Text Corpus Research Articles

Epic word: a comparative analysis of the nominations of the <i>world</i> in the Elder and Younger Eddas

The authorʼs goal is to compare the nominations of the world in the “Elder” and “Younger Edda”, including three types of objects: lexeme (hel, Hel ‘hel; owner of another worldʼ), combination of words (Ginnunga gap ‘gaping abyssʼ), composite (Ás-garðr ‘aces' fenced spaceʼ, etc.). A comparison of the designations of the world in the analyzed texts is quite justified, since only the “Elder” and “Younger Edda” represent a corpus of texts reflecting information about Scandinavian mythology as a system, therefore, in the absence of data from other Old Germanic areas, they become the only source for the reconstruction of German mythology, meaning which is difficult to overestimate. In addition, references to the “Elder Edda” constantly appear in the “Younger Edda”, therefore, the idea of comparison seemed relevant to its author himself, and experts have no right to ignore this circumstance. The research method is implemented in a scheme for describing an epic word, which arose on the basis of lexicological studies of a number of key concepts in the ancient Indian “Rigveda” and a thesaurus description of a folk word, which have undergone fundamental processing. Summing up the results of the comparative analysis of the designations of the world in the “Elder” and “Younger Eddas”, we can state a more archaic stage recorded in the Old Icelandic prose monument, manifested in the presence of nominations of the original loci of cosmogenesis (Mú-spellz-heimr ‘points (in space) destruction of the worldʼ, Nifl-heimr ‘fog worldʼ) and the place of their interaction (the World Abyss), the description of which as an object of the cosmic universe is given in sufficient detail; in the presentation of the opposition between one’s own and someone else’s space, fundamental to the mythopoetic model of the world (Mið-garðr ‘middle enclosed space’ – Út-garðr ‘outer enclosed space’); a more detailed and specific description of the corresponding lexemes and a wider range of their mythological connotations, as well as the factor of theocentrism, in which the reference point is the location of the dwelling of the gods (Ás-garðr ‘the enclosed space of the Asesʼ), for example, in relation to hel ‘another world, the afterlife’.

The function of Theme in Spanish

Abstract The main objective of this paper is to outline the system of theme in Spanish starting from its realisation in texts (Martin 2013). In order to do this, Halliday’s definition of Theme is revised and the distinction he proposes between the definition of a function and its means of realisation is followed (Halliday 1985, 1994). Then, a brief corpus of text instantiating different genres is analysed, considering different options of mood in the clause. The perspective assumed is language typology in systemic functional linguistics (Martin 1983; Caffarel et al. 2004a; Quiroz 2013; Martin et al. 2023), which highlights the peculiar characteristics of the language at stake based on patterns found in actual discourse and observes the phenomenon under study from above, from around and from below (Matthiessen and Halliday 2009).

UTILIZĂRI ALE PREPOZIȚIILOR LA, DE ȘI DE LA ÎN STRUCTURI DE GENITIV SAU DE DATIV DIN ROMÂNA NONSTANDARD

In the present article the author describes the use of the prepositions la, de, and de la in genitive or dative structures, encountered in the regional varieties of Romanian, emphasizing the frequency of the aforementioned structures and their specific features as shown in the corpus of dialectal texts investigated. The author also illustrates and discusses the different stages of grammaticalization of the preposition la in the expression of the genitive and dative and the connection between the genitive/dative relation and the locative relation.

Classification of human- and AI-generated texts for different languages and domains

Chatbots based on large language models (LLMs) like ChatGPT are available to the wide public. These tools can for instance be used by students to generate essays or whole theses from scratch or by rephrasing an existing text. But how does for instance a teacher know whether a text is written by a student or an AI? In this paper, we investigate perplexity, semantic, list lookup, document, error-based, readability, AI feedback and text vector features to classify human-generated and AI-generated texts from the educational domain as well as news articles. We analyze two scenarios: (1) The detection of text generated by AI from scratch, and (2) the detection of text rephrased by AI. Since we assumed that classification is more difficult when the AI has been prompted to create or rephrase the text in a way that a human would not recognize that it was generated or rephrased by an AI, we also investigate this advanced prompting scenario. To train, fine-tune and test the classifiers, we created the Multilingual Human-AI-Generated Text Corpus which contains human-generated, AI-generated and AI-rephrased texts from the educational domain in English, French, German, and Spanish and English texts from the news domain. We demonstrate that the same features can be used for the detection of AI-generated and AI-rephrased texts from the educational domain in all languages and the detection of AI-generated and AI-rephrased news texts. Our best systems significantly outperform GPTZero and ZeroGPT—state-of-the-art systems for the detection of AI-generated text. Our best text rephrasing detection system even outperforms GPTZero by 181.3% relative in F1-score.

