Word Co-occurrence Research Articles

Sparse Structure Learning via Graph Neural Networks for Inductive Document Classification

Recently, graph neural networks (GNNs) have been widely used for document classification. However, most existing methods are based on static word co-occurrence graphs without sentence-level information, which poses three challenges:(1) word ambiguity, (2) word synonymity, and (3) dynamic contextual dependency. To address these challenges, we propose a novel GNN-based sparse structure learning model for inductive document classification. Specifically, a document-level graph is initially generated by a disjoint union of sentence-level word co-occurrence graphs. Our model collects a set of trainable edges connecting disjoint words between sentences, and employs structure learning to sparsely select edges with dynamic contextual dependencies. Graphs with sparse structure can jointly exploit local and global contextual information in documents through GNNs. For inductive learning, the refined document graph is further fed into a general readout function for graph-level classification and optimization in an end-to-end manner. Extensive experiments on several real-world datasets demonstrate that the proposed model outperforms most state-of-the-art results, and reveal the necessity to learn sparse structures for each document.

PaintTeR: Automatic Extraction of Text Spans for Generating Art-Centered Questions

We propose PaintTeR, our Paintings TextRank algorithm for extracting art-related text spans from passages on paintings. PaintTeR combines a lexicon of painting words curated automatically through distant supervision with random walks on a large-scale word co-occurrence graph for ranking passage spans for artistic characteristics. The spans extracted with PaintTeR are used in state-of-the-art Question Generation and Reading Comprehension models for designing an interactive aid that enables gallery and museum visitors focus on the artistic elements of paintings. We provide experiments on two datasets of expert-written passages on paintings to showcase the effectiveness of PaintTeR. Evaluations by both gallery experts as well as crowdworkers indicate that our proposed algorithm can be used to select relevant and interesting art-centered questions. To the best of our knowledge, ours is the first work to effectively fine-tune question generation models using minimal supervision for a low-resource, specialized context such as gallery visits.

Lexical semantics enhanced neural word embeddings

Current breakthroughs in natural language processing have benefited dramatically from- neural language models, through which distributional semantics can leverage neural data representations to facilitate downstream applications. Since neural embeddings use context prediction on word co-occurrences to yield dense vectors, they are inevitably prone to capture more semantic association than semantic similarity. To improve vector space models in deriving semantic similarity, we post-process neural word embeddings through deep metric learning, through which we can inject lexical-semantic relations, including syn/antonymy and hypo/hypernymy, into a distributional space. We introduce hierarchy-fitting, a novel semantic specialization approach to modelling semantic similarity nuances inherently stored in the IS-A hierarchies. Hierarchy-fitting attains state-of-the-art results on the common- and rare-word benchmark datasets for deriving semantic similarity from neural word embeddings. It also incorporates an asymmetric distance function to specialize hypernymy’s directionality explicitly, through which it significantly improves vanilla embeddings in multiple evaluation tasks of detecting hypernymy and directionality without negative impacts on semantic similarity judgement. The results demonstrate the efficacy of hierarchy-fitting in specializing neural embeddings with semantic relations in late fusion, potentially expanding its applicability to aggregating heterogeneous data and various knowledge resources for learning multimodal semantic spaces.

Nineteen Years of Emerging Markets Finance and Trade: A Bibliometric Analysis

ABSTRACT This study draws a bibliographic profile of Emerging Markets Finance and Trade (EMFT) from bibliographic metadata, and provides a comprehensive overview of EMFT. We used scientometric techniques to analyze the publication and citation trends of EMFT. EMFT published 2049 cutting-edge documents between 2002 and 2020, with 12674 citations, which shows its influence in the academic field. The descriptive bibliometric analysis shows the publication and citation trends and the most prolific authors of EMFT. This study also presents the bibliographic coupling of authors, their affiliated institutions and countries, the co-citation of journals, and the co-occurrence of words from different sources. The study, with the help of network analysis, also identifies the major themes discussed in EMFT, develops seven clusters of EMFT publications, and analyzes their content.

