Designing Interactive Experiences to Explore Artwork Collections: a Multimedia Dialogue System Supporting Visits in Museum Exhibits Antonio Origlia 1,2 , Enrico Leone 1 , Antonio Sorgente 2 , Paolo Vanacore 2 , Maria Parascandolo 1 , Francesco Mele 2 , and Francesco Cutugno 1,2 1 PRISCA-Lab, Federico II University, Naples, Italy, {antonio.origlia, enrico.leone, maria.parascandolo, cutugno}@unina.it, 2 Institute of Applied Sciences and Intelligent Systems - CNR, Naples, Italy, {a.sorgente, p.vanacore, f.mele}@isasi.cnr.it Abstract. Speech and natural language processing have a central role in the implementation of systems designed to make the museum more reac- tive to users’ inputs and to improve the overall interaction quality. In this paper, we present the design and implementation of a dialogue system to provide multimedia presentations for museum visits. A corpus of speech recordings in Italian was collected with a mobile application to obtain a reference set of possible ways for the users to express their intentions. On the basis of this corpus, a set of recurring syntactic patterns associ- ated to device requests was extracted to let the dialogue system separate device commands from information queries. Disambiguation strategies depending on the context are also applied in presence of partial syntac- tic patterns. Information queries are answered by automatically assem- bling portions of semantically annotated texts and are synchronized with relevant multimedia resources. A case study on the ’800 exhibit at the Capodimonte museum in Naples is presented 3 . Keywords: dialogue systems, cultural heritage 1 Introduction Italian has a very poor visibility in the area of spoken dialogue systems basic research. The EVALITA evaluation campaign held in 2009 [1] showed the state of art for telephonic systems was limited to the features offered by VoiceXML standard. Participants in that evaluation campaign were able to set up three dif- ferent system initiative dialogue managers in the field of train services (ticketing, booking and timetable queries) but performances were found to be below the performance that is usually obtained on English. Also, recent dialogue systems for Italian equipped with semantic reasoning capabilities were presented in [2–4], but they only considers chat based interaction. In the passage from telephonic 3 This work is supported by the Italian PAC project Cultural Heritage Emotional Experience See-Through Eyewear (CHEESE).