This paper presents a multimodal human-robot interaction (HRI) system for educational contexts implemented on the humanoid robot Pepper. The system leverages multiple communicative channels, allowing learners to combine speech with tablet interaction while the robot responds through synchronized speech, textual captions and dynamic visual cues. To ensure robustness and reliability, we introduce the use of Metamorphic Testing for multimodal HRI. By validating system behavior through systematic input transformations, we demonstrate how metamorphic testing can uncover inconsistencies across linguistic, visual and cross-modal interactions. This work contributes both a novel methodological framework for evaluating multimodal HRI systems and an application to educational robotics.

This paper presents an experiment on the effects of inter-agents emotional alignment, a prerequisite for empathic communication, in Human-Robot Interaction (HRI). We describe a pipeline built around the Pepper robot with the idea of verifying the effect of emotionally-aligned communication toward a user. In particular, our goal is twofold, in that we investigate if and to what extent an emotionally-aligned, empathic dialogue impacts on (i) the communication style of the user, and (ii) the persuasive effectiveness of the robot, intended as its ability to alter or reinforce its interlocutors' attitudes and beliefs about the conversation topic. Both these aspects have been assessed in a controlled experiment with 46 participants, comparing a condition where the robot addresses participants with emotionally neutral sentences with a condition where the robot provides answers tailored to the emotions expressed in participants' input utterances. Results show how emotion alignment acts as an effective trigger for the elicitation of different communication styles of the users but also that, contrary to what we expected, it does not play any persuasive effect.

The paper presents an experiment on the effects of adaptive emotional alignment between agents, considered a prerequisite for empathic communication, in Human-Robot Interaction (HRI). Using the NAO robot, we investigate the impact of an emotionally aligned, empathic, dialogue on these aspects: (i) the robot’s persuasive effectiveness, (ii) the user’s communication style, and (iii) the attribution of mental states and empathy to the robot. In an experiment with 42 participants, two conditions were compared: one with neutral communication and another where the robot provided responses adapted to the emotions expressed by the users. The results show that emotional alignment does not influence users’ communication styles or have a persuasive effect. However, it significantly influences attribution of mental states to the robot and its perceived empathy.

This paper describes a corpus consisting of real-world dialogues in English between users and a task-oriented conversational agent, with interactions revolving around the description of finite state automata. The creation of this corpus is part of a larger research project aimed at developing tools for an easier access to educational content, especially in STEM fields, for users with visual impairments. The development of this corpus was precisely motivated by the aim of providing a useful resource to support the design of such tools. The core feature of this corpus is that its creation involved both sighted and visually impaired participants, thus allowing for a greater diversity of perspectives and giving the opportunity to identify possible differences in the way the two groups of participants interacted with the agent. The paper introduces this corpus, giving an account of the process that led to its creation, i.e. the methodology followed to obtain the data, the annotation scheme adopted, and the analysis of the results. Finally, the paper reports the results of a classification experiment on the annotated corpus, and an additional experiment to assess the annotation capabilities of three large language models, in view of a further expansion of the corpus.

Neural networks are notoriously data-hungry.This represents an issue in cases where data are scarce such as in low-resource languages.Data augmentation is a technique commonly used in computer vision to provide neural networks with more data and increase their generalization power.When dealing with data augmentation for natural language, however, simple data augmentation techniques similar to the ones used in computer vision such as rotation and cropping cannot be employed because they would generate ungrammatical texts.Thus, data augmentation needs a specific design in the case of neural logic-to-text systems, especially for a structurally rich input format such as the ones used for meaning representation.This is the case of the neural natural language generation for Discourse Representation Structures (DRS-to-Text), where the logical nature of DRS needs a specific design of data augmentation.In this paper, we adopt a novel approach in DRS-to-Text to selectively augment a training set with new data by adding and varying two specific lexical categories, i.e. proper and common nouns.In particular, we propose using WordNet supersenses to produce new training sentences using both in-and out-of-context nouns.We present a number of experiments for evaluating the role played by augmented lexical information.The experimental results prove the effectiveness of our approach for data augmentation in DRS-to-Text generation.Exp.No Implementation Type Precision Recall F1-Score 01 Gold (without augmentation) 95.2 95.4 95.3 02 Gold + PN (inside context) 95.8 95.9 95.8 03 Gold + PN (outside context) 95.9 95.9 95.9 04 Gold + CN (inside context with SS) 95.7 95.7 95.7 05 Gold + CN (inside context without SS) 95.7 95.5 95.6 06 Gold + CN (outside context with SS) 95.5 95.8 95.7 07 Gold + CN (outside context without SS) 95.8 95.7 95.7 08 Gold + PN-with-CN 96.1 95.9

