Understanding Of Context Research Articles

An analysis of the use of 3D game engine technology in visualising sustainability data

Purpose Responding to calls for accountants to engage with modern technologies and explore data visualisation within a three-dimensional (3D) environment, this study aims to explore whether social and environmental accounting (SEA) data visualisation is a promising use case for 3D game engine technology. Design/methodology/approach Drawing on visual perception and embodiment theories, this study uses photo-elicitation, a qualitative research method, to explore the usefulness of two-dimensional (2D) and 3D visualisations of sustainability information in a 3D virtual environment. This study provides three stimuli: numerical data, 2D visualisations and 3D visualisations, and asks open-ended questions regarding future applications. Twelve semi-structured interviews were conducted with academics, preparers and users of sustainability reports to obtain responses to these stimuli. Findings The key finding is that visualisation of SEA information may indeed be a strong use case for 3D game technology, but only for certain data and for certain audiences. Presenting information within a 3D virtual environment offered enhanced engagement and contextual understanding but reduced navigation speed and data clarity. Participants were enthusiastic about the potential of a museum-like experience, incorporating interactivity and community, but felt that the appropriate audience was more likely to be novices than experts. Practical implications This study suggests that deploying 3D game engine technology can be a powerful tool for presenting sustainability information but requires significant resources. The optimum audience is likely to be novices, and a key design principle is to ensure the virtual environment supports, rather than overwhelms, the information presented within that environment. Originality/value This study introduces a novel application of 3D visualisation technology within the SEA context, offering original insights into its potential to enhance user understanding and decision-making capabilities. This study highlights the technology’s value not as a replacement for traditional reporting but as a supplementary educational tool. The study also provides a novel setting for the photo-elicitation method, demonstrating this approach’s utility in a 3D environment.

Transformer-Enhanced Retinal Vessel Segmentation for Diabetic Retinopathy Detection Using Attention Mechanisms and Multi-Scale Fusion

Eye health has become a significant concern in recent years, given the rising prevalence of visual impairment resulting from various eye disorders and related factors. Global surveys suggest that approximately 2.2 billion individuals are visually impaired, with at least 1 billion affected by treatable diseases or ailments. Early detection, treatment, and screening for fundus diseases are crucial in addressing these challenges. In this study, we propose a novel segmentation model for retinal vascular delineation aimed at diagnosing diabetic retinopathy. The model integrates CBAM (Channel-Attention and Spatial-Attention) for enhanced feature representation, JPU (Joint Pyramid Upsampling) for multi-scale feature fusion, and transformer blocks for contextual understanding. Leveraging deep-learning techniques, our proposed model outperforms existing approaches in retinal vascular segmentation, like achieving a Mean IOU of 0.8047, Recall of 0.7254, Precision of 0.8492, F1 Score of 0.7824, and Specificity of 0.9892 for CHASEDB1 dataset. Extensive evaluations on benchmark datasets demonstrate its efficacy, highlighting its potential for automated diabetic retinopathy screening.

Fusion Text Representations to Enhance Contextual Meaning in Sentiment Classification

Sentiment classification plays a crucial role in evaluating user feedback. Today, online media users can freely provide their reviews with few restrictions. User reviews on social media are often disorganized and challenging to classify as positive or negative comments. This task becomes even more difficult when dealing with large amounts of data, making sentiment classification necessary. Automating sentiment classification involves text classification processes, commonly performed using deep learning methods. The classification process using deep learning models is closely tied to text representation. This step is critical as it affects the quality of the data being processed by the deep learning model. Traditional text representation methods often overlook the contextual meaning of sentences, leading to potential misclassification by the model. In this study, we propose a novel fusion text representation model, GloWord_biGRU, designed to enhance the contextual understanding of sentences for sentiment classification. Firstly, we combine the advantages of GloVe and Word2Vec to obtain richer and more meaningful word representations. GloVe provides word representations based on global frequency statistics within a large corpus, while Word2Vec generates word vectors that capture local contextual relationships. By integrating these two approaches, we enhance the quality of word representations used in our model. During the classification stage, we employ biGRU, considering the use of fewer parameters, which consequently reduces computational requirements. We evaluate the proposed model using the IMDB dataset. Several scenarios demonstrate that our proposed model achieves superior performance, with an F1 score of 90.21%.