Формування україномовної навчальної вибірки для визначення емоційного забарвлення текстів

Introduction. Every day, the number of news, pages on social networks and chats on the Internet is increasing, accordingly, there is an increase in information that carries an emotional load. At the same time, the number of information threats is also growing. Under such conditions, the construction of systems for determining the emotional color of texts becomes extremely relevant. Purpose. Emotional messages can be found and classified using artificial intelligence, namely based on neural network methods. For the process of learning neural networks, it is necessary to have a training sample of texts with a preliminary assessment of their emotional coloring. Such marked learning samples exist for news and texts in English, however, at the moment, no accessible learning sample of Ukrainian news and texts has been created. Methods. Using statistical methods of sentiment analysis for detecting text tonality with extended vocabulary. Results. Extended tonality vocabulary of the Ukrainian language was built. A large corpus of texts and their emotional coloring was built with an expertly assessed markup accuracy of 98%, containing 5,318,783 texts of various types in the Ukrainian language. Conclusion. The built text corpus can be used to train and test neural networks for sentiment analysis of Ukrainian-language texts.

Fine-Tuning Topic Modelling: A Coherence-Focused Analysis of Correlated Topic Models

The Correlated Topic Model (CTM) is a widely used approach for topic modelling that accounts for correlations among topics. This study investigates the effects of hyperparameter tuning on the model's ability to extract meaningful themes from a corpus of unstructured text. Key hyperparameters examined include learning rates (0.1, 0.01, 0.001), the number of topics (3, 5, 7, 10), and the number of top words (10, 20, 30, 40, 50, 80, 100). The Adam optimizer was used for model training, and performance was evaluated using the coherence score (c_v), a metric that assesses the interpretability and coherence of the generated topics. The dataset comprised 100 articles, and results were visualized using line plots and heatmaps to highlight performance trends. The highest coherence score of 0.803 was achieved with three topics and 10 top words. The findings demonstrate that fine-tuning hyperparameters significantly improves the model's ability to generate coherent and interpretable topics, resulting in more accurate and insightful outcomes. This research underscores the importance of parameter optimization in enhancing the effectiveness of CTM for topic modelling applications.

Una ucronía encarnada: Nación vacuna, de Fernanda García Lao

This article proposes a reading of the novel Nación Vacuna, which appeared in 2017 in Argentina and in February 2020 in Spain, as uchronia, given that it is a fiction that reimagines the future by turning to the past, specifically the so-called War of the Malvinas, which occurred in 1982, when the United Kingdom and Argentina clashed over the sovereignty of the islands of the same name. The author of the book, Fernanda García Lao, was born in Mendoza but has lived constantly jumping between her native Argentina and Spain. That is why, in the first part [Foreignness in Writing], some features of García Lao's literature are determined as the embodiment of her condition as a displaced writer, such as her use of metaphor, irony and the disregard for the rigid prescriptions of genres. In turn, the second, more extensive part [sections 2 to 4], focuses on the book and outlines three axes of analysis: first, the way in which it updates the uchronia; secondly, the choice of the narrative voice that imposes a certain perspective on what is narrated, as well as limited access to information; and, thirdly, the centrality of the body that is, also that of the textual corpus and the maxim that moves García Lao's writing: exploration, which gives such genuine vitality to her texts.

