Conversational Context Research Articles

M3GAT: A Multi-modal, Multi-task Interactive Graph Attention Network for Conversational Sentiment Analysis and Emotion Recognition

Sentiment and emotion, which correspond to long-term and short-lived human feelings, are closely linked to each other, leading to the fact that sentiment analysis and emotion recognition are also two interdependent tasks in natural language processing (NLP). One task often leverages the shared knowledge from another task and performs better when solved in a joint learning paradigm. Conversational context dependency, multi-modal interaction, and multi-task correlation are three key factors that contribute to this joint paradigm. However, none of the recent approaches have considered them in a unified framework. To fill this gap, we propose a multi-modal, multi-task interactive graph attention network, termed M3GAT, to simultaneously solve the three problems. At the heart of the model is a proposed interactive conversation graph layer containing three core sub-modules, which are: (1) local-global context connection for modeling both local and global conversational context, (2) cross-modal connection for learning multi-modal complementary and (3) cross-task connection for capturing the correlation across two tasks. Comprehensive experiments on three benchmarking datasets, MELD, MEISD, and MSED, show the effectiveness of M3GAT over state-of-the-art baselines with the margin of 1.88%, 5.37%, and 0.19% for sentiment analysis, and 1.99%, 3.65%, and 0.13% for emotion recognition, respectively. In addition, we also show the superiority of multi-task learning over the single-task framework.

Generative retrieval for conversational question answering

Effective passage retrieval is crucial for conversation question answering (QA) but challenging due to the ambiguity of questions. Current methods rely on the dual-encoder architecture to embed contextualized vectors of questions in conversations. However, this architecture is limited in the embedding bottleneck and the dot-product operation. To alleviate these limitations, we propose generative retrieval for conversational QA (GCoQA). GCoQA assigns distinctive identifiers for passages and retrieves passages by generating their identifiers token-by-token via the encoder–decoder architecture. In this generative way, GCoQA eliminates the need for a vector-style index and could attend to crucial tokens of the conversation context at every decoding step. We conduct experiments on three public datasets over a corpus containing about twenty million passages. The results show GCoQA achieves relative improvements of +13.6% in passage retrieval and +42.9% in document retrieval. GCoQA is also efficient in terms of memory usage and inference speed, which only consumes 1/10 of the memory and takes in less than 33% of the time. The code and data are released at https://github.com/liyongqi67/GCoQA.

Aoba_v3 bot: a multimodal chatbot system combining rules and various response generation models

In this paper, we present a multimodal dialogue system combining a neural response generation mechanism, a reranking mechanism, and a rule-based avatar control mechanism. Our system was submitted to the open track at the Fifth Dialogue System Live Competition and won second place. Remarkably, our system received the best human evaluation performance for visual information control (i.e. speaking style of avatar) in the preliminary round. The assessment of the competition evaluators revealed that our system generates natural utterances appropriate to the conversational context and topic with an appealing speaking style. Through the analysis, we found that our devices, such as post-processing on speech recognition results and the final response selection method, are effective, but we also found room for improvement, such as speech recognition errors and challenges in the reranking module.

Towards a theology of the Psalm titles: The Davidic voice and the totus Christus

AbstractThis article explores the ways in which the thirteen ‘biographical superscriptions’ which are found throughout the Psalter contribute to the blending of the Davidic voice which they invoke and the corporate voice of the community which receives them. It suggests that by receiving these thirteen Psalms, the canonical community enters an intensive identification with David and participates in the Davidic life and experience. Once this is established, the discussion turns to examine these insights in a Christian theological context in conversation with Augustine's totus Christus principle. It is suggested that the hermeneutical situation created by the biographical superscriptions provides a way for the totus Christus principle to be re-invoked in contemporary interpretation.

Practical (un)cancellability

Cancellability is an essential feature of implicatures. However, its reliability has been challenged by several cases and examples in which conversational implicatures seem to be hard or even impossible to cancel. Should it then be concluded that not all implicatures are cancellable, and therefore Grice's cancellability test should be weakened or abandoned? The present paper addresses this problem by drawing a distinction between theoretical and practical cancellability, where the latter concept captures the (un)reasonableness of explicit or contextual cancellation. By analyzing three legal cases of misleading advertising, I show how the indefensibility of an interpretation can account for the practical uncancellability of an implicature. To explain and account for the reasonableness of an interpretation, the presumptive inferences involved must be reconstructed and evaluated considering different dimensions, such as the conversational context and the associated presumptions.

