Multiple Texts Research Articles

Food safety news events classification via a hierarchical transformer model

In light of the significance of regulatory authorities and the rising demand for information disclosure, a vast amount of information on food safety news reports is readily accessible on the Internet. The extraction of such information for precise classification and provision of appropriate safety alerts based on their respective categories has emerged as a challenging problem for academic research. Given that most food safety-related events in news reports comprise lengthy text, the pre-trained language models currently employed for text analysis are generally limited in their capability to handle long documents. This paper proposes a long-text classification model utilising hierarchical Transformers. We categorise information in long documents into two distinct types: (1) multiple text chunks meeting the length constraint and (2) essential sentences within long documents, such as headings, paragraph start and end sentences, etc. Initially, our proposed model utilises the text chunks as input to the BERT model. Then, it concatenates the output of the BERT model with the important sentences from the document and use them as input to the Transformer model for feature transformation. Finally, we utilise a classifier for food safety news classification. We conducted several comparative experiments with various commonly used text classification models on a dataset constructed from publicly available information on food regulatory websites. Our proposed method outperforms existing methods, establishing itself as the leading approach in terms of performance.

'Status Quo' on Different Aspects of Gender Distribution in Rheumatology in Germany-Results from a Nationwide Online Survey among Physicians.

Despite the increasing number of female medical students and fellows in Europe, women are still under-represented in higher academic careers and positions in medicine. The aim of this survey was to assess the 'status quo' on gender distribution among rheumatologists in Germany. A web-based anonymous survey (21 questions with multiple answers and free text) using QuestionPro® was distributed among rheumatologists in Germany via newsletters, social media and personal contact, including questions regarding hierarchical positions and work characteristics. Among the total of 170 respondents (72% women, 28% men, 1% diverse), 48% were rheumatologists in training, 35% were trained rheumatologists and 7% were heads of rheumatology departments. Regarding the gender ratio at different hierarchical levels, 74% of respondents reported more men than women in leadership positions. Part-time work was possible in the departments of 86% of respondents, with more women working part-time (56%) compared to men (29%). Most respondents stated their impression that employees working part-time did not have the same career chances as full-time workers in their departments. In total, 66% agreed that activities to improve gender equity are necessary. The highest need was seen in reconciling work and family through, e.g., part-time models, flexible childcare options at work and a higher acceptance of part-time work in leadership positions. According to our results, a gender imbalance is prevalent among rheumatologists in Germany, with lower numbers of women evident at higher hierarchical levels. Traditional role assignments are still represented by a higher proportion of part-time work in women. The establishment of structural changes to achieve better gender equity is needed.

Power and the Aerial Sublime in Victor Pelevin

AbstractThis essay distinguishes flight as a salient trope throughout multiple Pelevin texts: Omon Ra (1992), Chapaev and the Void (1996), Generation P (1999), Empire V (2006), and Love for Three Zuckerbrins (2014). It examines flight through the aesthetics of the sublime—classical, (post)‐Soviet, and postmodern. The “aerial sublime” focuses on the motifs of flight, elevation, and their reversal such as descent and going underground. Through this analysis, the essay offers an entry point to the question of Pelevin’s preoccupation with the subject’s relation to power—political, military, and capital‐, media‐, or technology‐driven. The airborne protagonist is positioned in spatial counterpoint to tyrannical social and cosmic forces. One may remain in their clutches or escape, withdraw, comically yield, or pragmatically acquiesce. Pelevin’s aerial sublime grows increasingly bleak: from an internalized ethical experience to active thought work without resolution, touching the noumenal, or ethical to the apotheosis of the non‐reasoning, non‐ethical, and inhuman. That is, his later works envision the sublime transformed into an anti‐sublime through power annihilating the human as a sentient and ethical agent. Pelevin’s aerial sublime thereby conveys lessons about ironclad social and cosmic mechanics and the subject that is constituted, coerced, and occasionally resistant to it.

Особенности формирования умений младших школьников по работе с информацией на основе использования множественных текстов на уроках математики

The study aims to substantiate the need to form the skills of working with information based on the use of multiple texts in mathematics classes among younger schoolchildren. The paper discusses the definitions of the notion of “information”, outlines the specifics of working with tables, diagrams, pictograms. Among the types of information, special emphasis is given to the multiple text, since it serves as a tool for reading literacy assessment. The paper presents the results of an empirical study concerning the level of formation of the skills of working with information among 3rd-grade students according to three criteria: the cognitive criterion, the activity criterion and the evaluation criterion. The main difficulties of younger schoolchildren in the formation of information skills are described. The scientific novelty of the work consists in determining the criteria for the formation of the skills of working with information based on the use of multiple texts among younger schoolchildren and in describing the level characteristics of these criteria. As a result of the study, it has been determined that the multiple text can act as a means of forming the skills of working with information among younger schoolchildren, the difficulties encountered by younger schoolchildren when working with information have been indicated.