Let’s Discover More API Relations: A Large Language Model-Based AI Chain for Unsupervised API Relation Inference

APIs have intricate relations that can be described in text and represented as knowledge graphs to aid software engineering tasks. Existing relation extraction methods have limitations, such as limited API text corpus, and are affected by the characteristics of the input text. To address these limitations, we propose utilizing large language models (LLMs) (e.g., GPT-3.5) as a neural knowledge base for API relation inference. This approach leverages the entire Web used to pre-train LLMs as a knowledge base and is insensitive to the context and complexity of input texts. To ensure accurate inference, we design an AI chain consisting of three AI modules: API Fully Qualified Name (FQN) Parser, API Knowledge Extractor, and API Relation Decider. The accuracy of the API FQN Parser and API Relation Decider is 0.81 and 0.83, respectively. Using the generative capacity of the LLM and our approach’s inference capability, we achieve an average F1 value of 0.76 under the three datasets, significantly higher than the state-of-the-art method’s average F1 value of 0.40. Compared to the original CoT and modularized CoT methods, our AI chain design has improved the performance of API relation inference by 71% and 49%, respectively. Meanwhile, the prompt ensembling strategy enhances the performance of our approach by 32%. The API relations inferred by our method can be further organized into structured forms to provide support for other software engineering tasks.

Resepsi dan Transformasi Narasi Serat Menak Yasadipuran dalam Penciptaan Wayang Menak Gaya Surakarta

Serat Menak is a grand corpus of Javanese-Islamic text that is adopted from the Islamic epoch, Qissa-I Emir Hamza or Hikayat Amir Hamzah, which has widely spread in Asian region in the middle of second millenium CE. The popularity of Serat Menak made this text has many variants, and the one which is largely recognized was Yasadipura I (1729-1803)'s version which was composed in the beginning of Surakarta era. The text of Serat Menak Yasadipuran then become very well-known in Java, and became the inspiration for some branches of performing arts, such as wayang, whether in its kulit, golek, krucil, or wong formats. One of these wayang genres that took its repertoire from Menak tales was wayang kulit menak, which was introduced in Surakarta in the first quarter of 20th century, the creation of Sutarja Truna Dipa (STD) which has not being scientifically studied yet until this present day. This article aims to identify how far was the relation between STD's Surakarta wayang kulit menak with Serat Menak Yasadipuran, using Wolfgang Iser's Rezeption Aesthetic approach to obtain the reception and transformation patterns from textual narratives to visual forms. The result of this research concluded that STD's Surakarta wayang menak was a living artwork which evolving along times. On the other side, STD as the implied reader has been consistent in recepting Serat Menak Yasadipuran with some additions and adaptations, which was done to fulfil the needs of performance. Innovations done by STD to his wayang menak creation were not limited in the term of transformation from textual to visual media, but also filling the gaps (leerstellen) with the rich formal interpretation, which are completing the narrative of Serat Menak, so it will be more interesting to be played.

Structural analysis and intelligent classification of clinical trial eligibility criteria based on deep learning and medical text mining

Objective:To enhance the efficiency, quality, and innovation capability of clinical trials, this paper introduces a novel model called CTEC-AC (Clinical Trial Eligibility Criteria Automatic Classification), aimed at structuring clinical trial eligibility criteria into computationally explainable classifications. Methods:We obtained detailed information on the latest 2,500 clinical trials from ClinicalTrials.gov, generating over 20,000 eligibility criteria data entries. To enhance the expressiveness of these criteria, we integrated two powerful methods: ClinicalBERT and MetaMap. The resulting enhanced features were used as input for a hierarchical clustering algorithm. Post-processing included expert validation of the algorithm’s output to ensure the accuracy of the constructed annotated eligibility text corpus. Ultimately, our model was employed to automate the classification of eligibility criteria. Results:We identified 31 distinct categories to summarize the eligibility criteria written by clinical researchers and uncovered common themes in how these criteria are expressed. Using our automated classification model on a labeled dataset, we achieved a macro-average F1 score of 0.94. Conclusion:This work can automatically extract structured representations from unstructured eligibility criteria text, significantly advancing the informatization of clinical trials. This, in turn, can significantly enhance the intelligence of automated participant recruitment for clinical researchers.

Flexible Distribution Approaches to Enhance Regression and Deep Topic Modelling Techniques

ABSTRACTThis paper presents an extension of the Dirichlet multinomial regression (DMR) and deep Dirichlet multinomial regression (dDMR) topic modelling approaches by incorporating the generalised Dirichlet (GD) and Beta‐Liouville (BL) distributions using collapsed Gibbs sampling for parameter inference. The DMR and dDMR approaches have been shown to be effective in discovering latent topics in text corpora. However, these approaches have limitations when it comes to handling complex data structures and overfitting issues. To address these limitations, we introduce the GD and BL distributions, which have more flexibility in modelling complex data structures and handling sparse data. Additionally, we use collapsed Gibbs sampling to estimate the model parameters, which provides a computationally efficient method for inference. Experimental results on benchmark datasets demonstrate the effectiveness of the proposed approach in improving topic modelling performance, particularly in handling complex data structures and reducing overfitting. The proposed models also exhibit good interpretability of the learned topics, making them suitable for various applications in natural language processing and machine learning.

