Text Corpus Research Articles

Optimizing large language models in digestive disease: strategies and challenges to improve clinical outcomes.

Large Language Models (LLMs) are transformer-based neural networks with billions of parameters trained on very large text corpora from diverse sources. LLMs have the potential to improve healthcare due to their capability to parse complex concepts and generate context-based responses. The interest in LLMs has not spared digestive disease academics, who have mainly investigated foundational LLM accuracy, which ranges from 25% to 90% and is influenced by the lack of standardized rules to report methodologies and results for LLM-oriented research. In addition, a critical issue is the absence of a universally accepted definition of accuracy, varying from binary to scalar interpretations, often tied to grader expertise without reference to clinical guidelines. We address strategies and challenges to increase accuracy. In particular, LLMs can be infused with domain knowledge using Retrieval Augmented Generation (RAG) or Supervised Fine-Tuning (SFT) with reinforcement learning from human feedback (RLHF). RAG faces challenges with in-context window limits and accurate information retrieval from the provided context. SFT, a deeper adaptation method, is computationally demanding and requires specialized knowledge. LLMs may increase patient quality of care across the field of digestive diseases,where physicians are often engaged in screening, treatment and surveillance for a broad range of pathologies for which in-context learning or SFT with RLHF could improve clinical decision-making and patient outcomes. However, despite their potential, the safe deployment of LLMs in healthcare still needs to overcome hurdles in accuracy, suggesting a need for strategies that integrate human feedback with advanced model training.

ASPECTS OF SPORT METAPHORS TRANSFER

Sports terminology, as a general knowledge that covers our daily lives, is becoming increasingly popular in the world of linguistics. The use of sport metaphors is becoming more and more common - in mass media, in specialized texts, or in colloquial language. The sport metaphor, as a subject of this study, is the relationship of concepts in the field of sport with other conceptual areas. Our attempt to classify such metaphors in accordance with their transfer in various other areas has led to the achievement of a number of objectives: to study the problem in practice, to identify and analyze the most common sports metaphors, to reveal the fundaments on which the relationship between sport and other conceptual areas is based, to divide knowledge, organized around a sports concept that show our understanding of the world through exemplary models. The proposed models are considered as metaphors oriented by the type of sport in two categories – collective and individual sports, together with their subcategories. The studied textual corpus of the sports metaphor allows conclusions about their relationship with and impact on public speech. The increased use of sport metaphors can be explained by the fact that sport is an area that is attractive and emotionally explicit, intriguing the perceptions of the recipient (the mass reader and listener).

Searching COVID-19 Clinical Research Using Graph Queries: Algorithm Development and Validation.

Since the beginning of the COVID-19 pandemic, >1 million studies have been collected within the COVID-19 Open Research Dataset, a corpus of manuscripts created to accelerate research against the disease. Their related abstracts hold a wealth of information that remains largely unexplored and difficult to search due to its unstructured nature. Keyword-based search is the standard approach, which allows users to retrieve the documents of a corpus that contain (all or some of) the words in a target list. This type of search, however, does not provide visual support to the task and is not suited to expressing complex queries or compensating for missing specifications. This study aims to consider small graphs of concepts and exploit them for expressing graph searches over existing COVID-19-related literature, leveraging the increasing use of graphs to represent and query scientific knowledge and providing a user-friendly search and exploration experience. We considered the COVID-19 Open Research Dataset corpus and summarized its content by annotating the publications' abstracts using terms selected from the Unified Medical Language System and the Ontology of Coronavirus Infectious Disease. Then, we built a co-occurrence network that includes all relevant concepts mentioned in the corpus, establishing connections when their mutual information is relevant. A sophisticated graph query engine was built to allow the identification of the best matches of graph queries on the network. It also supports partial matches and suggests potential query completions using shortest paths. We built a large co-occurrence network, consisting of 128,249 entities and 47,198,965 relationships; the GRAPH-SEARCH interface allows users to explore the network by formulating or adapting graph queries; it produces a bibliography of publications, which are globally ranked; and each publication is further associated with the specific parts of the query that it explains, thereby allowing the user to understand each aspect of the matching. Our approach supports the process of query formulation and evidence search upon a large text corpus; it can be reapplied to any scientific domain where documents corpora and curated ontologies are made available.