The metaphysical novel as educator: Simone de Beauvoir’s philosophy of lived experience

This essay analyzes the educational significance of the metaphysical novel, that is, how it can be used to educate ourselves and our students. Mordechai Gordon begins by describing the nature of the metaphysical novel while contrasting it to “pure” philosophy and theory building. Gordon also situates Beauvoir’s insights in the broader context of the ongoing conversation on philosophy and literature. In the next part, he examines Beauvoir’s philosophy of lived experience and compare her philosophical approach to more traditional phenomenological theories. From the analysis of Beauvoir’s philosophy of lived experience, Gordon turns to explore the role that ambiguity, contingency, and complexity play in helping us make sense of people’s lives while drawing on examples from several metaphysical novels to illustrate these notions. He concludes this essay by reflecting on the question: what can the study of metaphysical novels bring to philosophy of education?

Using artificial intelligence-enhanced video feedback for reflective practice in coach development: benefits and potential drawbacks

ABSTRACT Sports coaching has used video feedback for decades to improve athlete and coach performance. More recently, artificial intelligence (AI) and machine-learning technologies have enabled analytics alongside visual review to accelerate development further. So too in coaching conversations, software allows the implementation of behavioural analytics, tracking speech patterns, body language and facial expressions to deliver performance data measured against core coaching competencies. In this study, we interviewed 15 coaches who used AI-enhanced video review software in coaching sessions with clients over several weeks and reflected on those sessions using the recordings and AI-generated data. Our aim was to discover the benefits and drawbacks of using such an approach in reflective practice. Clear benefits emerged: insights gained from video and data analysis drove deeper reflection and heightened self-awareness; coaches focused on skills development, made specific changes to their practice, developed over time and gained in confidence. Challenges included coaches’ nervousness around using new technology, viewing and analysing their own performance, and a sense that the software does not understand the subtle nuances and context of conversations. Limitations are discussed as well as the implications for coach training, reflective practice and supervision. We suggest possibilities for further study in this area.

Conditional Collapse

AbstractIndicative and subjunctive conditionals are in non-complimentary distribution: there are conversational contexts at which both are licensed (Stalnaker 1975; Karttunen and Peters 1979; von Fintel 1998). This means we can ask an important, but under-explored, question: in contexts which license both, what relations hold between the two? In this paper, I’ll argue for an initially surprising conclusion: when attention is restricted to the relevant contexts, indicatives and subjunctives are co-entailing. §1 introduces the indicative/subjunctive distinction, along with a discussion of the relevant notion of entailment; §2 presents the main argument of the paper, and §3 considers some of the philosophical implications of the argument in §2. Finally, §4 argues that we can reconcile the equivalence of indicatives and subjunctives with apparently conflicting judgements.

Knowledge-Bridged Causal Interaction Network for Causal Emotion Entailment

Causal Emotion Entailment aims to identify causal utterances that are responsible for the target utterance with a non-neutral emotion in conversations. Previous works are limited in thorough understanding of the conversational context and accurate reasoning of the emotion cause. To this end, we propose Knowledge-Bridged Causal Interaction Network (KBCIN) with commonsense knowledge (CSK) leveraged as three bridges. Specifically, we construct a conversational graph for each conversation and leverage the event-centered CSK as the semantics-level bridge (S-bridge) to capture the deep inter-utterance dependencies in the conversational context via the CSK-Enhanced Graph Attention module. Moreover, social-interaction CSK serves as emotion-level bridge (E-bridge) and action-level bridge (A-bridge) to connect candidate utterances with the target one, which provides explicit causal clues for the Emotional Interaction module and Actional Interaction module to reason the target emotion. Experimental results show that our model achieves better performance over most baseline models. Our source code is publicly available at https://github.com/circle-hit/KBCIN.