Categorising speakers’ language background: Theoretical assumptions and methodological challenges for learner corpus research

In this article, we investigate how speakers can be categorised based on their language background in the field of Learner Corpus Research (LCR). Specifically, we discuss three key aspects: first, the theoretical assumptions and methodological choices made in learner corpus design, second the integration of a holistic perspective for speaker categorisation in LCR and third the consequences that different categorisations might have on study outcomes. Through a comprehensive review of corpora used in the field, we identify the most common terms, definitions and criteria of categorisation used to describe a speaker's language background. Focusing on the most central metadata encoding language backgrounds, the L1 metadata, we inspect different operationalisations made and scrutinise the theoretical assumptions underlying them. Drawing on research on plurilingualism, we propose a holistic view of speaker's language background for Learner Corpus Research, combining various aspects of speaker's language use by methods inspired from the Dominant Language Constellation framework. We apply this methodology to re-evaluate the language categorisation system in LEONIDE, a multilingual corpus of Italian, German and English texts from secondary school students of diverse language backgrounds. We use the same corpus to evaluate the consequences of using different categorisations of the students on the outcome of possible linguistic studies. Despite a generally high overlap between study results across categorisations, we observe that variables combining multiple aspects of the speakers’ language backgrounds seem to explain group differences for more of the linguistic features investigated.

Subword Dictionary Learning and Segmentation for Expanding the Vocabulary of Automatic Speech Recognition in Tamil and Kannada

We present automatic speech recognition (ASR) systems for Tamil and Kannada based on subword modeling to effectively handle unlimited vocabulary arising due to the highly agglutinative nature of the languages. We propose a variant of the byte pair encoding (BPE) algorithm named extended-BPE, and Morfessor tool to segment each word into its subwords. We have effectively incorporated maximum likelihood and Viterbi estimation techniques with weighted finite state transducers framework in these algorithms to learn the subword dictionary from a large text corpus. Using the learned subword dictionary, the words in the transcriptions of the training data are segmented into subwords. We train deep neural network ASR systems that recognize the subword sequence for any given test speech utterance. The output subword sequence is then post-processed using deterministic rules to get the final word sequence. Because of this subword design, the number of words that can be recognized is much larger than the number of words in the training corpus. For Tamil ASR, we use 152 hours of data for training and 65 hours for testing, whereas for Kannada ASR, we use 275 hours for training and 72 hours for testing. Upon experimenting with different combinations of segmentation and estimation techniques, we find that the word error rate (WER) reduces drastically when compared to the baseline word-level ASR, achieving a maximum absolute WER reduction of 6.24% and 6.63% for Tamil and Kannada, respectively. Further, comparing our results with that of an end-to-end ASR model available on Github, we have seen that our subword language models can perform comparable to or exceed that of recent end-to-end ASR models for Kannada and Tamil languages.

Consistency Verification Method of Chinese and Russian Traditional Culture Translation Text Corpus Based on CNN-BiGRU