Perceiving Multiple Representations for scene text image super-resolution guided by text recognizer

Single image super-resolution (SISR) aims to recover clear high-resolution images from low-resolution images, which has made great progress with the development of deep learning these years. Scene text image super-resolution (STISR) is a subfield of SISR with the goal of increasing the resolution of a low-resolution text image and enhancing the readability of characters in the image. Despite significant improvements in recent approaches, STISR remains a challenging task due to the diversity of background, text appearances and layouts, etc. This paper presents a Perceiving Multiple Representations (PerMR) method for better super-resolution performances in scene text images. PerMR is a unified network that combines super-resolution with text recognition and exploits the recognizer’s feedback to facilitate super-resolution. Specifically, contextual information from the text decoder is extracted to provide sequence-specific guidance and enable the super-resolution model to pay more attention to the text region. Meanwhile, low-level and high-level visual features from the vision backbone of the recognition network are integrated to further improve visual quality. Additionally, we incorporate a frequency branch into the vanilla convolution unit, which efficiently enhances global and local feature representations. Experiments on the STISR benchmark dataset TextZoom validate that PerMR can not only generate more distinguishable images, but also outperforms the current state-of-the-art methods. PerMR boosts the average recognition accuracy by 5.9% using ASTER, 5.8% using MORAN and 10.6% using CRNN compared to the baseline model TSRN. PerMR outperforms the advanced method TPGSR-3 by 1.4% on ASTER, 0.1% on MORAN, 0.2% on CRNN and boosts TATT by 0.6% on ASTER and 1.1% on MORAN respectively. Furthermore, PerMR demonstrates good robustness and generalization when tackling low-quality text images in multiple scene text recognition datasets. The experiment results verify the capabilities of PerMR to boost text recognition performance.

A Function Area Division Approach for Autonomous Transportation System Based on Text Similarity

Along with emerging technologies and increasing demands, autonomation has become a significant trend in current transportation systems. Within this context, the autonomous transportation system (ATS) framework hinges on functions that serve as fundamental units to support its operation. Recognizing the divisions among these function areas can enhance our understanding of their meanings and interrelationships. This study introduces a method for dividing function areas within the ATS framework, grounded in text similarity, to mitigate reliance on subjective experience. Precisely, this method quantifies the similarity between functions based on their textual descriptions, and implements hierarchical clustering to delineate them into distinct function areas. To validate the effectiveness of this proposed method, a case study analyzing a vehicle automatic driving scenario was conducted. The results demonstrate that our approach can efficiently divide function areas, producing clustering outcomes that possess superior accuracy and purity when juxtaposed with reference classifications. Consequently, this method has the potential to facilitate the formulation of function areas within ATS, thereby supporting the autonomous operation and construction of ATS. Moreover, its applicability extends beyond ATS, showing promise for other clustering problems that involve multiple texts, such as in text classification.

Multiple Interpretation and Fragmented Texts Within a Historical Corpus: The Case of Old East Slavic Vernacular Writing

Abstract The paper presents the issue of fragmented and/or ambiguously interpreted texts within the corpora of Old East Slavic vernacular writing. One of these corpora, the corpus of the Old East Slavic birchbark letters, is already available, the other, comprising the texts of Old East Slavic inscriptions, is under preparation. Due to the fragmentary state of many birchbark and epigraphy texts, their lemmatization and grammatical tagging may be uncertain and multiple interpretations may coexist. Some lemmas survive only in fragments which are nevertheless relevant for the study of lexicon. The grammatical status of many fragments may be firmly established despite lacking lexical information. However the relevant data on these fragments is not available in the word indices and corpora that take into consideration only best-preserved word forms. In the paper, the representation and annotation of such word forms within the Old East Slavic vernacular corpora is presented, and relative frequencies of such phenomena within the birchbark letter corpus are shown, with some case studies showing the relevance of the annotation of fragmented forms. The existing approaches, namely for the classical epigraphy within the EpiDoc standard and in the Hittite syntactic treebanks, are also briefly presented and compared to the solution found within the Old East Slavic vernacular corpora.