In this article, we investigate the case of human-machine dialogues in the specific domain of commercial customer care. We built a corpus of conversations between users and a customer-care chatbot of an Italian Telecom Company, focusing on a sample of conversations where users contact the service asking for explanations about billing issues or overcharges. We observed that users’ requests are often vague, generic or incomprehensible. In such cases, commercial dialogue systems typically ask for clarifications or further details to fully understand users’ specific requests. However, from the corpus analysis it appeared that chatbot's clarifying requests may result in ineffective interactions, with users eventually giving up the conversation or switching to a human agent for a faster query resolution. A recovery strategy is thus needed to anticipate users’ information needs, or intentions. We address this issue resorting to GEN-DS, a dialogue system based on symbolic data-to-text generation. GEN-DS analyzes the user-company contextual relational knowledge, with the aim to generate more relevant answers to unclear questions. In this article, we describe the GEN-DS architecture along with the experiments we carried out to evaluate its output. Results from an offline human evaluation show significant improvements of GEN-DS compared to the original system. These improvements concern properties such as utility, necessity, understandability, and quickness of the information communicated in the dialogue. We believe that GEN-DS techniques may find application in all the dialogue systems that need to manage vague requests and must rely on relational knowledge.

Artificial intelligence largely relies on bounds on differences (BoDs) to model binary constraints regarding different dimensions, such as time, space, costs, and calories. Recently, some approaches have extended the BoDs framework in a fuzzy, “noncrisp” direction, considering probabilities or preferences. While previous approaches have mainly aimed at providing an optimal solution to the set of constraints, we propose an innovative class of approaches in which constraint propagation algorithms aim at identifying the “space of solutions” (i.e., the minimal network) with their preferences, and query answering mechanisms are provided to explore the space of solutions as required, for example, in decision support tasks. Aiming at generality, we propose a class of approaches parametrized over user-defined scales of qualitative preferences (e.g., Low, Medium, High, and Very High), utilizing the resume and extension operations to combine preferences, and considering different formalisms to associate preferences with BoDs. We consider both “general” preferences and a form of layered preferences that we call “pyramid” preferences. The properties of the class of approaches are also analyzed. In particular, we show that, when the resume and extension operations are defined such that they constitute a closed semiring, a more efficient constraint propagation algorithm can be used. Finally, we provide a preliminary implementation of the constraint propagation algorithms.

People with sight impairments can access to a mathematical expression by using its L A T E X source.However, this mechanisms have several drawbacks: (1) it assumes the knowledge of the L A T E X, (2) it is slow, since L A T E X is verbose and (3) it is error-prone since L A T E X is a typographical language.In this paper we study the design of a natural language generation system for producing a mathematical sentence, i.e. a natural language sentence expressing the semantics of a mathematical expression.Moreover, we describe the main results of a first human based evaluation experiment of the system for Italian language.

This paper presents a project about the automatic generation of persuasive messages in the context of the diet management.In the first part of the paper we introduce the basic mechanisms related to data interpretation and content selection for a numerical data-to-text generation architecture.In the second part of the paper we discuss a number of factors influencing the design of the messages.In particular, we consider the design of the aggregation procedure.Finally, we present the results of a human-based evaluation concerning this design factor.

This paper describes the SimpleNLG-IT realiser, i.e. the main features of the porting of the SimpleNLG API system (Gatt and Reiter, 2009) to Italian.The paper gives some details about the grammar and the lexicon employed by the system and reports some results about a first evaluation based on a dependency treebank for Italian.A comparison is developed with the previous projects developed for this task for English and French, which is based on the morpho-syntactical differences and similarities between Italian and these languages.

One open problem in the AI & Law community is how to provide computers with a basic understanding of legal concepts, and their relationship with legal texts and with the legal lexicon. We propose to add a layer to connect the linguistic description of the provisions to syntactic patterns using FramNet that can be exploited thought NLP tools. A deep-parsing and shallow-semantics approach has been devised to interpret and retrieve the characterizing components of legal modificatory provisions. In this paper we single out the case of efficacy suspension and show how FrameNet approach can provide profit especially to isolate temporal parameters and their interpretation.

In this paper we illustrare a research based on NLP techniques aimed at automatically annotate modificatory provisions. We propose an approach which pairs deep syntactic parsing with rule-based shallow semantic analysis relying on a fine-grained taxonomy of modificatory provisions. The implemented system is evaluated on a large dataset hand-crafted by legal experts; the results are discussed and future directions of the research outlined.

We are concerned with the automatic semantic interpretation of legal modificatory provisions. We propose a novel approach which pairs deep syntactic parsing and a fine-grained taxonomy of legal modifications. Although still in a developmental stage, the implemented system can be used to annotate with meta-information modificatory provisions of NormaInRete documents.

This paper describes the philosophy behind our tool called "Legal Taxonomy Syllabus", the analytical instruments it provides and some case studies. The Legal Taxonomy Syllabus is an ontology based tool designed to annotate and recover multi-lingua legal information and build conceptual dictionaries. The Legal Taxonomy Syllabus allows to build legal dictionaries in a bottom up fashion starting from the annotation of legal terms by legal terminological experts and to let legal ontology engineers refine the resulting taxonomies of concepts. The Legal Taxonomy Syllabus and its analytical tools provide help to lawyers to study the peculiarities of European Union Directives concerning the polysemy of legal terms, and the terminological and conceptual misalignment. By means of two case studies we show how the Legal Taxonomy Syllabus can help the processes of drafting and translating of the Directives.