Developing corpus literacy: A perspective of Latgalian language and cultural studies

ABSTRACT Corpus literacy has been described as the ability to use the technology of corpus linguistics to investigate language and enhance the language development of students. In recent decades, text and speech corpora for regional and lesser-used languages have increasingly been created, and they are considered an important resource of authentic texts for language and culture studies and learning. The Modern Latgalian Text Corpus (MuLa 2022) and the Modern Latgalian Speech Corpus (MuLaR) are, in this sense, available for researchers, learners and teachers of Latgalian. Researchers, teachers, and students as potential discoverers of language and culture often lack knowledge, skills and practice to work with language corpora. The purpose of this article is therefore to discuss the opportunities of using Latgalian language corpora in language and culture education. In this vein, we call for the development of corpus literacy of researchers and teachers as part of their professional development. In support of this call, we offer suggestions on how to facilitate students’ cultural awareness and foster their proficiency in the Latgalian language through language corpora. These recommendations aim to enrich both the research and teaching of language (socio-onomastics in a cultural context) and culture (folklore concept) with the help of corpus tools.

The Passive Voice in the Ancient Egyptian Pyramid Texts I

Summary The present article studies the ancient Egyptian Pyramid Texts passive strategy with an added .t or .tj suffix, the so-called T passive. It sets out to be an analysis of its uses and an interpretation of its meaning. A comprehensive list of data is also provided, along with tables analyzing the number of occurrences of certain syntactic structure, and the number of instances of the same verb using other passive strategies. A consideration of the number of occurrences of every passive form and their distinct uses within the different corpora of texts (Sacerdotal and Personal Texts) determines possible grammatical differences between both sources.

„Protestovali před a v budově řeckého ministerstva financí.“ Krátké nahlédnutí do problematiky zeugmatu

In this paper, we focus on one of the deviations from sentence structure, the zeugma, specifically its frequency in different types of texts. Existing linguistic publications treat usually zeugma as an error to be avoided. In addition to defining the term, we also draw attention to ellipsis and condensation, which are related to zeugma. The frequency of its occurrence in texts will be verified by corpus analysis; the corpus used will be the SYN2020 corpus of written Czech, the ORTOFON_v2 corpus of spoken, informal Czech, the ORATOR_v2 corpus of formal texts, and the web corpus, Araneum Bohemicum Maximum. The analyzed examples illustrate that zeugma occurs in different types of texts and, thus, is not bound to one particular mode. The prevalence of zeugma in professional texts suggests that this device does not affect the clarity and precision of expression that are typical for them. This finding could open up a debate as to whether to consider the zeugma as a standard stylistic device by which language users condense their utterances.

In search of Zion: reconsidering the political category of Zionist utopias

ABSTRACT This article focuses on the so-called Zionist utopias in order to reconsider these unprecedented Jewish utopian novels, which arose in the late nineteenth century, from a different perspective to the canonical one that formed within the Zionist historiography of the 1940s. To do so, it examines the literary textual corpus and includes other new utopian sources that have not been considered thus far. It then offers a brief reconstruction of the rediscovery of this utopian imaginary by Zionist scholars. Finally, this article proposes a new historiographical interpretation based on the disjunction between dreaming of Zion and founding Israel in order to deconstruct the underlying Zionist teleological view and thus reconsider these utopias in the long history of Jewish emancipation.

Advancing language models through domain knowledge integration: a comprehensive approach to training, evaluation, and optimization of social scientific neural word embeddings

This article proposes a comprehensive strategy for training, evaluating, and optimizing domain-specific word2vec-based word embeddings, using social science literature as an example. Our primary objectives are: (1) to train the embeddings utilizing a corpus of social science text, (2) to test their performance against domain-unspecific embeddings using our developed intrinsic and extrinsic evaluation strategy, and (3) to enhance their performance even further by using domain knowledge. As an integral part of this approach, we present SociRel-461, a domain-knowledge dictionary designed for the intrinsic evaluation and subsequent refinement of social science word embeddings. Using a dataset of 100,000 full-text scientific articles in sociology, we train multiple vector space models, which we then benchmark against a larger, pre-trained general language embedding model as part of our extrinsic evaluation. Furthermore, we developed a transfer learning multi-label classification task for extrinsic evaluation. Our findings reveal that domain-specific embeddings outperform their domain-unspecific counterparts in both intrinsic and extrinsic evaluations. We also investigated the retrofitting post-processing method to enhance domain-unspecific embeddings with the domain knowledge embedded in SociRel-461. While retrofitting does not enhance our domain-specific vector space models, it significantly improves the performance of the domain-unspecific embeddings. This highlights the potential of retrofitting for the transfer of domain knowledge to domain-unspecific embeddings. Our results emphasize the importance of utilizing domain-specific word embeddings for better performance in domain specific transfer learning tasks, as they outperform conventional embeddings trained on everyday language.