Actividades musicales en escuelas infantiles de Hong Kong: Análisis de contenido de los informes de Revisión de Calidad

In Hong Kong, the Education Bureau (EDB) assesses the quality of services provided to children in local kindergartens. Quality Review (QR) reports of kindergartens that pass the assessment are published on EDB’s website. We conducted a content analysis of 164 QR reports to examine the alignment between the music activities alluded to and the curriculum objectives established for music in official policies. A coding scheme was developed using both inductive and deductive approaches. High inter-reliability was obtained. MAXQDA was used to conduct word frequency, descriptive, and co-occurrence analyses. The most common terms identified in the music-related segments focused on children’s development of sensory abilities through music experiences, in relation to singing, rhythm, beat, movement, and instrumental music. However, activities intended to foster musical creativity and self-expression were seldom mentioned. We conclude that the QR reports reveal important discrepancies between official curriculum policies and actual classroom practices, which EDB assessors seemed to ignore or overlook. Implications focus on the need for kindergarten stakeholders to address curriculum/practice gaps and further prepare teachers to foster children’s musical creativity.

Re-organise: Game-Based Learning of Circular Business Model Innovation

This study furthers game-based learning for circular business model innovation (CBMI), the complex, dynamic process of designing business models according to the circular economy principles. The study explores how game-play in an educational setting affects learning progress on the level of business model elements and from the perspective of six learning categories. We experimented with two student groups using our game education package Re-Organise. All students first studied a reader and a game role description and then filled out a circular business model canvas and a learning reflection. The first group, i.e., the game group, updated the canvas and the reflection in an interactive tutorial after gameplay. The control group submitted their updated canvas and reflection directly after the interactive tutorial without playing the game. The results were analyzed using text-mining and qualitative methods such as word co-occurrence and sentiment polarity. The game group created richer business models (using more waste processing technologies) and reflections with stronger sentiments toward the learning experience. Our detailed study results (i.e., per business model element and learning category) enhance understanding of game-based learning for circular business model innovation while providing directions for improving serious games and accompanying educational packages.

The Global Research Trends on the Growth of Remote Learning in Higher Education Institutions: A Bibliometric Analysis

Remote learning has rapidly emerged as one of the most fundamental approaches in facilitating self-directed learning in Higher Learning Institutions. The purpose of this paper was to conduct a bibliometric study on the global trends in research about remote learning during 1961 to 2021 period. A bibliometric analysis of 719 research articles from the Dimensions database was carried out. The VOSviewer1.6.16 software was used to carry out visual analysis of the publications to establish the network and overlay visualization maps. The results indicate that there has been a steady growth of research articles on remote learning during COVID-19 pandemic of which ‘Polytechnic Institute of Porto’ was the leading organization in terms of publications. Countries which are most productive include the United States, Russia and the United Kingdom while ‘Corter’ and ‘Kreijns’ were the most cited authors and ‘Lima’ and ‘Viegas’ were the authors with highest links in terms of collaboration. Regarding the co-occurrence of key words, it was found that keywords such as ‘experience’, ‘COVID’, ‘internet’ ‘poverty’ ‘interaction’, ‘inequality’, ‘social emotions’ and ‘online interactions’ had strong association with remote learning. The findings suggest the need for developing countries to allocate adequate funds on remote learning.

Combining shallow and deep learning approaches against data scarcity in legal domains

We are recently witnessing a radical shift towards digitisation in many aspects of our daily life, including law, public administration and governance. This has sometimes been done with the aim of reducing costs and human errors by improving data analysis and management, but not without raising major technological challenges. One of these challenges is certainly the need to cope with relatively small amounts of data, without sacrificing performance. Indeed, cutting-edge approaches to (natural) language processing and understanding are often data-hungry, especially those based on deep learning. With this paper we seek to address the problem of data scarcity in automatic Legalese (or legal English) processing and understanding. What we propose is an ensemble of shallow and deep learning techniques called SyntagmTuner, designed to combine the accuracy of deep learning with the ability of shallow learning to work with little data. Our contribution is based on the assumption that Legalese differs from its spoken language in the way the meaning is encoded by the structure of the text and the co-occurrence of words. As result, we show with SyntagmTuner how we can perform important tasks for e-governance, as multi-label classification of the United Nations General Assembly (UNGA) Resolutions or legal question answering, with data-sets of roughly 100 samples or even less.

