Artworks’ Features Discovery Through Engaging Conversations for Children

Abstract

Dialog-based interfaces have seen a renewed interest in the last years. Smart assistants like Siri, Google Assistant or Cortana are pervasive nowadays. Messaging platforms like Facebook Messenger, Telegram or Skype are integrating automatic responders commonly known as bots or chatbots. Industries started incorporating smart assistants as web-services to help, retain and profile their customers. We see a considerable potential in using conversational interfaces to enhance guided tours in cultural institutions. Museums user experiences may vary depending on the exhibition. Nonetheless, hearing is the common channel museums use to inform visitors, in particular by exploiting audio-guides. The existing guides are very well structured and provide indeed a complete guided experience but are generally intended for adults and neuro-typical users. As a consequence, museums may have difficulties in targeting all kinds of visitors, e.g. children or users with cognitive disability. They may require assistance to pass over their heritage to all audiences. Our research purpose is to design an infrastructure to help cultural institutions in broadcasting their heritage to all guests. We propose interactivity over information: the user takes an active part to the visit and is engaged though the whole experience. Using conversational interfaces, we switch from the perspective of a passive listener to the point of view of a dynamic "player": the user can ask for information. The engagement is a necessary aspect to maintain a high interest in children. We explore how bi-directional conversations can provide accessible pieces of information to users with special needs. We design a plug-and-play pervasive solution for a seamless adoption for the interested institutions. We allow holders of the system to fine-tune its characteristics based on the artwork, the exhibition and the target users.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.