Discrimination analysis of quantitative indicators of understanding the written form of deaf children's communication through the use of types of words

In deaf students, there is no contextual understanding and recognition of word types through linguistic competence testing compared to the hearing population, because 67.10% do not understand, and 10.00% of deaf children partially understand the contextual application of word types in a written text task. The aim of the study is to determine the distance of language discourse between deaf and hearing respondents and to establish a discrimination factor that can be used for practical purposes in a classification sense, with the aim of determining the priority of linguistic education and language elements of deaf children in relation to their lagging behind hearing children. The study was conducted on a sample of 140 respondents. The first subsample of respondents, the experimental group consisted of 70 deaf students, and the second subsample, the control group of 70 hearing students, of the same chronological age and gender. The measuring instrument “Test of comprehension of the written form of expression” was applied. The collected data were processed using the discriminant analysis method. The results of the study showed that the discrimination factor is in the sequence of the use of words, adjectives, exclamations and adverbs. Nouns and prepositions have a negative correlation, which points to the fact that these four types of words are in direct implication with nouns and prepositions, and represents information that these four types of words must be more represented in the educational materials of written expression of deaf children. The results of the study also open up a series of questions focused on the quality of the educational processes of deaf children, as well as the level of lag in written communication compared to hearing children. The results of the study can influence the raising of general rehabilitation procedures to a higher level of responsibility in education centers where deaf children are educated.

Lessons learned from COVID-19 modelling efforts for policy decision-making in lower- and middle-income countries

IntroductionThe COVID-19 pandemic had devastating health and socioeconomic effects, partly due to policy decisions to mitigate them. Little evidence exists of approaches that guided decisions in settings with limited pre-pandemic...

Is the information provided by large language models valid in educating patients about adolescent idiopathic scoliosis? An evaluation of content, clarity, and empathy : The perspective of the European Spine Study Group.

Large language models (LLM) have the potential to bridge knowledge gaps in patient education and enrich patient-surgeon interactions. This study evaluated three chatbots for delivering empathetic and precise adolescent idiopathic scoliosis (AIS) related information and management advice. Specifically, we assessed the accuracy, clarity, and relevance of the information provided, aiming to determine the effectiveness of LLMs in addressing common patient queries and enhancing their understanding of AIS. We sourced 20 webpages for the top frequently asked questions (FAQs) about AIS and formulated 10 critical questions based on them. Three advanced LLMs-ChatGPT 3.5, ChatGPT 4.0, and Google Bard-were selected to answer these questions, with responses limited to 200 words. The LLMs' responses were evaluated by a blinded group of experienced deformity surgeons(members of the European Spine Study Group) from seven Europeanspine centers. A pre-established 4-level rating system from excellent to unsatisfactory was used with a further rating for clarity, comprehensiveness, and empathy on the 5-point Likert scale. If not rated 'excellent', the raters were asked to report the reasons for their decision for each question. Lastly, raters were asked for their opinion towards AI in healthcare in general in six questions. The responses among all LLMs were 'excellent' in 26% of responses, with ChatGPT-4.0 leading (39%), followed by Bard (17%). ChatGPT-4.0 was rated superior to Bard and ChatGPT 3.5 (p = 0.003). Discrepancies among raters were significant (p < 0.0001), questioning inter-rater reliability. No substantial differences were noted in answer distribution by question (p = 0.43). The answers on diagnosis (Q2) and causes (Q4) of AIS were top-rated. The most dissatisfaction was seen in the answers regarding definitions (Q1) and long-term results (Q7). Exhaustiveness, clarity, empathy, and length of the answers were positively rated (> 3.0 on 5.0) and did not demonstrate any differences among LLMs. However, GPT-3.5 struggled with language suitability and empathy, while Bard's responses were overly detailed and less empathetic. Overall, raters found that 9% of answers were off-topic and 22% contained clear mistakes. Our study offers crucial insights into the strengths and weaknesses of current LLMs in AIS patient and parent education, highlighting the promise of advancements like ChatGPT-4.oand Gemini alongside the need for continuous improvement in empathy, contextual understanding, and language appropriateness.