Social justice reasoning when students learn about social issues using multiple texts

ABSTRACT In this paper, I apply the Multiple Documents Text-Based Relevance Assessment and Content Extraction (MD-TRACE) model, to describe the types of cognitive processes that students engage to critically reason about social issues, as they are portrayed through mass media. In addition to examining such processes, I further consider the extent to which these are reflective of social justice reasoning, or students’ critical reasoning about social issues in ways that recognize and analyze inequities in society. Three studies are introduced to provide empirical examples of how cognitive processes, identified in the MD-TRACE, may function within the context of students’ reasoning about mass media. The processes examined include selection (Study 1), processing (Study 2), and integration (Study 3). Study 1 examines the types of perspectives that students propose seeking out in association with various social issues and the extent to which these perspectives may belong to marginalized groups (i.e., selection). Study 2 examines students’ critical reasoning about or abilities to critique two deliberately constructed texts during processing. Study 3 investigates students’ specific abilities to identify and critique narrative frames, or common reporting tropes, introduced across texts (i.e., integration). Together, these three studies serve as exemplars of students’ engagement in reasoning about mass media and social justice reasoning. They suggest that social justice reasoning involves, in part, students’ engagement in perspective taking, their application of prior knowledge to contextualize information in texts, and their recognition of common narrative frames across texts and the values that these uphold. Additional social justice reasoning strategies are suggested and directions for future research proposed.

Enhancing Answer Selection via Ad-Hoc Knowledge Extraction from Unstructured Web Texts

Answer selection aims to identify the most relevant answers to a given question from a set of candidates. It is the fundamental component of intelligent question answering system. To improve performance, it gradually becomes an effective strategy to integrate external structured knowledge bases (KBs) into the answer selection model. Due to expensive cost of construction and maintenance of such KBs, these models are suffering from domain barriers and information incompleteness. In this paper, we propose a two-stage extraction–comprehension answer selection model, which can extract ad-hoc knowledge from unstructured web texts to enhance the performance of answer selection. For the extraction, two types of snippets are extracted from unstructured web pages and utilized as the source of ad-hoc knowledge. For the comprehension, a selective attention mechanism is employed to extract and integrate ad-hoc knowledge from multiple text snippets obtained in the first stage, which can enrich the representation of question–answer pairs and more accurately identify the correct answers. By incorporating ad-hoc knowledge extracted from both types of snippets, the proposed model achieves state-of-the-art results on two public available benchmark datasets. In particular, on WikiQA, in terms of the two evaluation metrics (mean average precision and mean reciprocal rank), it achieves 9.9[Formula: see text] and 8.4[Formula: see text] higher than the previous non-pretraining-based models, and 3.4[Formula: see text] and 3.2[Formula: see text] higher than the pretraining-based models.

Enhancing Knowledge Management in Healthcare: An Embedding Fusion Approach to Business Rule Representation

Enterprise Decision Management systems are vital for delivering efficient healthcare services. However, the ever-changing clinical terminology creates complex business rules, making healthcare IT systems difficult to maintain. In this study, we present an embedding fusion technique using unsupervised Natural Language Processing (NLP) to represent business rules as semantic vectors by incorporating multiple text data sources for each rule. We apply this method to a dental insurance administration case study and find that our approach is over 200 times more likely to identify redundant rule pairs compared to random pairs. This case study suggests that an embedding-based technique can significantly improve knowledge management efficiency in healthcare IT systems.

Practice patterns in revision dacryocystorhinostomy

ABSTRACT Purpose This study aimed to report the practice patterns while performing revision dacryocystorhinostomy (RevDCR) amongst oculoplastic surgeons from several regions across the globe. Methods The survey consisted of 41 specific questions sent via email that contained a link to the Google forms. The questions covered several aspects of the respondents’ practice profiles, evaluation methods, pre-operative choices, surgical techniques, and follow-up preferences while dealing with patients of prior failed DCRs. Questions were answerable either as multiple choice or free text typing. The survey respondents were anonymized. The responses were collected and analyzed, and data were tabulated to understand the preferred practice trends. Results A total of 137 surgeons completed the survey. Most respondents identified themselves as experienced surgeons managing failed DCRs (76.6%, total respondents (n) = 137). The commonly preferred modalities for evaluation of a failed DCR were lacrimal irrigation (91.2%) and nasal endoscopy (66.9%). About 64% (87/137) of the respondents performed a combination of nasal endoscopy, lacrimal irrigation, and probing to localize the area of the failed DCR. A majority of the respondents used anti-metabolites (73.3%, n = 131) and stents (96.4%, 132/137) during the revision surgery. Most surgeons preferred endoscopic approach (44.5%), (61/137) for revising a failed DCR and most preferred general anaesthesia with local infiltration (70.1%, 96/137). Aggressive fibrosis with cicatricial closure was identified as the most frequently encountered cause of failure (84.6%, 115/137). The osteotomy was performed on an as-needed basis by 59.1% (81/137) of the surgeons. Only 10.9% of the respondents used navigation guidance while performing a revision DCR, mainly for post-trauma scenarios. Most of the surgeons completed the revision procedure within 30–60 minutes (77.4%, 106/137). The self-reported outcomes of revision DCRs were good (80–95%, median-90%, n = 137). Conclusion A significantly high percentage of oculoplastic surgeons who responded to this survey from across the globe performed nasal endoscopy in their pre-operative evaluations, preferred endoscopic surgical approach, and used antimetabolites and stents while performing revision DCRs.