Abusive Language Detection in Khasi Social Media Comments

Abusive Language Detection in Khasi Social Media Comments

Exploring Semanticity for Content and Function Word Distinction in Catalan

In the realm of linguistics, the concept of “semanticity” was recently introduced as a novel measure designed to study linguistic networks. In a given text, semanticity is defined as the ratio of the potential number of meanings associated with a word to the number of different words with which it is linguistically linked. This concept provides a quantitative indicator that reflects a word’s semantic complexity and its role in a language. In this pilot study, we applied the semanticity measure to the Catalan language, aiming to investigate its effectiveness in automatically distinguishing content words from function words. For this purpose, the measure of semanticity has been applied to a large corpus of texts written in Catalan. We show that the semanticity of words allows us to classify the word classes existing in Catalan in a simple way so that both the semantic and syntactic capacity of each word within a language can be integrated under this parameter. By means of this semanticity measure, it has been observed that adverbs behave like function words in Catalan. This approach offers a quantitative and objective tool for researchers and linguists to gain insights into the structure and dynamics of languages, contributing to a deeper understanding of their underlying principles. The application of semanticity to Catalan is a promising pilot study, with potential applications in other languages, which will allow progress to be made in the field of theoretical linguistics and contribute to the development of automated linguistic tools.

Europe's Largest Research Infrastructure for Curated Medical Data Models with Semantic Annotations.

Structural metadata from the majority of clinical studies and routine health care systems is currently not yet available to the scientific community. To provide an overview of available contents in the Portal of Medical Data Models (MDM Portal). The MDM Portal is a registered European information infrastructure for research and health care, and its contents are curated and semantically annotated by medical experts. It enables users to search, view, discuss, and download existing medical data models. The most frequent keyword is "clinical trial" (n = 18,777), and the most frequent disease-specific keyword is "breast neoplasms" (n = 1,943). Most data items are available in English (n = 545,749) and German (n = 109,267). Manually curated semantic annotations are available for 805,308 elements (554,352 items, 58,101 item groups, and 192,855 code list items), which were derived from 25,257 data models. In total, 1,609,225 Unified Medical Language System (UMLS) codes have been assigned, with 66,373 unique UMLS codes. To our knowledge, the MDM Portal constitutes Europe's largest collection of medical data models with semantically annotated elements. As such, it can be used to increase compatibility of medical datasets and can be utilized as a large expert-annotated medical text corpus for natural language processing.

Analysis of Determinant Factors Customer Loyalty Towards Brand in The Telecommunication Industry With The Digitalization Paradigm

The goal of this research is to determine whether company parties have implemented engagement strategies in the current digital context. This research uses a qualitative marketing research approach. We used a digital platform for the initial data collection stage. The analysis then continued with the creation of a textual corpus. A deep understanding of consumer engagement with brands can help companies build stronger relationships with consumers, create greater value for brands, and maintain a competitive advantage in the marketplace. By paying attention to emotional, cognitive, and behavioral dimensions, companies can develop more holistic and effective strategies for managing consumer engagement and strengthening their brand positioning. Research results reveal that consumers actively seek knowledge and information to enhance their decision-making about brands and products. Mobile telecoms have well-implemented Customer Brand Engagement strategies, but they also need to have a proactive approach in dialogue with customers to maintain their engagement. This means that in addition to understanding and responding to consumer needs and desires, companies also need to be proactive in communicating with customers to strengthen their relationship and engagement with the brand. According to a careful analysis of the lexical units in cluster one, the majority come from Twitter. Cluster two highlights the comparisons that consumers make regarding the technical characteristics of products and services with those of various competitors. Meanwhile, cluster three shows that consumers are increasingly looking for personal services that can meet their needs. Cluster four warns management about loss phenomena that are specific to certain areas or regions. Finally, Cluster Five allows management to understand that consumers are currently less involved in innovative services such as payments via smartphone.