Artificial intelligence for language learning and teaching: A narrative literature study

The purpose of this study was to examine teachers' beliefs, the factors that influence the use of AI tools, and the advantages, disadvantages, opportunities, and threats of artificial intelligence in language learning and teaching. To accomplish this, the study used a narrative literature review and ATLAS.ti 9 to analyze the data. The results of the study showed that teachers believed that Artificial Intelligence in Language Learning and Teaching (AILLT) included assessing student needs, developing appropriate learning resources, promoting collaboration among resources, providing real-time assessment tools to improve the educational process, and increasing teaching efficiency and effectiveness. Second, AI tools for writing tools, including Paperpal, Quillbot, Jenni AI, ChatGPT, Elicit, as well as in applications to improve speaking skills, such as Speeko and Vocaroo, and "Siri and Say It" for listening and pronunciation skills. Third, factors that may influence students' use of AI-based language learning tools include adaptability, engagement, motivation, autonomy, immediate or direct feedback, accessibility and inclusivity, and teacher and parent support. This research found that AI enhances accessibility, adaptability, personalization, and immediate feedback, but it also has limitations, such as dependence on technology, inadequate human interaction, limited contextual understanding, and algorithmic biases. In addition, AILLT faces potential threats such as quality control, privacy concerns, and job or task displacement. Narrative literature studies provide theoretical insights, and it is expected that future studies at different levels of language learning will incorporate empirical evidence from experimental and case study-based research.

The impact of contextual understanding on neural machine translation accuracy: A case study of Indonesian cultural idioms in English translation

The translation of Indonesian idiomatic expressions into English using Neural Machine Translation (NMT) systems presents significant challenges due to the intricate nature of idiomatic language. Idioms are culturally embedded constructs conveying meanings different from their literal interpretations. This study explores the effectiveness of NMT systems in capturing and translating Indonesian idiomatic expressions into English. Using a qualitative approach, 150 diverse Indonesian idioms were evaluated through Google Translate and DeepL. Each idiom was assessed for semantic accuracy, syntactic coherence, and contextual fidelity. Qualitative analysis provided a comprehensive evaluation of translation quality. Findings reveal significant challenges for NMT systems in translating idiomatic expressions. Both Google Translate and DeepL show strengths in some areas, but also have limitations. While they generally capture literal meanings, they often miss metaphorical nuances and cultural connotations. Syntactic errors such as incorrect word order and tense inconsistencies are common, especially in complex idioms. In addition, contextual fidelity analysis shows that NMT systems struggle with contextual appropriateness and pragmatic usage, resulting in translations lacking cultural sensitivity and relevance. These challenges highlight the need for improved algorithms to better interpret and translate idiomatic language across cultural contexts. Accurate idiom translation is crucial for language learners to understand cultural nuances and idiomatic usage. Enhancing the capabilities of NMT systems requires refining algorithms, developing context-aware models, and expanding training datasets with diverse idiomatic expressions. Overcoming these challenges will advance machine translation capabilities and improve cross-cultural communication and language learning experiences.

Multifaceted Natural Language Processing Task-Based Evaluation of Bidirectional Encoder Representations From Transformers Models for Bilingual (Korean and English) Clinical Notes: Algorithm Development and Validation.