Investigating EFL Students Reading Preferences (Digital or Printed)

Reading is crucial to the process of learning a language. Students can use digital technology to support reading activities in today's modern culture. As a result of students' need for information in the digital age, digital text is now virtually everywhere. This study investigated EFL students reading preferences (digital or printed). The participants of this study were 102 EFL students from private universities in South Sumatera. The data was collected through 20 items of questionnaire distributed through google form. Based on the result of the study, it could be concluded that first, most participants enjoy reading. They read for a variety of purposes, including to get the knowledge, to get information, to get the entertainment, and for other purposes. The second, the participants prefer reading digital texts. The practicality of the digital texts is the primary motivating factor why they chose it. They acknowledged that they do not need to hold thick texts or books while reading, they only need to bring their devices. Furthermore, they said one device can store and read multiple texts. They added that reading materials on digital devices are considerably more intriguing and enjoyable than those on printed. Third, the researchers noted that all readers—whether they preferred reading printed or digital texts—felt at comfortable with the reading equipment they had selected. . Finally, it is revealed from the data that most participants are more inspired to read when they use digital reading texts.

Metacomprehension in Multiple Document Reading

Abstract: Accurately assessing someone’s comprehension is central to learning from multiple documents. This study examines whether judgment accuracy can be increased if readers summarize texts before assessing their own comprehension. We asked 144 adults to read multiple documents on a historical controversy. They then produced Twitter-like summaries either (a) after a time delay or (b) immediately after reading; a control group (c) did not write summaries. Results showed that delaying writing summaries increased the absolute accuracy of self-assessments of undisputed historical content knowledge. We found negative effects for the accuracy of self-assessments of source–content links, intertextual conflicts, and integration performance in an essay. The relative accuracy was lower in the delayed summaries group than in the controls. In summary, the positive effect of writing summaries appears to be limited to the self-assessment of content knowledge, whereas it is reduced with respect to forms of knowledge specific to reading multiple texts.

Save Our Ships SOS-AN Android Application for Safety

Abstract: In this paper, android applications are named SOS for smartphones and web applications, which are developed by java and XML using android studio. The main aim of SOS is to save lives with a single tap. It sends multiple text messages and emails while pressing the SOS button. SOS uses user interface and user experience design by a Machine Learning (ML) algorithms such as Application programming interface(API) to understand SQL queries and it provides a perfect response. Through the Dijkstra algorithm, this application employs Google Maps to trace the nearest way to an available time. SOS has a high level of accuracy in understanding user’s needs and providing a response on time. SOS provides safety for children and girls without parental care.

Efficient Memory-Enhanced Transformer for Long-Document Summarization in Low-Resource Regimes.

Long document summarization poses obstacles to current generative transformer-based models because of the broad context to process and understand. Indeed, detecting long-range dependencies is still challenging for today's state-of-the-art solutions, usually requiring model expansion at the cost of an unsustainable demand for computing and memory capacities. This paper introduces Emma, a novel efficient memory-enhanced transformer-based architecture. By segmenting a lengthy input into multiple text fragments, our model stores and compares the current chunk with previous ones, gaining the capability to read and comprehend the entire context over the whole document with a fixed amount of GPU memory. This method enables the model to deal with theoretically infinitely long documents, using less than 18 and 13 GB of memory for training and inference, respectively. We conducted extensive performance analyses and demonstrate that Emma achieved competitive results on two datasets of different domains while consuming significantly less GPU memory than competitors do, even in low-resource settings.