Exploring the evolution of research topics during the COVID-19 pandemic

The COVID-19 pandemic has changed the research agendas of most scientific communities, resulting in an overwhelming production of research articles in a variety of domains, including medicine, virology, epidemiology, economy, psychology, and so on. Several open-access corpora and literature hubs were established; among them, the COVID-19 Open Research Dataset (CORD-19) has systematically gathered scientific contributions for 2.5 years, by collecting and indexing over one million articles—this corpus, however, does not provide an easy-to-access overview of its content. Here, we present the CORD-19 Topic Visualizer (CORToViz), a method and associated visualization tool for inspecting the CORD-19 textual corpus of scientific abstracts. Our method is based upon a careful selection of up-to-date technologies (including large language models), resulting in an architecture for clustering articles along orthogonal dimensions and extraction techniques for temporal topic mining. Topic inspection is supported by an interactive dashboard, providing fast, one-click visualization of topic contents as word clouds and topic trends as time series, equipped with easy-to-drive statistical testing for analyzing the significance of topic emergence along arbitrarily selected time windows. Overall, our pipeline is very fast and its results match our expectations on topic identification (F1-score 0.854). The processes of data preparation and results visualization are completely general and virtually applicable to any corpus of textual documents—thus suited for effective adaptation to other contexts.

Anotacja socjolingwistyczna w Korpusie dawnych polskich tekstów dramatycznych (1772–1939)

SOCIOLINGUISTIC ANNOTATION IN THE CORPUS OF OLD POLISH DRAMATIC TEXTS (1772–1939) The article presents the criteria applied in the sociolinguistic annotation in the Corpus of old Polish dramatic texts (1772–1939) (KorTeDa). This level of annotation was about assigning the three basic sociolinguistic variables: age, sex and social status to all the utterances of drama characters (except for collective statements). Six values were selected for determining age (underage, young adult, middle- aged adult, older adult (senior), indeterminate adult, indeterminate); three for sex (female, male, indeterminate); and four for social status (low, medium, high, indeterminate). The article shows the interdependence between those factors and the historical, social and cultural context. Additionally, the authors illustrated substantial and technical problems they faced during the process of this type of annotation. KorTeDa may become an important tool for analysing the history of the Polish language taking into account the methods of corpus linguistics and sociopragmatics.

A new inscribed Aramaic potsherd from Tell al-Assara, Jordan

Recent fieldwork undertaken by the joint Finnish-Jordanian Tell Yaˁmoun Regional Archaeological Survey (TYRAS) in the northern part of the Jordanian plateau explored the unexcavated site of Tell al-Assara. An initial survey of the site and its surface pottery indicates that it was occupied for much of the 1st millennium BCE. The initial survey also found an inscribed Aramaic potsherd, which we present and analyse here. This analysis of the inscription builds both on digital techniques like photogrammetry and traditional philological approaches. Despite its brevity, the al-Assara inscription is an important new datum in the corpus of Transjordanian Iron Age texts and palaeography. It can be dated to the middle of the 1st millennium BCE and links Tell al-Assara to broader regional practices of writing and administration.

Visualization of educational data in a German-language corpus of student texts

Introduction. Digital transformation of education involves the use of digital tools in managing the educational process as a whole and in the learning process in individual disciplines. The specifics of teaching a foreign language require special functionality from digital tools. The purpose of the article is to explore the possibilities of a corpus of student texts in a foreign language for analyzing the results of students’ educational work and forming an adapted learning path within the discipline related to the study of a foreign language. Materials and methods. The material for the study was the Petrozavodsk Annotated Corpus of Texts (PACT), containing written works in German and French written by students from 2019 to 2023. Texts are accompanied by attributive information about the author, writing conditions, evaluation and error corrections. A web application has been developed to work with the corpus. The application includes personal accounts for teachers and students. To analyze the state of the learning process and make subsequent decisions on managing the learning process, the web application contains several tools for visualizing error statistics at different levels of detail with the ability to select texts according to different conditions. Results of the study. As part of the study, patterns were identified that connect the types of errors with the genres of texts, with the severity of the errors, and with the emotional and physiological state of the student. Russian-speaking students studying German as a foreign language make the most mistakes in the choice of lexemes, spelling, punctuation and the place of the verb in a sentence. At the same time, genre differences can be observed to distribute the number of errors by their type. There is an increase in the relative number of gross errors by 1.5 times for senior students compared to the first year. Errors in the choice of lexeme are critical for understanding the content of the text. The severity of the error does not depend on the genre of the text. Conclusion. The results of the study can be used to build corpora of student texts with educational analytics functions in dashboard format. The results obtained can be useful in developing teaching materials for the “Foreign Language” discipline.