China and Russia have completely different cultural backgrounds, historical traditions and ways of thinking, which makes it difficult to directly correspond or accurately convey many culturally unique concepts, customs and expressions during the translation process. For example, some idioms, sayings or cultural symbols may have completely different meanings or no equivalents in the two cultures. Therefore, there are problems such as poor verification resulting in the consistency verification of Chinese and Russian traditional cultural external publicity translation text corpus. In order to solve this problem, this paper designs a consistency verification method for Chinese and Russian traditional cultural external publicity translation text corpus based on CNN-BiGRU. Through the part-of-speech correspondence of the Chinese and Russian traditional cultural external publicity translation text corpus, it is represented by vectors, the feature weights of the Chinese and Russian traditional cultural external publicity translation text corpus are calculated, the collocation information between the feature items is determined, and the maximum likelihood of the features is calculated using data smoothing technology. The rules are adjusted to extract the characteristics of the Chinese and Russian traditional cultural external publicity translation text corpus. Preprocess the key feature source documents of each Chinese and Russian traditional culture external publicity translation text corpus, determine the source documents and their corresponding key features of the traditional culture external publicity translation text corpus, and use TF-IDF and IDF to calculate the key feature sources of the corpus. Based on the rarity of document data, the BERT model is introduced as an encoder to determine the key features of the Chinese and Russian traditional cultural external publicity translation text corpus, integrating the CNN-BiGRU algorithm to design a consistency verification algorithm for Chinese and Russian traditional cultural external publicity translation text corpus. Set up the convolution layer, weight sharing layer, pooling layer, and BiGRU verification layer to build a CNN-BiGRU model to achieve consistency verification of the Chinese and Russian traditional cultural external publicity translation text corpus. Experimental results show that this method can improve the consistency verification effect of Chinese and Russian traditional cultural external publicity translation text corpus.

COVID-19 AND THE MENTAL HEALTH OF ELDERLY PEOPLE PARTICIPATING IN COMMUNITY CENTERS

This study focuses on the importance of elderly people's participation in Community Centers for mental health. In an attempt to understand the subject to be researched, as well as the themes interconnected with this process, including the discussion of concepts necessary for a better understanding of the object, it was necessary to read specific articles in the area of ​​gerontology, mental health, relevant legislation and COVID-19, in addition to topics such as community groups; the recent change that occurred in basic social protection, which was the reorganization of social assistance services; Community Service and Strengthening of Bonds; among others. This research aims to investigate the perceptions of the impact of COVID-19 on the mental health of elderly people participating in Community Centers. With a qualitative approach, based on the critical dialectical method, the research included an exploratory literature review and a semi-structured interview with users of eight Community Centers for elderly people, located in the city of Ponta Grossa/PR, totaling sixteen people interviewed of both sexes. To support the organization of the information, the IRAMUTEC software was used to create the textual corpus, construct the word cloud and the similarity tree. The analysis of the results followed the content analysis procedures proposed by Bardin (1977) and then the data were categorized, contemplating three chronological phases: 1) pre-analysis; 2) exploration of the material; 3) treatment of the results, inference and interpretation. Through the analysis of the data, it was possible to capture, with the results in the form of responses, how elderly people participating in the Community Centers had their mental health affected by the consequences left by the COVID-19 pandemic. The interviews were duly authorized by the elderly people, who were also previously informed about the objectives and procedures of the research. It was concluded that Community Centers for the elderly are an extremely important place, which can provide human emancipation, prevent isolation and social exclusion and is also an environment in which participants can interact with each other, share life experiences, in addition to achieving active and healthy aging.

Models for Implementing the Concept of “Absence”

The article examines typical verbalization models of the concept of “Absence” in the Ukrainian, Russian and English languages for their further comparison with the verbal implementation of this concept. There are several tasks which the article aims to achieve: 1) to identify common structures of conveying the concept of “Absence” in a text corpus of each selected language; 2) to build typical models based on the means of representation of the considered concept in the three languages; 3) to determine universal and original models in the three languages. There are a number of methods which are used to analyse the material under investigation: the continuous sampling method, the comparative method, etymologization in terms of the comparative-historical method, modelling within the structural method etc. We concluded that the languages under consideration have both universal and original typical models for implementing the concept of “Absence”. Their features depend on the structure of each specific language and the corresponding means of verbalization of different concepts. According to these models, the verbal implementation of the concept experiences its modification without violation of the identity of its meaning / sense. These models are a starting point for studying the implementation of the “Absence” concept in speech. Deviation from the models represents a violation of the norms within each language.