Contrastive Graph Convolutional Networks with adaptive augmentation for text classification

Text classification is an important research topic in natural language processing (NLP), and Graph Neural Networks (GNNs) have recently been applied in this task. However, in existing graph-based models, text graphs constructed by rules are not real graph data and introduce massive noise. More importantly, for fixed corpus-level graph structure, these models cannot sufficiently exploit the labeled and unlabeled information of nodes. Meanwhile, contrastive learning has been developed as an effective method in graph domain to fully utilize the information of nodes. Therefore, we propose a new graph-based model for text classification named CGA2TC, which introduces contrastive learning with an adaptive augmentation strategy into obtaining more robust node representation. First, we explore word co-occurrence and document word relationships to construct a text graph. Then, we design an adaptive augmentation strategy for the text graph with noise to generate two contrastive views that effectively solve the noise problem and preserve essential structure. Specifically, we design noise-based and centrality-based augmentation strategies on the topological structure of text graph to disturb the unimportant connections and thus highlight the relatively important edges. As for the labeled nodes, we take the nodes with same label as multiple positive samples and assign them to anchor node, while we employ consistency training on unlabeled nodes to constrain model predictions. Finally, to reduce the resource consumption of contrastive learning, we adopt a random sample method to select some nodes to calculate contrastive loss. The experimental results on several benchmark datasets can demonstrate the effectiveness of CGA2TC on the text classification task.

Predicting Social Events with Multimodal Fusion of Spatial and Temporal Dynamic Graph Representations

Big data has been satisfactorily used to solve social issues in several parts of the word. Social event prediction is related to social stability and sustainable development. However, current research rarely takes into account the dynamic connections between event actors and learning robust feature representations of social events. Inspired by the graph neural network, we propose a novel Siamese Spatial and Temporal Dynamic Network for predicting social events. Specifically, we use multimodal data containing news articles and global events to construct dynamic graphs based on word co-occurrences and interactions between event actors. Dynamic graphs can model the evolution of social events. By employing the fusion of spatial and temporal dynamic graph representations from heterogeneous historical data, our proposed model predicts the occurrence of future social events for the target country. Qualitative and quantitative analysis of experiment results on multiple real-word datasets shows that our proposed method is competitive against several approaches for social event prediction.

Gender Stereotypes in Hollywood Movies and Their Evolution over Time: Insights from Network Analysis

The present analysis of more than 180,000 sentences from movie plots across the period from 1940 to 2019 emphasizes how gender stereotypes are expressed through the cultural products of society. By applying a network analysis to the word co-occurrence networks of movie plots and using a novel method of identifying story tropes, we demonstrate that gender stereotypes exist in Hollywood movies. An analysis of specific paths in the network and the words reflecting various domains show the dynamic changes in some of these stereotypical associations. Our results suggest that gender stereotypes are complex and dynamic in nature. Specifically, whereas male characters appear to be associated with a diversity of themes in movies, female characters seem predominantly associated with the theme of romance. Although associations of female characters to physical beauty and marriage are declining over time, associations of female characters to sexual relationships and weddings are increasing. Our results demonstrate how the application of cognitive network science methods can enable a more nuanced investigation of gender stereotypes in textual data.

Analysis and data visualization in bibliometric studies

In recent years, academics with different professional formation other than Information Sciences have been involved in research related to metric studies using approaches and techniques based on bibliometrics; however, these professionals do not have knowledge and necessary competences for conduct metric information studies in a rigorous and methodical way. This article aims to find what are methods and tools used by bibliometrics researchers through a descriptive and textual analysis, using methods such as word co-occurrence in bibliographical data extracted from scientific articles indexing in Web of Science, research area "Library & Information Science." We find most frequently software and tools used in bibliometrics have four levels for data treatment: recovery, preparation, processing and analysis, visualization; and use different methods analysis: networks social, geospatial, thematic, temporal, bibliographic coupling, among others.