Contextualizing Online Conversational Networks

Online social connections occur within a specific conversational context. Prior work in network analysis of social media data attempts to contextualize data through filtering. We propose a method of contextualizing online conversational connections automatically and illustrate this method with Twitter data. Specifically, we detail a graph neural network model capable of representing tweets in a vector space based on their text, hashtags, URLs, and neighboring tweets. Once tweets are represented, clusters of tweets uncover conversational contexts. We apply our method to a dataset with 4.5 million tweets discussing the 2020 US election. We find that even filtered data contains many different conversational contexts, with users engaging in multiple conversations. While users engage in multiple conversations, the overlap between any two pairs of conversations tends to be only 30-40%, giving very different networks for different conversations. Even accounting for this variation, we show that the relative social status of users varies considerably across contexts, with tau=0.472 on average. Our findings imply that standard network analysis on social media data can be unreliable in the face of multiple conversational contexts.

Keigo Learning in Building Omotenashi no Kokoro through a Communicative Language Teaching Approach

In building interpersonal relationships through omotenashi no kokoro (Excellent Service), Keigo is a crucial aspect to consider when communicating in Japanese. Keigo, both in terms of grammar and its usage concept, is quite complicated. Therefore, the role of instructors in delivering Keigo learning materials is expected to motivate students to improve their communication skills while mastering the art of creating interpersonal connections. Communicative Language Teaching (CLT) is chosen as the teaching method. This research aims to provide an alternative teaching experience focusing on omotenashi no kokoro. The research method used is descriptive with a qualitative approach. The research subjects are students of the Japanese Language Program at Universitas Teknologi Yogyakarta (UTY) taking the hospitality course. The research is conducted through observation by analyzing CLT-based learning activities using three steps: mechanical practice, meaningful practice, and communicative practice. The study results indicate that with the CLT method, students can use Keigo through word or phrase repetition, understand the meaning of words and expressions in sentences, and freely actualize themselves through improvisation and exploration of conversational contexts. A new finding from this research is that the CLT teaching method is effective when applied to the hospitality course, emphasizing building interpersonal relationships through omotenashi no kokoro.

Travelling in a Cosmopolitan Milieu: An Intercultural Exchange of Two Artists as Educators in Conversation at a UK University

By travelling in a cosmopolitan milieu in a UK university, pedagogies of possibilities are explored. Over a period of five years the exploration narrates the journeys of, what Clover (2010) calls, “artists as educators,” documenting conversations and creative pedagogic practises. This is despite the closure of the university campus in 2023, due to an overt neoliberal agenda. The findings of two feminist cosmopolitans in conversation, illustrates that when stories require repetition, that is the retelling of a certain narrative as well as stories that are whispered, they are of a particular significance. Represented are resistance, resilience and rebellion. In addition to the tone and context of conversations being important, the findings in the research suggest that by adhering to a feminist cosmopolitan ontology, encouraged is a sensitive, ethical encounter with others. This approach suggests that those marginalised in the academy and the artworld, then are seen and heard. Their voice and artwork are visualised as challenges to the norms of the academy.

AI-based chatbots in conversational commerce and their effects on product and price perceptions.

The rise of AI-based chatbots has gradually changed the way consumers shop. Natural language processing (NLP) technology and artificial intelligence (AI) are likely to accelerate this trend further. However, consumers still prefer to engage with humans and resist chatbots, which are often perceived as impersonal and lacking the human touch. While the predominant tendency is to make chatbots appear more humanlike, little is known about how anthropomorphic verbal design cues in chatbots influence perceived product personalization and willingness to pay a higher product price in conversational commerce contexts. In the current work, we set out to test this through one pre-test (N = 135) and two online experiments (N = 180 and 237). We find that anthropomorphism significantly and positively affects perceived product personalization, and that this effect is moderated by situational loneliness. Moreover, the results show that the interaction between anthropomorphism and situational loneliness has an impact on the willingness to pay a higher product price. The research findings can be used for future applications of AI-driven chatbots where there is a need to provide personalized and data-driven product recommendations.