Reflection of local cultural tradition in oral childhood memories (based on the speech of speakers of Kuban subdialects)

The article examines the functional potential, subject matter, content, and linguistic specific features of oral stories-memories of native speakers of Kuban subdialects of their childhood years (the 30, 40, 50, 60, 70s of the twentieth century). The main corpus of texts was collected through the field method within the framework of dialectological expeditions (2005–2023). Textual analysis, thematic classification, and linguistic commentary on linguistic facts were used to analyse the language material. The study shows that the subject matter and content of old Kuban residents’ oral childhood memories reflect the local cultural tradition, which, while organically fitting into the national cultural context, has peculiarities determined by the specific features of the region as a Cossack land. We describe recurring plots which appear due to life circumstances common to this dialect community. The research has also established the functional potential of childhood memories in terms of identifying conceptual meanings of individual concepts. It is demonstrated that the Kuban dialect is an important component of the local cultural tradition. The storytellers learnt it in the childhood, and it remains a natural means of communication for most of them. The analysis of the linguistic features of the memories shows a consistent manifestation of dialectal features at all linguistic levels, which enables us to recognise the informants’ stories about their childhood as one of the effective ways of preserving the dialect. The conclusion notes the scientific potential of oral memories of childhood as valuable materials for multidimensional analysis of the childhood phenomenon, as well as for studying such a unique phenomenon as linguistic genetic memory which preserves dialect not only in space but also in time. Additionally, the article focuses on the didactic potential of the analysed texts. Familiarisation with such materials during Russian language, history, and library lessons, as well as homeroom periods will contribute to strengthening generational continuity. Moreover, it will enhance ethnic identity formation and involvement in the family history of people whose destinies shape the history of the country and the nation.

Identifying Persons of Interest in Digital Forensics Using NLP-Based AI

The field of digital forensics relies on expertise from multiple domains, including computer science, criminology, and law. It also relies on different toolsets and an analyst’s expertise to parse enormous amounts of user-generated data to find clues that help crack a case. This process of investigative analysis is often done manually. Artificial Intelligence (AI) can provide practical solutions to efficiently mine enormous amounts of data to find useful patterns that can be leveraged to investigate crimes. Natural Language Processing (NLP) is a subdomain of research under AI that deals with problems involving unstructured data, specifically language. The domain of NLP includes several tools to parse text, including topic modeling, pairwise correlation, word vector cosine distance measurement, and sentiment analysis. In this research, we propose a digital forensic investigative technique that uses an ensemble of NLP tools to identify a person of interest list based on a corpus of text. Our proposed method serves as a type of human feature reduction, where a total pool of suspects is filtered down to a short list of candidates who possess a higher correlation with the crime being investigated.

Dotless Arabic Text for Natural Language Processing

Abstract This article introduces a novel representation of Arabic text as an alternative approach for Arabic NLP, inspired by the dotless script of ancient Arabic. We explored this representation through extensive analysis on various text corpora, differing in size and domain, and tokenized using multiple tokenization techniques. Furthermore, we examined the information density of this representation and compared it with the standard dotted Arabic text using text entropy analysis. Utilizing parallel corpora, we also drew comparisons between Arabic and English text analysis to gain additional insights. Our investigation extended to various upstream and downstream NLP tasks, including language modeling, text classification, sequence labeling, and machine translation, examining the implications of both the representations. Specifically, we performed seven different downstream tasks using various tokenization schemes comparing the standard dotted text with dotless Arabic text representations. Performance using both the representations were comparable across different tokenizations. However, dotless representation achieves these results with significant reduction in vocabulary sizes, and in some scenarios showing reduction of up to 50%. Additionally, we present a system that restores dots to the dotless Arabic text. This system is useful for tasks that require Arabic texts as output.

Biting into the Bitter Reality: A Metaphorical Exploration of Food Disorders

This study investigated the metaphorical representation of eating disorders in a bilingual corpus of blogs. The analysis yielded three common metaphors: ED IS A JOURNEY; ED IS A WAR and ED IS A CONTAINER. Two language-specific metaphors emerged: ED IS A LIVING ENTITY in the Spanish subcorpus and ED IS A BODY in the English subcorpus. The journey and war metaphors demonstrated subtle linguistic nuances, with the Spanish speakers emphasizing the disease’s challenges and the English speakers emphasizing the prospects for recovery. The specific metaphors ED IS A LIVING ENTITY and ED IS A BODY highlighted cultural variations in agency, with the Spanish speakers portraying themselves as passive sufferers and the English speakers viewing themselves as active agents in the recovery process. This cultural divergence aligns with the distinct value orientations of American and Spanish cultures. While the limited text corpus precludes definitive conclusions, the findings align with previous research on conceptual metaphors in ED discourse. Further investigation is warranted to establish the figurative potential of different languages and determine whether either language employs more metaphors in the ED communicative domain.