Journal

Volume 25, Issue 2 (June 30, 2024)

4 articles

• Predicting Affective Responses based on Physiological Data using Regression-based Decoding
  by Hyeonjung Kim & Jongwan Kim
  J. CS. 2024, 25(2), 179-198;
  Abstract Physiological responses have been regarded as better measures of arousal than valence of core affect dimensions. Recently, valence was explained by two valence hypotheses (bipolarity and bivalent), represented as signed and unsigned valence. In this paper, with reference to two valence models and an... [Read more].
  Abstract Physiological responses have been regarded as better measures of arousal than valence of core affect dimensions. Recently, valence was explained by two valence hypotheses (bipolarity and bivalent), represented as signed and unsigned valence. In this paper, with reference to two valence models and an arousal dimension, we explored whether affective representations can be identified based on physiological responses across participants. By re-analyzing a shared dataset that includes physiological responses and behavioral ratings of affect, we performed a regression-based decoding to predict affective representations of participants based on their physiological responses. Additionally, a one-way repeated measures ANOVA with a trend analysis was conducted to compare the accuracies of two valence models and the arousal dimension. The results revealed that all predictions were significant, indicating that physiological responses can identify valence and arousal. A trend analysis showed that arousal was predicted more accurately than two valence models, suggesting that physiological responses capture arousal better than valence. [Collapse]
• Refined Word Embeddings with Intensity Awareness for Fine-Level Sentiment Classification
  by Prashantkumar M. Gavali & Suresh K. Shirgave
  J. CS. 2024, 25(2), 199-236;
  Abstract Sentiment analysis employs classification models to discern people's opinions automatically. Recent strides made with Large Language Models (LLMs) have significantly enhanced the accuracy of binary-level sentiment classification, particularly through zero-shot and few-shot learning approaches. Howev... [Read more].
  Abstract Sentiment analysis employs classification models to discern people's opinions automatically. Recent strides made with Large Language Models (LLMs) have significantly enhanced the accuracy of binary-level sentiment classification, particularly through zero-shot and few-shot learning approaches. However, when it comes to fine-level sentiment classification, LLMs face challenges because they are not specifically trained for this downstream task. In contrast, other classification models utilize word embedding, a vector representation of words, as input data. Contemporary word embedding algorithms create these embeddings by considering the surrounding context of each word. Nonetheless, these word embeddings often fail to capture the nuances of intensity differences between words. For example, words like 'more' and 'less' have embeddings closely positioned in the semantic space, despite representing distinct intensity levels. These intensity words such as 'much', 'more', and 'less' are frequently used to convey the strength of opinions. Their intensity distinctions are crucial in fine-level sentiment classification. This paper introduces an innovative intensity-aware feed-forward neural network, equipped with a novel referential loss function designed to capture these intensity differences between words. The proposed model effectively separates words of varying intensities while bringing together words sharing the same intensity in the semantic space. To assess the effectiveness of this refined word embedding in sentiment analysis tasks, diverse fine-level sentiment datasets are employed. The results demonstrate that the refined word embedding surpasses original embeddings and popular LLMs for fine-level sentiment analysis. [Collapse]
• Bayesian Joint Modeling of Item Response and Response Time in a Statistical Learning Task
  by Jinglei Ren & Hong Jiao
  J. CS. 2024, 25(2), 237-274;
  Abstract This study utilized a hierarchical Bayesian joint modeling approach to concurrently analyze item responses with the Rasch model and response time (RT) data with a lognormal model in a statistical learning task. Further, different models of RT including the Boxcox, Exponential, and Gamma RT distribut... [Read more].
  Abstract This study utilized a hierarchical Bayesian joint modeling approach to concurrently analyze item responses with the Rasch model and response time (RT) data with a lognormal model in a statistical learning task. Further, different models of RT including the Boxcox, Exponential, and Gamma RT distributions were empirically compared. Response accuracy (RA) based on the Rasch-Only model was compared with that based on joint models. The parameter estimation for all models was performed using Markov chain Monte Carlo methods. The results indicated that the Rasch-Boxcox model was the best-fitting joint model. The correlation between the item difficulty and item speed parameters as well as the correlation between the person ability and person speed parameters were both negative in the Rasch-Boxcox joint models, which led to smaller standard errors in both item difficulty and ability parameter estimates in the joint modeling compared to the Rasch-Only model, indicating the auxiliary information from RT helps improve the measurement precision of RA. [Collapse]
• Request: A Kuwaiti Cultural Study of Speech Act Realization Patterns (KCSSARP)
  by Dhari AlOtaibi, Shamlan AlQenaie, and Soonhyuck Park
  J. CS. 2024, 25(2), 275-308;
  Abstract Expressions for requests are known to exhibit various patterns depending on the language. This research analyzes the various ways requests are made in Kuwaiti-spoken Arabic, which has received comparatively limited attention to the relevant research. The study employs a Discourse Completion Test (DC... [Read more].
  Abstract Expressions for requests are known to exhibit various patterns depending on the language. This research analyzes the various ways requests are made in Kuwaiti-spoken Arabic, which has received comparatively limited attention to the relevant research. The study employs a Discourse Completion Test (DCT) to examine how native Kuwaiti speakers use different request techniques in various social situations. Based on the results, the findings are divided into three social groups: High Ranking to Low Ranking (HR-LR), Equal Ranking to Equal Ranking (ER-ER), and Low Ranking to High Ranking (LR-HR). The study demonstrates significant differences in the choice of request strategies based on the social status of the individuals involved, indicating the impact of social hierarchy on language usage in Kuwaiti Arabic. This research contributes to our understanding of pragmatics in the field of Arabic sociolinguistics, specifically concerning speech act theory and interlanguage pragmatics. [Collapse]