The bidirectional encoder representations from transformers (BERT) model has attracted considerable attention in clinical applications, such as patient classification and disease prediction. However, current studies have typically progressed to application development without a thorough assessment of the model's comprehension of clinical context. Furthermore, limited comparative studies have been conducted on BERT models using medical documents from non-English-speaking countries. Therefore, the applicability of BERT models trained on English clinical notes to non-English contexts is yet to be confirmed. To address these gaps in literature, this study focused on identifying the most effective BERT model for non-English clinical notes. In this study, we evaluated the contextual understanding abilities of various BERT models applied to mixed Korean and English clinical notes. The objective of this study was to identify the BERT model that excels in understanding the context of such documents. Using data from 164,460 patients in a South Korean tertiary hospital, we pretrained BERT-base, BERT for Biomedical Text Mining (BioBERT), Korean BERT (KoBERT), and Multilingual BERT (M-BERT) to improve their contextual comprehension capabilities and subsequently compared their performances in 7 fine-tuning tasks. The model performance varied based on the task and token usage. First, BERT-base and BioBERT excelled in tasks using classification ([CLS]) token embeddings, such as document classification. BioBERT achieved the highest F1-score of 89.32. Both BERT-base and BioBERT demonstrated their effectiveness in document pattern recognition, even with limited Korean tokens in the dictionary. Second, M-BERT exhibited a superior performance in reading comprehension tasks, achieving an F1-score of 93.77. Better results were obtained when fewer words were replaced with unknown ([UNK]) tokens. Third, M-BERT excelled in the knowledge inference task in which correct disease names were inferred from 63 candidate disease names in a document with disease names replaced with [MASK] tokens. M-BERT achieved the highest hit@10 score of 95.41. This study highlighted the effectiveness of various BERT models in a multilingual clinical domain. The findings can be used as a reference in clinical and language-based applications.

Contemporary understanding of health literacy across the Island of Ireland

Abstract This presentation will detail two reviews that provided contextual understanding for this project. The first, an umbrella review which explored i) How is health literacy currently defined and operationalised nationally and internationally? And ii) What are the key components and principles of health literacy? The search was limited, most notably, to systematic reviews published after 2012. A total of 12 studies were included for data extraction and narrative analysis. Findings provided an update of the conceptual understanding of health literacy, emphasising the complex and multidimensional nature of the concept. Dimensions are common, but growing, and moving beyond just individual capabilities. The second, moving beyond the definition, details a systematic review that aimed to gain a contemporary understanding of health literacy through identifying and appraising existing research conducted in an Irish context. Seven electronic databases were searched between October and December 2023. Initial search identified 551 records, following screening, 37 studies were included in this review. Narrative analysis indicated that across the Island of Ireland in the last decade, many studies had narrow populations of focus (i.e. Dublin based, adults, chronic illness populations), limited research design and methodologies (cross-sectional, narrative, and primary research with short time frames), and lacked rigorous monitoring and evaluation of health literacy as a primary or secondary outcome. While studies highlighted clinical, education and community implications, a limited focus on mental health was also observed. Findings from this systematic review offer valuable insights and recommendations for future efforts to enhance health literacy and the health and well-being of individuals across the island of Ireland.

Assessing Performance of Cloud-based Heterogeneous Chatbot Systems.

Cloud-based heterogeneous chatbot systems represent a significant advancement in automating diverse tasks across various platforms, including customer service, healthcare, and education. These systems are designed with multi-modal capabilities, allowing them to interact via text, voice, and visual interfaces, and are built on a cloud infrastructure that ensures scalability and reliability. The cloud-based architecture allows dynamic resource allocation, enabling the system to handle high volumes of traffic without compromising performance. These chatbots also benefit from global availability, as cloud regions distributed across the world ensure low latency and high availability.The integration of advanced AI and NLP models, such as GPT and BERT, enhances the chatbot's response accuracy and contextual understanding, making interactions more precise and human-like. Moreover, the cloud infrastructure supports continuous learning through machine learning models that refine chatbot performance over time. With multitenant capabilities, these chatbots can cater to different users or organizations, providing tailored services while ensuring that performance are not affected. Additionally, their ability to integrate with external enterprise systems, like CRMs or ERPs, allows for real-time data exchange and task automation.In terms of security, cloud-based chatbots benefit from encryption protocols and compliance with industry regulations, ensuring safe handling of sensitive data, particularly in sectors like healthcare and finance. Despite the advantages, challenges such as network dependency and cost management need careful consideration, as performance can be affected by poor connectivity or high processing costs, especially when using complex AI models. Nonetheless, cloud-based heterogeneous chatbots excel in delivering scalable, reliable, and intelligent interactions, adapting to growing user demands and a wide range of applications.

