Chatbots are increasingly part of digital customer service, yet their evaluation is often reduced to response speed, availability, or automation rate. This study assumes that customers do not experience the algorithm directly, but the language through which the system communicates. The empirical analysis is based on survey data from 411 respondents. Composite indices were constructed for language quality, trust, satisfaction, service communication performance, and customer re-engagement intention, all showing acceptable internal consistency. Regression models indicate that language quality is positively associated with trust ( β =0.283;p<0.001 ). Language quality and trust jointly explain 50.7 β =0.287;p<0.001 ), while the direct relationship between trust and re-engagement becomes non-significant after satisfaction is included. Mediation results support a sequential mechanism in which language quality is associated with re-engagement via trust and satisfaction. The study shifts chatbot evaluation from technical efficiency toward language quality as part of managing digital service operations.
This study examines the determinants of repurchase intention in the beauty clinic industry by developing an integrated framework linking perceived service quality, perceived price fairness, and influencer credibility with brand trust, and by testing brand trust as a mediating mechanism. Survey data were collected from 360 beauty clinic consumers in the Jabodetabek metropolitan area, Indonesia, using purposive sampling. The proposed model was tested using covariance-based structural equation modelling (SEM) with AMOS, with bootstrapping used to assess the mediating role of brand trust. Perceived service quality, perceived price fairness, and influencer credibility each had significant positive effects on brand trust and repurchase intention (all p < 0.05). Perceived service quality was the strongest predictor of brand trust (β = 0.399), while perceived price fairness was the strongest direct predictor of repurchase intention (β = 0.332). Brand trust significantly influenced repurchase intention (β = 0.296) and partially mediated all three antecedents, most strongly for perceived service quality (indirect β = 0.117). The model explained 30.6
Abstract This paper investigates an efficient approach for distilling Large Language Models (LLMs) into smaller, application-specific models using zero-shot Chain of Thought (CoT) rationale generation and Optimization by Prompting (OPRO). To address the challenges of deploying computationally intensive generative AI for narrow tasks or resource-constrained environments, the approach leverages LLM reasoning capabilities to generate both labels and natural language explanations for unlabeled data. By reducing reliance on human-generated annotations, the approach substantially lowers annotation requirements and prompting costs while maintaining comparable performance in the evaluated settings. We formulate distillation as a multi-task learning problem in which student models are trained to jointly predict labels and learn from teacher-generated rationales, with the goal of improving data efficiency and generalization. Building on established zero-shot Chain of Thought (CoT) prompting and the OPRO prompt optimization technique, we use teacher-generated rationales to reduce annotation token requirements and examine the associated performance and efficiency gains. Additionally, we systematically investigate how explanation properties affect distillation efficiency. Across natural language inference and question answering benchmarks, results indicate that near-optimal performance can be achieved even when rationales are provided for only a subset of the training data, and that shorter explanations are often sufficient. These findings provide practical insights into the trade-offs between rationale generation cost and student model performance. Overall, this work contributes empirical evidence on the effectiveness and cost characteristics of rationale-based distillation for training compact, task-specific language models with minimal human intervention.
While existing literature establishes that patent quality drives firm performance better than mere patent counts, the boundary conditions of this consensus remain underexplored under severe macroeconomic stress. This study investigates the relationship between patent quantity and quality and firm profitability – in terms of return on assets (ROA) and return on equity (ROE) – within the strategically vulnerable, capital-intensive Western European electronics manufacturing sector over the 2016–2024 period. Using the European Patent Office’s PATSTAT database and EMIS, we analyse a panel of 112 firms via two-way fixed-effects regressions to test how extreme economic disruptions impact innovation premiums. Our baseline findings confirm that under stable macroeconomic conditions, patent quality – measured by average forward citations – acts as a robust exogenous driver of operating and equity profitability, whereas sheer patent quantity does not. Crucially, however, we demonstrate that the systemic post-2020 economic shock and subsequent disruptions completely neutralised this established quality premium. This study provides a novel theoretical contribution by showing that the superiority of high-quality intellectual property is not absolute; rather, it is strictly contingent upon stable macroeconomic environments. By highlighting these vulnerabilities, our findings offer critical insights into the limitations of patent-driven profitability during periods of systemic economic uncertainty.
The article examines the relationship between population body dimensions and the practical design of wooden furniture, with a particular focus on wooden chairs intended for bariatric individuals. Given the growing prevalence of obesity, there is an increasing need to reassess the ergonomic, dimensional, and strength requirements of seating furniture for this population group. The aim of the study is to analyse selected body dimensions of bariatric respondents in Slovakia and to identify their opinions and preferences regarding seating furniture designed for overweight users. The research was based on a comprehensive analysis of selected body dimensions of 319 bariatric respondents in Slovakia. In addition, a structured questionnaire survey was conducted to assess 350 respondents’ preferences and their level of satisfaction with the existing portfolio of wooden seating furniture. The findings indicate that the current portfolio of wooden furniture does not sufficiently meet the needs of bariatric users. Respondents emphasized the need for improved chair dimensions and greater structural strength. The results also show consumer interest in furniture adapted to above-average body weight and dimensions. The study confirms the necessity of redesigning wooden seating furniture to better accommodate bariatric individuals. Special attention should be paid to ergonomic parameters, load-bearing capacity, and appropriate dimensions in order to increase user comfort, safety, and satisfaction.