Content and quantity of highlights and annotations predict learning from multiple digital texts

Learning from multiple texts is an essential yet challenging academic task. Thus, a persistent need in the field is identifying those strategies that will support students' learning from multiple texts, and facilitating their use. In this study, we adopt a quasi-experimental design to examine how the use of digital annotations, allowing undergraduates (N = 278) to identify and, potentially, link important concepts across texts, can be used to foster learning from multiple texts. Additionally, we examine students' digital highlights and annotations under four different strategy conditions, assigned to elicit different forms of processing found to be effective in prior work. These conditions included asking students to highlight and annotate (a) important or relevant information in texts, (b) connections or links across texts (i.e., intertextual reasoning condition), (c) information in texts facilitating evaluation, and (d) difficult to understand information (i.e., metacognitive monitoring condition). Some differences were found in students' rendering of highlights and annotations across strategy conditions, however, no differences in multiple text comprehension and integration performance were found across strategy conditions. Still, membership in the metacognitive monitoring strategy condition, relative to the relevance condition, as well as students’ highlighting of information supporting evaluation (i.e., source information, statistical evidence), and accessing of references during reading were predictive of multiple text comprehension and integration. Thus, this study contributes to theory by providing further evidence of strategies (e.g., metacognitive monitoring, evaluation) supportive of multiple text learning and introduces a promising means whereby such strategy use may be fostered.

To Clarity and Beyond: Situating Higher-Order, Critical, and Critical-Analytic Thinking in the Literature on Learning from Multiple Texts

To Clarity and Beyond: Situating Higher-Order, Critical, and Critical-Analytic Thinking in the Literature on Learning from Multiple Texts

Combining a Direct and Indirect Training Approach for Cross-Domain Competences: The Case of the Course “Pedagogy for Psychotherapists”

In psychology programs, students should acquire both domain-specific knowledge and cross-domain competences important for later practice (e.g., multiple document literacy). Typically, such competences are trained directly in courses explicitly devoted to them or indirectly in courses on psychological topics that require them without systematically teaching and training them (e.g., when reading multiple texts about a topic). To exploit the advantages of both direct and indirect training approaches, we combined them in a new psychology course in which students were taught domain-specific knowledge on pedagogy and psychotherapy, and they were trained in the competence facets of (a) epistemic beliefs, (b) multiple document literacy, and (c) argumentative thinking. The direct training took a tried-and-tested example-based learning approach. A central element of the indirect training consisted of course assessments requiring the application of these three competences to the contents about pedagogy and psychotherapy. The combined training approach led to significant increases in declarative knowledge, advanced epistemic beliefs, and greater self-efficacy in implementing strategies relating to multiple document literacy and argumentative thinking. This approach can be adapted to accommodate different psychological content areas or different cross-domain competences.

A Survey of Implicit Discourse Relation Recognition

A discourse containing one or more sentences describes daily issues and events for people to communicate their thoughts and opinions. As sentences are normally consist of multiple text segments, correct understanding of the theme of a discourse should take into consideration of the relations in between text segments. Although sometimes a connective exists in raw texts for conveying relations, it is more often the cases that no connective exists in between two text segments but some implicit relation does exist in between them. The task of implicit discourse relation recognition (IDRR) is to detect implicit relation and classify its sense between two text segments without a connective. Indeed, the IDRR task is important to diverse downstream natural language processing tasks, such as text summarization, machine translation and so on. This article provides a comprehensive and up-to-date survey for the IDRR task. We first summarize the task definition and data sources widely used in the field. We categorize the main solution approaches for the IDRR task from the viewpoint of its development history. In each solution category, we present and analyze the most representative methods, including their origins, ideas, strengths and weaknesses. We also present performance comparisons for those solutions experimented on a public corpus with standard data processing procedures. Finally, we discuss future research directions for discourse relation analysis.

Improving text mining in plant health domain with GAN and/or pre-trained language model.

The Bidirectional Encoder Representations from Transformers (BERT) architecture offers a cutting-edge approach to Natural Language Processing. It involves two steps: 1) pre-training a language model to extract contextualized features and 2) fine-tuning for specific downstream tasks. Although pre-trained language models (PLMs) have been successful in various text-mining applications, challenges remain, particularly in areas with limited labeled data such as plant health hazard detection from individuals' observations. To address this challenge, we propose to combine GAN-BERT, a model that extends the fine-tuning process with unlabeled data through a Generative Adversarial Network (GAN), with ChouBERT, a domain-specific PLM. Our results show that GAN-BERT outperforms traditional fine-tuning in multiple text classification tasks. In this paper, we examine the impact of further pre-training on the GAN-BERT model. We experiment with different hyper parameters to determine the best combination of models and fine-tuning parameters. Our findings suggest that the combination of GAN and ChouBERT can enhance the generalizability of the text classifier but may also lead to increased instability during training. Finally, we provide recommendations to mitigate these instabilities.