Evaluating the Translation Accuracy of ChatGPT and DeepL Through the Lens of Implied Subjects

This research aims to offer deeper insights into the translation quality of large language models and neural machine translation systems. It compares the performance of two state-of-the-art models, ChatGPT and DeepL, to explore which system achieves higher translation accuracy. This research employs a highly focused and specialized method to evaluate translation performance, enabling the identification of nuanced differences between advanced AI/machine models. It focuses on implied subjects in contemporary Chinese writer Qiuyu Yu’s Iraq travelogue and examines how the two models handle these implied subjects when translating the text into English. The analysis shows that both ChatGPT and DeepL exhibit high-level intelligence when dealing with the implied subjects. They can flexibly adjust sentence structures to maintain the implicitness or to make the implied subjects explicit. The analysis further reveals that ChatGPT performs better in accurately inferring the implied subjects, which thereby reflects ChatGPT’s superior contextual understanding capability. The comparison of the source text and target texts also shows that ChatGPT’s translation is more succinct, nuanced, and easier to read. DeepL’s translation, by comparison, prioritizes reducing potential misunderstanding. This research tentatively concludes that, though not specialized for translation, ChatGPT translates with higher accuracy and demonstrates more sophisticated intelligence compared to DeepL when translating literary texts.

A pedagogical evaluation of an institution's digital assessment platform (DAP): Integrating pedagogical, technical and contextual factors

An assessment for learning approach is foundational for student learning. The necessity to shift teaching and learning online as a response to COVID-19 has propelled digital assessment into the mainstream within higher education institutions. User experience is a common indicator of effectiveness of technologically enhanced initiatives; however, measuring the pedagogical impact of digital assessment initiatives has not been straightforward. This paper presents a pedagogical evaluation, incorporating staff and student perspectives, of a digital assessment platform (DAP) at a large Australian metropolitan university. Data were collected using surveys, interviews, focus groups and case studies and incorporated findings from student evaluations of courses and a student-staff partnership project. Results highlight strong pedagogical reasons for continuing the digital assessment initiative, while also identifying constraints and opportunities for improvement. The insights generated through this evaluation emphasise the importance of considering technical, pedagogical and contextual factors and the interplay between the factors. In a quality assurance climate, where the demand for this type of institution-wide evaluation is increasing, this study contributes an approach that addresses some of the complexities and challenges of evaluating digital assessment initiatives in higher education. Future research is needed to understand more fully the relationships between pedagogical and contextual factors when undertaking an evaluation. Implications for practice or policy: The DAP enabled the (re)design of authentic, inclusive, engaging and secure assessment tasks. Educator support is critical in learning the system and creating assessment with high pedagogical merit. Student familiarisation opportunities with using the DAP are beneficial. The DAP’s longevity within the institution is crucial considering the significant effort required of educators and students. Contextual influences blur understanding of pedagogical factors.

An ensemble deep learning model for medical image fusion with Siamese neural networks and VGG-19.