Investigating Research Trend of Rock Climbing in China Using Semantic Network Analysis

PURPOSE The objectives of this study were as follows: (1) studying the core authors and institutions of rock climbing to understand the connections between them, (2) investigate the keyword changes, research hotspots, and development trends of rock climbing in different periods, and (3) discovering the subgroup areas of rock climbing and study their knowledge structure. METHOD The bibliometrics visualization software CiteSpace was used in this study. Semantic network analysis was performed on 376 valid papers published in CNKI during 1996-2021. The word frequency, mediating centrality, co-occurrence map, and clustering for the keywords were separately investigated. RESULT In terms of author collaboration, there are mainly four academic cooperation circles between Ma Chuhong, Guo Wei, Zhang Bo, and Li Yuanyuan. In terms of institutional cooperation networks, little cooperation occurs between rock climbing research institutions in China. Since 2009, rock climbing sport, rock climbing, current situation, feasibility, university, extreme sports, enthusiasts, artificial rock climbing, and other keywords have the highest frequency. Into rock climbing sport, rock climbing, and colleges and universities, keywords such as outdoor sports, college students, training, development, and sports tourism have become the focus of research since 2010. The rock-climbing knowledge structure consists of six sub-areas, which are G1 rock climbing, G2 college, G3 strategy, G4 climbing, G5 sports tourism, and G6 school training. CONCLUSION First, the field of rock-climbing keyword research has changed dramatically over the past 20 years. College student training and sports tourism in rock climbing are the current and future research hotspots and trends. Second, a few keywords have made significant contributions to rock climbing. Keyword semantic network analysis provides a new idea and research method for exploring the development of rock climbing.

Knowledge Mapping of Articles on Application of the Quran and Hadiths in Health Care: VOSviewer Visualization Techniques.

This study sought to conduct a knowledge mapping of articles to shed light on the application of the Quran and Hadiths in health care. The study is applied in terms of purpose and scientometrics in terms of type, which has been carried out using retrospective bibliometric analysis methods and visualization through knowledge mapping. In order to do so, the PubMed database with related keywords was thoroughly searched. Then, 302 relevant articles were reviewed in terms of the author, the year of publication, the journal, and the subject. In addition, VOSviewer software was employed to analyze and visualize keywords and maps of the articles. The findings showed that 302 articles were published between 1972 and 2021 in 190 journals, among which the Journal of Religion and Health was ranked first with 43 articles. Word co-occurrence map of scientific products of the Quran and Hadith in health care were categorized into sixclusters, namely: "complementary and alternative medicine," "ethics," "the effectiveness of spirituality on the treatment and control of mental illness," "women's rights and status," "family planning, infertility treatment and abortion," and "the views of the Quran and Hadiths on sexual issues." Adopting a holistic approach toward the application of various aspects of the Quran and Hadiths on health care, the present study can be considered as a pioneering research whose results can account for determining relevant applied research projects.

Improving NMF clustering by leveraging contextual relationships among words

Non-negative Matrix Factorization (NMF) and its variants have been successfully used for clustering text documents. However, NMF approaches like other models do not explicitly account for the contextual dependencies between words. To remedy this limitation, we draw inspiration from neural word embedding and posit that words that frequently co-occur within the same context (e.g., sentence or document) are likely related to each other in some semantic aspect. We then propose to jointly factorize the document-word and word-word co-occurrence matrices. The decomposition of the latter matrix encourages frequently co-occurring words to have similar latent representations and thereby reflecting the relationships among them. Empirical results, on several real-world datasets, provide strong support for the benefits of our approach. Our main finding is that we can drastically improve the clustering performance of NMF by leveraging the contextual relationships among words explicitly.

The Translation of Clausal Extension Collocation: A Contrastive Study of English-IndonesianTranslation

Abstract. Collocation is the co-occurrence of words comprising a node and one or more collocates. Insufficient studies have documented on the equivalence of source language (SL) collocation into more than one version of target language (TL) collocation. Hence, this study aims to explore the equivalence of English clausal extension collocations in the three versions of Indonesian translated novels, to identify the typology of the translations, to deal with the cohesive ties of the translations, to investigate the translation techniques used, and to observe the impact of translation techniques on the quality of the translation in terms of accuracy, acceptability, and readability. The methods applied in this research were quantitative and descriptive qualitative. The data were taken from a novel Pride and Prejudice and the three versions of Indonesian translated novels. Documentary analysis was implemented in collecting the data. They were pinpointed, coded, evaluated, classified and analysed. Further, the findings indicate that English clausal extension collocations are translated into Indonesian in the form of clausal extension, verbal extension, clausal enhancement, verbal enhancement collocation, and non-verbal node collocation. Meanwhile, the cohesive ties of the collocations include strong and weak cohesive ties. The translation techniques applied are generalization, established equivalence, modulation, amplification, and discursive creation. The use of generalization, established equivalence, modulation and amplification results in accurate, acceptable, and readable translations. On the other hand, discursive creation and deletion result in less accurate, less acceptable and less readable. It was concluded that it is the arbitrariness of language that results the different typology and cohesive tie of collocation among languages. Altogether, accuracy, acceptability and readability should be the goal of translation. Keywords: Arbitrariness; Clausal Extension Collocation, Translation Quality Assessment, Translation Techniques