Mental health promotion as perceived by Norwegian adolescents with somatically ill parents – an interpretative phenomenological analysis

ABSTRACT Objective Having a somatically ill parent can influence adolescents’ daily lives and mental health. This study aimed to explore the lived experiences of mental health promotion among adolescents with a somatically ill parent, based on a salutogenic orientation. Methods Individual interviews with 11 adolescents (13–18 years of age) with a somatically ill parent were conducted. The data were analysed according to interpretative phenomenological analysis. Results A superordinate theme, feeling at home in a house of mental health–promoting conversations, frames the participants’ perceptions of a prominent mental health promotion experience, including decisive characteristics of the significant conversation partners and different important conversation contexts. The metaphor of feeling at home expresses that the participants ascribe the conversations taking place there as promoting mental health. The themes elucidate the superordinate theme: (a) significant conversation partners characterised by the three subthemes: (i) being available, (ii) being competent and (iii) being caring and (b) rooms reflecting conversation contexts containing four subthemes: (i) room of increased knowledge, (ii) room of disclosure, (iii) room for meeting points and (iv) room for breaks. Conclusion Adolescents with a somatically ill parent perceived that conversations about important topics with significant others with special characteristics in distinct contexts promoted mental health.

Cultivating Relationship-Centered Care: Patient, Caregiver, and Provider Communication Preferences for and Experiences with Prognostic Conversations

ABSTRACT Prognostic conversations present many challenges for patients, caregivers, and providers alike. Most research examining the context of prognostic conversations have used a more siloed approach to gather the range of perspectives of those involved, typically through the lens of patient-centered care. However, the mutual influence evident in prognostic conversations suggests a relationship-centered care model may be useful in cancer communication research. Similarities and differences in preferences for and experiences with prognostic conversations among oncology patients, caregivers, and providers (N = 32) were explored. Identified themes were then mapped to the principles of the relationship-centered care framework to extend our understanding of prognostic conversations and contribute to a new direction in the application of relationship-centered care. Findings suggest fewer similarities than differences, point to important discrepancies among participant perspectives, and reinforce the utility of relationship-centered care in identifying communication practices that enhance the prognostic conversation experience.

Clear speech promotes speaking rate normalization.

When speaking in noisy conditions or to a hearing-impaired listener, talkers often use clear speech, which is typically slower than conversational speech. In other research, changes in speaking rate affect speech perception through speaking rate normalization: Slower context sounds encourage perception of subsequent sounds as faster, and vice versa. Here, on each trial, listeners heard a context sentence before the target word (which varied from "deer" to "tier"). Clear and slowed conversational context sentences elicited more "deer" responses than conversational sentences, consistent with rate normalization. Changing speaking styles aids speech intelligibility but might also produce other outcomes that alter sound/word recognition.

The linguistic complexity of adult and child contextualized and decontextualized talk.

The association between decontextualized talk (DT; i.e., talk extending beyond immediate context) and child language outcomes is well-attested but not well-understood. This study tested the hypothesis that DT is more linguistically complex than contextualized talk (CT). Thirty-eight Norwegian children (Mage = 5.5 years; 25 girls; 30 Norwegian-speaking monolinguals and eight multilinguals) and their teachers were videotaped during picture book reading, story card conversations and toy play (collected 2010-2011 and 2017). Results show that DT was more complex than CT among children and teachers. Both types of talk were more complex during book reading and story conversations than during play. The conversational context should be accounted for when theorizing about the role of DT in language development.

Exploring Bi-Directional Context for Improved Chatbot Response Generation Using Deep Reinforcement Learning

The development of conversational agents that can generate relevant and meaningful replies is a challenging task in the field of natural language processing. Context and predictive capabilities are crucial factors that humans rely on for effective communication. Prior studies have had a significant limitation in that they do not adequately consider the relationship between utterances in conversations when generating responses. This study aims to address this limitation by proposing a novel method that comprehensively models the contextual information of the current utterance for response generation. A commonly used approach is to rely on the information of the current utterance to generate the corresponding response, and as such it does not take advantage of the context of a multi-turn conversation. In our proposal, different from other studies, we will use a bi-directional context in which the historical direction helps the model remember information from the past in the conversation, while the future direction enables the model to anticipate its impact afterward. We combine a Transformer-based sequence-to-sequence model and the reinforcement learning algorithm to achieve our goal. Experimental results demonstrate the effectiveness of the proposed model through qualitative evaluation of some generated samples, in which the proposed model increases 24% average BLEU score and 29% average ROUGE score compared to the baseline model. This result also shows that our proposed model improves from 5% to 151% for the average BLEU score compared with previous related studies.