Multimodal medical image fusion methods, which combine complementary information from many multi-modality medical images, are among the most important and practical approaches in numerous clinical applications. Various conventional image fusion techniques have been developed for multimodality image fusion. Complex procedures for weight map computing, fixed fusion strategy and lack of contextual understanding remain difficult in conventional and machine learning approaches, usually resulting in artefacts that degrade the image quality. This work proposes an efficient hybrid learning model for medical image fusion using pre-trained and non-pre-trained networks i.e. VGG-19 and SNN with stacking ensemble method. The model leveraging the unique capabilities of each architecture, can effectively preserve the detailed information with high visual quality, for numerous combinations of image modalities in image fusion challenges, notably improved contrast, increased resolution, and lower artefacts. Additionally, this ensemble model can be more robust in the fusion of various combinations of source images that are publicly available from Havard-Medical-Image-Fusion Datasets, GitHub. and Kaggle. Our proposed model performance is superior in terms of visual quality and performance metrics to that of the existing fusion methods in literature like PCA+DTCWT, NSCT, DWT, DTCWT+NSCT, GADCT, CNN and VGG-19.

Issues in Machine Translation with reference to the Story “Thakur ka kuan” by Munshi Premchand

This paper explores the issues encountered in machine translation, specifically using Google Translate, in translating Munshi Premchand's short story Thakur Ka Kuan ("Thakur's Well"). The study identifies key challenges, including the misinterpretation of cultural references, idiomatic expressions, and contextual meaning that are integral to Premchand's narrative style. Machine translation often produces literal translations that lack the nuance and emotional depth present in the original text, leading to misrepresentation of the story’s themes. For instance, socio-cultural terms and metaphors central to the story's message are frequently mistranslated or omitted entirely. The translation also struggles with maintaining consistency in tense and sentence structure, further diluting the narrative. This paper argues that while machine translation offers convenience for basic translations, it remains insufficient for complex literary works, where cultural, emotional, and contextual understanding is crucial. A comparative analysis with a human-translated version highlights these deficiencies and suggests that human input remains essential for accurate literary translations. The findings underscore the need for improved machine learning models to address these linguistic and cultural intricacies.

Ecological empathy: relational theory and practice

ABSTRACT Sustainability science researchers are increasingly interested in human–nature connection as a leverage point for societal transformation. Empathy has potential as a way to reconnect people to nature by building relationships among more-than-humans. However, current approaches to empathy with more-than-humans usually prefer sympathy and compassion. I argue that these approaches limit the potential of empathy when considering human–nature (re)connection. I use the established concept of social empathy (Segal 2011, 2018) to structure a new presentation of empathy with more-than-humans: ecological empathy. Ecological empathy, as presented in this paper, consists of two subcomponents: contextual understanding of more-than-human interdependencies and more-than-human awareness and earth system perspective-taking. From this new perspective, I suggest defining ecological empathy as a cognitive and affective ability, which allows for internal coherence across bodily separation in humans and their environment. Integrating literature from biophilia, deep ecology, embodied cognition and multi-species ethnography, I elaborate on ecological empathy with inspirational practices that can be advanced across a range of decision, policy and design environments to address human–nature (re)connection.

At-risk or a risk? SENCOs’ conceptualisations of vulnerability and risk in relation to school exclusion

ABSTRACT In this paper we explore how Special Educational Needs Coordinators (SENCOs) conceptualise vulnerability and risk, and how these conceptualisations inform their responses to students at risk of exclusion from school. The literature typically makes a distinction between within-child and systemic or structural factors. We draw on interview data from SENCOs working in 11 secondary schools in England, and an activity theory approach to unpick the contradictions and tensions in their conceptualisations and responses. Overall, we identify a mismatch between SENCOs’ conceptualisations of risk and vulnerability intersecting at multiple levels, and interventions that are primarily located at an individual level. National policy, with its shift away from a contextual understanding of difficulties and difference towards a requirement for clinical diagnosis to access additional funding plays a key role in mediating or shaping SENCO responses, including the tools that schools use to identify at-risk students. The conceptualisations indicate a source of differential impact with some students shifting from being seen as ‘at-risk’ and requiring support, to being ‘a risk’ and better educated elsewhere. In the face of these conceptualisations, we argue for the place of a social relational understanding, emphasising the dialogic spaces in which behaviour is framed, and interventions planned.

