unitarizable representation of G that are subquotients of the parabolic induction of cuspidal representations of Levi subgroup of corank at most 3 in G., Comment: This paper is based on our earlier manuscript Unitarizability in generalised rank three for classical p-adic groups (arXiv:1709.00630), which is now substantially revised (organisation of the paper and geometric exhaustion argument are only some of principal differences). It has iv+119 pages and 15 figures. To appear in the Memoirs of the AMS

Multilingualism is a cultural cornerstone of Europe and firmly anchored in the European treaties including full language equality. However, language barriers impacting business, cross-lingual and cross-cultural communication are still omnipresent. Language Technologies (LTs) are a powerful means to break down these barriers. While the last decade has seen various initiatives that created a multitude of approaches and technologies tailored to Europe's specific needs, there is still an immense level of fragmentation. At the same time, AI has become an increasingly important concept in the European Information and Communication Technology area. For a few years now, AI, including many opportunities, synergies but also misconceptions, has been overshadowing every other topic. We present an overview of the European LT landscape, describing funding programmes, activities, actions and challenges in the different countries with regard to LT, including the current state of play in industry and the LT market. We present a brief overview of the main LT-related activities on the EU level in the last ten years and develop strategic guidance with regard to four key dimensions., Comment: Proceedings of the 12th Language Resources and Evaluation Conference (LREC 2020). To appear

M. Hanzer and I. Matic have proved that the genuine unitary principal series representations of the metaplectic groups are irreducible. A simple consequence of that paper is a criterion for the irreducibility of the non-unitary principal series representations of the metaplectic groups that we give in this paper., Comment: 9 pages, to appear in Rad HAZU, Matematicke Znanosti

In the introduction of this paper we discuss a possible approach to the unitarizability problem for classical p-adic groups. In this paper we give some very limited support that such approach is not without chance. In a forthcoming paper we shall give additional evidence in generalized cuspidal rank (up to) three., Comment: This paper is a merged and revised version of ealier preprints arXiv:1701.07658 and arXiv:1701.07662. The paper is going to appear in the Proceedings of the Simons Symposium on Geometric Aspects of the Trace Formula

In an earlier paper we propose an approach to the unitarizability problem in the case of classical groups over a p-adic field of characteristic zero based on cuspidal reducibility points. We have reduced earlier the unitarizability for these groups to the case of so called weakly real representations. Following C. Jantzen, to an irreducible weakly real representation $\pi$ of a classical group one can attach a sequence ($\pi_1,\dots,\pi_k)$ of irreducible representations of classical groups, each of them supported by a line of cuspidal representations $X_\rho$ of general linear groups containing a selfcontragredient representation $\rho$, and an irreducible cuspidal representation $\sigma$ of a classical group. The first question is if $\pi$ is unitarizable if and only if all $\pi_i$ are unitarizable. Further, the pair $\rho,\sigma$ determines the non-negative reducibility exponent $\alpha_{\rho,\sigma}\in\frac12\mathbb Z$ among $\rho$ and $\sigma$. The question is if the unitarizability of irreducible representations supported by $X_\rho\cup \sigma$ depends only on $\alpha_{\rho,\sigma}$. Following the above proposed strategy, in this paper we solve the unitarizability problem for irreducible subquotients of representations Ind$_P^G(\tau)$, where G is a classical group over a p-adic field of characteristic zero, P is a parabolic subgroup of G of the generalized rank (at most) 3 and $\tau$ is an irreducible cuspidal representation of a Levi factor M of P. As a consequence, this gives also a solution of the unitarizability problem for classical p-adic groups of the split rank (at most) three. This paper also provides some very limited support for the possibility of the above approach to the unitarizability could work in general., Comment: The substantial revised content of this paper is in a new version under the new name "Unitarizability in Corank Three for Classical p-adic Groups " ID 2006.12609

Given a (complex, smooth) irreducible representation $\pi$ of the general linear group over a non-archimedean local field and an irreducible supercuspidal representation $\sigma$ of a classical group, we show that the (normalized) parabolic induction $\pi\rtimes\sigma$ is reducible if there exists $\rho$ in the supercuspidal support of $\pi$ such that $\rho\rtimes\sigma$ is reducible. In special cases we also give irreducibility criteria for $\pi\rtimes\sigma$ when the above condition is not satisfied.