A Systematic Review on Features Extraction Techniques for Aspect Based Text Classification Using Artificial Intelligence

Aspect extraction is an important and challenging and meaningful task in aspect-based text classification analysis. To apply variants of topic models on task, while reasonably successful, these methods usually do not produce highly coherent aspects. In this review, present a novel neural/cognitive approach to discover coherent aspects methods. Exploiting the distribution of word co-occurrences through neural/cognitive word embeddings. Unlike topics that typically assume independently generated words, word embedding models encourage words that appear in similar factors close to each other in the embedding space. Also, use an attention mechanism to de-emphasize irrelevant words during training, further improving aspects coherence. Methods results on datasets demonstrate that approach discovers more meaningful and coherent aspects and substantially outperforms baseline. Aspect-based text analysis aims to determine people's attitudes towards different aspects in a review.

Analysis of scientific production of refused derived fuel through scientometric and bibliometric indicators

Constant urbanization has intensified the volume of solid urban waste generated, in addition to causing the exhaustion of landfills and the elevation of carbon in the atmosphere. New treatment technologies allow waste to be transformed, having economic value and, consequently, increasing the useful life of landfills. Energy recovery as a solution for the recovery of waste, through conversion to Refused Derived Fuel, in order to be used, for example, in industries. The aim of this study was to analyze the current state of scientific production on RDF production through scientometric and bibliometric indicators. The distribution of study sites suggests a global interest from both developed and developing countries. The most frequent terms analyzed in the word cloud of the article titles focus on words related to the energy recovery of waste and the main words in the abstracts emphasized the methods of energy recycling, life cycle and incineration. The analysis of networks by co-occurrence of words showed that the synergy occurs around the term waste, with terms that demonstrate a tendency to use solid waste as an energy source, in order to contribute to the municipal management of these materials. In addition, one of the energy conversion processes, gasification, was highlighted, pointing out that this process is one of the most applied. Finally, the potential collaborations of the factor analysis that verified two clusters focused on energy issues and related to the management of urban solid waste.

Is it you you’re looking for?

Abstract Previous evidence has implicated personal relevance as a predictive factor in lexical access. Westbury (2014) showed that personally relevant words were rated as having a higher subjective familiarity than words that were not personally relevant, suggesting that personally relevant words are processed more fluently than less personally relevant words. Here we extend this work by defining a measure of personal relevance that does not rely on human judgments but is rather derived from first-order co-occurrence of words with the first-person singular personal pronoun, I. We show that words estimated as most personally relevant are recognized more quickly, named faster, judged as more familiar, and used by infants earlier than words that are less personally relevant. Self-relevance is also a strong predictor of several measures that are usually measured only by human judgments or their computational estimates, such as subjective familiarity, age of acquisition, imageability, concreteness, and body-object interaction. We have made all self-relevance estimates (as well as the raw data and code from our experiments) available at https://osf.io/gdb6h/.

Reiteration Used by the Participants of Warmadewa University Reading News Contest

This paper discusses about reiteration used by the participants of reading news contest conducted in Warmadewa University. Reiteration is part of lexical cohesion. The lexical cohesion has two functions; first to link between word and word, phrase and phrase within a text and second to indicate co-occurrence of words in which their meaning related in the same environment. The reiteration refers to first function of lexical cohesion. The theories applied to support this paper are the theory of cohesion from a book entitled Introducing Discourse Analysis by Nunan (1993) and a book entitled Cohesion in English by Halliday and Hasan (1976). The data of this paper taken from six texts made by the participants of the contest. The text contains with the participants’ perspectives on Indonesian culture in the globalization era. The data was analyzed by using qualitative method. Based on the result of the analysis, it is found that reiteration used in the text by the participants of reading news contest consists of repetition, synonym/nears synonym, superordinate and general word.