Majority-vote LLM ensembles improve accuracy monotonically with ensemble size when agents are homogeneous samples of the same prompt (Li et al., 2024). We test whether this holds for four heterogeneous clinical-role prompts on one frozen MedGemma backbone, across binary tumor detection, tumor subtyping, MS, and stroke triage. Our comparison is against single-agent triage: we ask whether role-diverse voting clears that bar, not whether it matches sampling diversity, since we do not run a paired homogeneous-sampling forest. On a paired 500-image evaluation per task, a four-agent forest does not reliably beat single-agent triage: significantly worse on binary tumor (both stages) and MS end-to-end, indistinguishable on tumor subtyping and stroke. Pairwise error correlation between roles is far from independent on every task (φ = 0.43–0.75 vs. φ = 0 under independence), explaining why voting fails to average errors out. Roles diverge behaviorally on the easiest task but collapse onto a shared “normal” bias on the two hardest, so voting cannot recover sensitivity no member has. Dissent rate, a candidate escalation signal, is valid on two tasks, uninformative on MS, and inverted on stroke, where agreement is less reliable than disagreement. These results argue against assuming role diversity on a shared backbone helps by default, and for task-conditional validation before deployment.

Text style transfer is the task of modifying the stylistic attributes of a given text while preserving its original meaning. This task has also gained interest with the advent of large language models. Although knowledge graph augmentation has been explored in various tasks, its potential for enhancing text style transfer has received limited attention. This paper proposes a method to create a Style Knowledge Graph (SKG) to facilitate and improve text style transfer. The SKG captures words, their attributes, and relations in a particular style, that serves as a knowledge resource to augment text style transfer. We conduct baseline experiments to evaluate the effectiveness of the SKG for augmenting text style transfer by incorporating relevant parts from the SKG in the prompt. The preliminary results demonstrate its potential for enhancing content preservation and style transfer strength in text style transfer tasks, while the results on fluency indicate promising outcomes with some room for improvement. We hope that the proposed SKG and the initial experiments will inspire further research in the field.

The viral spread of false, unverified and misleading information on the Internet has attracted a heightened attention of an interdisciplinary research community on the phenomenon. This paper contributes to the research efforts of automatically determining the veracity of rumourous tweets and classifying their replies according to stance. Our research objective was to investigate the interplay between a number of phenomenological and contextual features of rumours, in particular, we explore the extent to which network structural characteristics, metadata and user profiles could complement the linguistic analysis of the written content for the task at hand. The current findings strongly demonstrate that supplementary sources of information play significant role in classifying the veracity and the stance of Twitter interactions deemed to be rumourous.

This paper reports an experiment carried out to investigate the relevance of several syntactic, stylistic and pragmatic features on the task of distinguishing between mainstream and partisan news articles. The results of the evaluation of different feature sets and the extent to which various feature categories could affect the performance metrics are discussed and compared. Among different combinations of features and classifiers, Random Forest classifier using vector representations of the headline and the text of the report, with the inclusion of 8 readability scores and few stylistic features yielded best result, ranking our team at the 9th place at the SemEval 2019 Hyperpartisan News Detection challenge.