Adaptability to the audience is an important feature for conversational systems, especially in the healthcare dissemination field, where scientific concepts have to be delivered to a potentially wide range of users. This work presents an evaluation of the capability of LLMs to support style transfer according to the target user’s age group. Two complementary evaluation methods were employed: an automatic assessment based on the ARI readability index, and a human experts evaluation focusing on appropriateness depending on the user’s educational level as well as content accuracy. Results show that LLMs efficiently switch style when provided with information about the user’s age while managing content still requires the adoption of safety measures.

This paper explores the application of the Influence Diagrams model for decision-making in the context of conversational agents. The system consists of a Conversational Recommender System (CoRS), in which the decision-making module is separate from the language generation module. It provides the capability to evolve a belief based on user responses, which in turn influences the decisions made by the conversational agent. The proposed system is based on a pre-existing CoRS that relies on Bayesian Networks informing a separate decision process. The introduction of Influence Diagrams aims to integrate both Bayesian inference and the dialogue move selection phase into a single model, thereby generalising the decision-making process. To test the effectiveness and plausibility of the dialogues generated by the developed CoRS, a dialogue simulator was created and the simulated interactions were evaluated by a pool of human judges.

In linguistics, research on dialogue systems has accentuated the need to focus on various pragmatic aspects for their management and modelling. Among the most important pragma-linguistic speech acts in dialogue systems studies are Clarification Requests, corrective feedback that in some circumstances require access to the set of shared knowledge known as Common Ground. Regarding Common Ground management, pragmatic studies suggest differences in the type of polar questions that people prefer be used in Clarification Requests, where polar questions can have two possible answers: true or false. This preference appears to depend on the relationship between bias and contextual evidence. In this work, we show that varying the form of polar questions in a given pragmatic setting can influence the capability of people to track Common Ground inconsistencies. As a result, we demonstrate that using a negative polar question in Italian has functional consequences when communicating conflicting material in the Common Ground. This can improve the quality of human interactions with dialogue systems, in terms of an improved identification of the conflict. The results obtained in this work provide insights into design of error reporting approaches in natural interactions.

Commonsense Knowledge (CSK) is defined as a complex and multifaceted structure, encompassing a wide range of knowledge and reasoning generally acquired through everyday experiences. As CSK is often implicit in communication, it poses a challenge for AI systems to simulate human-like interaction. This work aims to deepen the CSK information structure from a linguistic perspective, starting from its organisation in conversations. To achieve this goal, we developed a three-level analysis model to extract more insights about this knowledge, focusing our attention on the second level. In particular, we aimed to extract the distribution of explicit actions and their execution order in the communicative flow. We built an annotation scheme based on FrameNet and applied it to a dialogical corpus on the culinary domain. Preliminary results indicate that certain frames occur earlier in the dialogues, while others occur towards the process’s end. These findings contribute to the systematic nature of actions by establishing clear patterns and relationships between frames.

In dialogue analysis, characterising named entities in the domain of interest is relevant in order to understand how people are making use of them for argumentation purposes. The movie recommendation domain is a frequently considered case study for many applications and by linguistic studies and, since many different resources have been collected throughout the years to describe it, a single database combining all these data sources is a valuable asset for cross-disciplinary investigations. We propose an integrated graph-based structure of multiple resources, enriched with the results of the application of graph analytics approaches to provide an encompassing view of the domain and of the way people talk about it during the recommendation task. While we cannot distribute the final resource because of licensing issues, we share the code to assemble and process it once the reference data have been obtained from the original sources.