Exploring corpus linguistics via computational tool analysis: key finding review

Corpus linguistics investigates language using extensive text databases. Tools assist researchers in analyzing, extracting, and interpreting linguistic information efficiently. Furthermore, if researchers only use traditional tools in corpus linguistic analysis, they will lack the comprehensiveness and efficiency required to effectively navigate and derive valuable insights from language data. This paper employed the preferred reporting items for systematic reviews and meta-analyses (PRISMA) approach to find the primary data based on a few keywords in corpus linguistic, corpus analysis, computational linguistic, text corpora and tool support. Based on this method, we used advanced searching techniques on Scopus and Web of Science (WoS) and discovered (N=28) data pertinent to the study. Expert scholars decide on a theme based on the problem, which is (i) types of corpus tools and their uses; (ii) their contributions and their capabilities (iii) limitations of corpus tools. All the tools were used in interdisciplinary studies. In summary, this systematic review uncovers pivotal key findings at the intersection of computational tools and corpus analysis, enriching linguistic knowledge. It highlights the interdisciplinary potential of corpus-based analysis in advancing linguistic tools and, their applications, as well as language analysis.

Semantic prosody, semantic transfer and semantic change

Abstract This article investigates semantic prosody in a diachronic perspective. Although prosodies have been shown to change over time, there is no consensus regarding the source of such changes. The present study explores this further through a corpus study of the development of the lemmas fabric, fabricate and fabrication from the late 15th century to the late 20th century, drawing on material from Early English Books Online, the Corpus of Late Modern English Texts and the British National Corpus. The results of the study show that prosodic changes coincide with the emergence of new senses and indicate that these processes are related to and possibly caused by semantic transfer induced by persistent prosodies over time.

Fast, Simple, and Accurate Time Series Analysis with Large Language Models: An Example of Mean-motion Resonances Identification

Classical machine learning has been actively utilized in astronomy to address various challenges, including predicting orbital stability, classifying asteroids, galaxies, and other objects, and analyzing images. However, the emerging trend in artificial intelligence involves the use of large language models such as GPT-4 and ChatGPT. These models are trained on a large corpus of text and can perform a wide range of natural language processing tasks, including text generation, translation, summarization, and classification. Surprisingly, these capabilities present significant potential for application in astronomy. This paper demonstrates how the new model gpt-4-vision-preview can analyze visual patterns and accurately classify asteroids as resonant or nonresonant with high accuracy. This process requires no training, fine-tuning, or coding beyond writing the appropriate prompt in natural language. Moreover, this approach can be extended to other common problems within astronomy.

Understanding lexico-semantic opposition empty/full in official business texts: Quantitative and qualitative research

This paper discusses the lexical-semantic opposition EMPTY/FULL in the corpus of legal, military and medical texts. The study analyzes 130 official business texts, in which we examined the problem of defining and constructing the lexical-semantic opposition EMPTY/FULL. The analysis promoted the construction of quantitative sampling, which demonstrates the weight of the lexemes in the corpus of official business texts. The purpose of the proposed analysis was to determine the distribution of the lexical-semantic opposition EMPTY/FULL, especially to find out the total number of occurrences, the average number of occurrences and the edge limits of the lexeme EMPTY and the lexeme FULL in the corpus of texts. The main results of the study outline the diversity of the lexical-semantic opposition EMPTY/FULL in different contexts, that underlines the functional diversification of the lexeme EMPTY and lexeme FULL. The results showed that in legal texts lexemes EMPTY/FULL nominate a legal concept; in medical texts they perform the function of medical conditions assessment; within the limits of military texts, they establish an associative connection between the nomination and semantic components (military actions, targets and objects of military use, consequences of war). Another important finding was the determination of the dominant lexeme in the corpus of legal texts and the dominant lexeme in the corpus of military texts. The similarities between EMPTY and FULL are primarily due to their antonymy. However, their differences are fixed in their distinct core meanings, contextual usage, pragmatic significance and connotations. These differences make them adaptable in various contexts of military, legal and medical texts to convey precise and often contrasting information about the state of objects and concepts.