Pre-sizing of a modular high power density DC/DC converter with GaN components

Finding the most appropriate architecture to achieve a function is a major engineering challenge. In the power conversion context, the question that arises is: what is the best trade-off between an architecture based on a single high-power converter and another based on the combination of multiple low-power converters? This article aims to answer this question in the specific case of the isolated interface between the low voltage (28VDC) and high voltage (±270VDC) buses of an aircraft. Towards this goal, the targeted power electronics converter is carefully modeled in order to obtain an accurate and fast computing model. A technological database of the different usable components is then used to feed an optimization algorithm. The execution of the latter achieves an attractive and robust result showing an excellent performance in terms of power-to-weight ratio, which is the key index of this study. The precision and speed of this computer-aided design is based on analytical models that are quick to run and therefore enable exploring almost exhaustively the search space. More specifically, the high-frequency transformer has a large relative mass and generates significant losses that are difficult to assess. A major modeling effort has been undertaken and has been enabled to define a simplified but convincing and accurate model (successfully assessed on a wide bandwidth using an experimental setup). It was used for the first time in this optimization process. According to the targeted aeronautical specifications, the two best solutions would make it possible to achieve a mass power density of 4.5kW/kg, i.e. twice as high as traditional solutions. This result very clearly shows that the appropriate use of the new GaN transistors makes it possible to make a technological breakthrough. Considering the voltage and current ratings of these components, this shows that the combination of multiple partial converters is very promising and will make it possible in the future to achieve a significant increase in the compactness of electrical power conversion functions. This sizing study therefore clearly shows the potential of the standardized design of power electronics converters and the search for the best combinations (series, parallel) to meet any specific specifications. Additionally, the developed approach, based on a modeling effort, especially as far as the high-frequency transformer is concerned, and collection of manufacturer data, also makes it possible to limit the implementations and associated tests.Hence, the main contributions of this article are (i) the assembly of the analytical models of the various components constituting an isolated DC–DC power supply, (ii) the simplification of an analytical model of the planar HF transformer, its experimental validation and its first use in a loop of optimization, (iii) the fact of showing supporting figures that GaN technology enables technological breakthroughs to achieve high power density, which opens the way to modular architectures based on partial converters of electrical power, (iv) the comprehensive description of a powerful pre-study design tool enabling to greatly reduce power converters’ time to market.

The Effect of Familiarity on Conversation Turns in Dyadic Conversation Among Chinese Older Adults

This study explored the effect of mutual familiarity of interlocutors on quantitative contributions of conversation turns in dyadic conversation among Mandarin Chinese-speaking older adults. A quantitative quasi-experimental study was conducted. A total of 42 healthy older adults aged 65 years or over were recruited. Percentages of contributed conversation turns for each interlocutor were computed as frequency of interlocutor conversation turns divided by total frequency of dyad conversation turns multiplied by 100. Quantitative asymmetries were differences of percentages of contributed conversation turns. A total of 60 ten-minute dyadic conversation sessions were conducted, including 30 mutually familiar-older-adult sessions (FOAS) and 30 mutually unfamiliar-older-adult sessions (UOAS). Quantitatively asymmetrical contributions of conversation turns occurred in both FOAS and UOAS, and quantitative contributions of conversation turns in FOAS were significantly more asymmetrical than those in UOAS. There were three limitations to the current study, including limited representations of everyday conversation contexts (e.g., at home); no consideration of the types of conversation modes; and no consideration of sensitivity to conversation as one of the inclusion criteria for research participants. Quantitatively asymmetrical contributions of conversation turns occurred in both mutually FOAS and UOAS dyads among Mandarin Chinese-speaking older adults. Moreover, quantitative contributions of conversation turns in mutually FOAS dyads were significantly more asymmetrical than those in mutually UOAS dyads. Sufficient knowledge of changes in conversation turns in dyadic conversation among healthy older adults might reduce misperceptions of older adults suffering from neurological disease.