A ‘Back Door’ Exploration of Social Problems in Comic Strip Dialogues With Marginalized Children

ABSTRACTBased on 35 comic strip dialogues with marginalized children in Italy, Norway and Slovenia, as well as three individual interviews with Italian practitioners, this article introduces a methodological approach that offers a ‘back door’ for exploring social problems embedded in the everyday lives of children. The analysis demonstrates the importance of adopting an everyday life perspective as an analytical strategy in conversations with children in marginalized life situations. This means shifting the focus away from solely examining risk factors and instead promoting a contextual understanding of risk, with an emphasis on amplifying the diverse voices of children.

Congruences of factor solutions of quantitative indicators for understanding the written form of deaf children's communication through the use of words types

In deaf students, there is no contextual understanding and recognition of word types through linguistic competence testing compared to the hearing population, because 67.10% do not understand, and 10.00% of deaf respondents partially understand the contextual application of word types in a written text task. In written communication, deaf students use nouns (65.70%), pronouns (34.30%), verbs (45.7%), adjectives (28.60%), adverbs (22.90%), prepositions (54.30%), exclamations (15.70%), particles (12.90%) and numbers (32.90%). The aim of the study was to examine the factor solutions of quantitative indicators of contextual understanding of the word’s types used and the relation of isolated factors of the words types words in the written text task of deaf students. The study was conducted on a sample of 140 respondents. The first subsample of respondents, the experimental group consisted of 70 deaf students, and the second subsample, the control group of 70 hearing students, of the same chronological age and gender. The measurement instrument “Test of comprehension of the written form of expression” was applied. The collected data were processed using the method of oblique rotation and the method of factor analysis. The results of the study showed that the language discourse of the deaf in the quantitative use of word types in the written form of expression is organized significantly differently compared to the hearing population. By factor analysis, three main factors were isolated in the deaf population, and four main factors were isolated in the hearing population. Considering the saturation of variability, it is possible to accurately determine in practice the direction of teaching deaf children the types of words in their native language. Isolated factors are part of separate, statistically unrelated entities in both deaf and hearing respondents and confirm the justification of successive teaching of the deaf respondents according to the types of words that are present in the comparative analysis of hearing respondents.

Joint intent detection and slot filling with syntactic and semantic features using multichannel CNN-BiLSTM

Understanding spoken language is crucial for conversational agents, with intent detection and slot filling being the primary tasks in natural language understanding (NLU). Enhancing the NLU tasks can lead to an accurate and efficient virtual assistant thereby reducing the need for human intervention and expanding their applicability in other domains. Traditionally, these tasks have been addressed individually, but recent studies have highlighted their interconnection, suggesting better results when solved together. Recent advances in natural language processing have shown that pretrained word embeddings can enhance text representation and improve the generalization capabilities of models. However, the challenge of poor generalization in joint learning models for intent detection and slot filling remains due to limited annotated datasets. Additionally, traditional models face difficulties in capturing both the semantic and syntactic nuances of language, which are vital for accurate intent detection and slot filling. This study proposes a hybridized text representation method using a multichannel convolutional neural network with three embedding channels: non-contextual embeddings for semantic information, part-of-speech (POS) tag embeddings for syntactic features, and contextual embeddings for deeper contextual understanding. Specifically, we utilized word2vec for non-contextual embeddings, one-hot vectors for POS tags, and bidirectional encoder representations from transformers (BERT) for contextual embeddings. These embeddings are processed through a convolutional layer and a shared bidirectional long short-term memory (BiLSTM) network, followed by two softmax functions for intent detection and slot filling. Experiments on the air travel information system (ATIS) and SNIPS datasets demonstrated that our model significantly outperformed the baseline models, achieving an intent accuracy of 97.90% and slot filling F1-score of 98.86% on the ATIS dataset, and an intent accuracy of 98.88% and slot filling F1-score of 97.07% on the SNIPS dataset. These results highlight the effectiveness of our proposed approach in advancing dialogue systems, and paving the way for more accurate and efficient natural language understanding in real-world applications.