In the case of p-adic general linear groups, each irreducible representation is parabolically induced by a tensor product of irreducible representations supported by cuspidal lines. One gets in this way a parameterization of the irreducible representations of p-adic general linear groups by irreducible representations supported by cuspidal lines. It is obvious that in this correspondence an irreducible representation of a p-adic general linear group is unitarizable if and only if all the corresponding irreducible representations supported by cuspidal lines are unitarizable. C. Jantzen has defined an analogue of such correspondence for irreducible representations of classical p-adic groups. It would have interesting consequences if one would know that the unitarizability is also preserved in this case. A purpose of this paper and its sequel, is to give some very limited support for possibility of such preservation of the unitarizability. More precisely, we show that if we have an irreducible unitarizable representation $\pi$ of a classical p-adic group whose one attached representation $\pi_L$ supported by a cuspidal line $L$ has the same infinitesimal character as the generalized Steinberg representation supported by that cuspidal line, then $\pi_L$ is unitarizable., Comment: This preprint is merged with the preprint "On unitarity of some representations of classical p-adic groups II " ID 1701.07662, and published under new title "On unitarizability in the case of classical p-adic groups" ID 1709.00626

In this paper we give a simple (local) proof of two principal results about irreducible tempered representations of general linear groups over a non-archimedean local division algebra. We give a proof of the parameterization of the irreducible square integrable representations of these groups by segments of cuspidal representations, and a proof of the irreducibility of the tempered parabolic induction. Our proofs are based on Jacquet modules (and the Geometric Lemma, incorporated in the structure of a Hopf algebra). We use only some very basic general facts of the representation theory of reductive p-adic groups (the theory that we use was completed more then three decades ago, mainly in 1970-es). Of the specific results for general linear groups over A, basically we use only a very old result of G.I. Olshanskii, which says that there exist complementary series starting from $Ind(\rho\otimes\rho)$ whenever $\rho$ is a unitary irreducible cuspidal representation. In appendix of the paper "On parabolic induction on inner forms of the general linear group over a non-archimedean local field" of E. Lapid and A. Minguez, there is also a simple local proof of these results, based on a slightly different approach., Comment: 26 pages

In the first part of the paper we give some bounds for domains where the unitarizabile subquotients can show up in the parabolically induced representations of classical p-adic groups. Roughly, it can show up only if the central character of the inducing irreducible cuspidal representation is dominated in an appropriate way by the square root of the modular character of minimal parabolic subgroup. For representations supported by fixed parabolic subgroup, a more precise bound is given. There are also bounds for specific Bernstein components. A number of these upper bounds are best possible. The second part of the paper addresses a question how far is the trivial representation from the rest of the unramified automorphic dual. By a result of L. Clozel, trivial representation is isolated in the automorphic dual of a split rank one semisimple group over a completion of a global field, but it is very seldom isolated in the unitary dual (it can happen only in the archimedes cases). Further, the level of isolation in the case of SL(2) is important for the number theory. For the higher rank groups, the trivial representation is always isolated in the unitary dual by an old result of D. Kazhdan. Still, we may ask if the level of isolation is higher in the case of the automorphic duals. We show that the answer is negative to this question for symplectic p-adic groups., Comment: 28 pages

We study representations of segment type of groups Sp(n) and SO(2n+1, F) over a local non-archimedean field, which play a fundamental role in the constructions of discrete series, and obtain a complete description of the Jacquet modules of these representations. Also, we provide an alternative way for determination of Jacquet modules of strongly positive discrete series and a description of top Jacquet modules of general discrete series. In this version are corrected few typographical errors which exist in the published version of this paper (see page 5 of the paper for more details)., Comment: 41 pages

In this paper we consider reducibility points beyond the ends of complementary series of general linear groups over a p-adic field, which start with Speh representations. We describe explicitly the composition series of the representations at these reducibility points. They are multiplicities one representations, and they can be of arbitrary length. We give Langlands parameters of all the irreducible subquotients., Comment: In this revised version, some proofs are expanded (two chapters are added). The paper is going to appear in the Journal of Lie Theory

In this paper we review some connections between harmonic analysis and the modern theory of automorphic forms. We indicate in some examples how the study of problems of harmonic analysis brings us to the important objects of the theory of automorphic forms, and conversely. We consider classical groups and their unitary, tempered, automorphic and unramified duals. The most important representations in our paper are the isolated points in these duals., Comment: 49 pages, 1 figure

This paper has two aims. The first is to give a description of irreducible tempered representations of classical p-adic groups which follows naturally the classification of irreducible square integrable representations modulo cuspidal data obtained by C. Moeglin and the author. The second aim of the paper is to give description of an invariant (partially defined function) of irreducible square integrable representation of a classical p-adic group (defined by C. Moeglin using embeddings) in terms of subquotients of Jacquet modules. As an application, we describe behavior of partially defined function in one construction of square integrable representations of a bigger group from such representations of a smaller group (which is related to deformation of Jordan blocks of representations)., Comment: 58 pages

Let A be a finite dimensional central division algebra over a local non-archimedean field F. Fix any parabolic subgroup P of GL(n,A) and a Levi factor M of P. Let \pi be an irreducible unitary representation of M and \phi (not necessarily unitary) character of M. We give an explicit necessary and sufficient condition for the parabolically induced representation Ind(\phi\pi) of GL(n,A) to be irreducible., Comment: 40 pages

The main aim of this paper is to present the ideas which lead first to the solution of the unitarizability problem for $\GL(n)$ over nonarchimedean local fields and to the recognition that the same result holds over archimedean local fields, a result which was proved by Vogan using an internal approach. Let us say that the approach that we are going to present may be characterized as external. At no point do we go into the internal structure of representations., Comment: 38 pages. Abstract added in migration.