Children, nowadays, are great consumers of media for them [6], and there is growing interest towards novel mechanisms that can consider their speci c needs and improve both the recommendation process and output of videos for them. Children, in fact, have unique characteristics, which change with age. In particular, in the 8{12 age range, they like challenging interactions with systems, they enjoy exploring and \to feel the experts". The majority of current recommendation solutions are unable to leverage on such speci c features of children, and others, when interacting with them and recommending adequate videos. In this paper, we introduce ITTuB, an exploratory project, set at the intersection of recommender system and interaction design for children. Starting from previous research in the two areas, it plans to introduce tangibles for enhancing the interaction of children with videos, and to leverage stylistic features of videos in order to deliver recommendations that are optimized for children. This workshop paper presents the main ideas of ITTuB.
The majority of existing movie-sharing applications tend to combine
collaborative ltering techniques with semantic features, such as the video genre, director,
cast, or tags and textual reviews, for recommending videos [
More in general, careful considerations of children's characteristics is
necessary for designing recommender systems that can e ectively interact with
children. Moreover, children are not a homogeneous group, e.g., their characteristics
change with age. While children become adults, they go through a number of
stages, which include unique cognitive, emotional, social and physical challenges,
that developers of interactive recommender systems for children must take into
consideration in order to make truly engaging systems for children. In
particular, tweens, in the age range 8{12, have their own speci c characteristics. For
example, richer interactions are preferred by tweens than by younger children,
but a large number of choices, which could be adequate to adults, still confuses
tweens and a ects the perceivability of the interactions; at the same time, play
is still a fundamental component of the design of interactions for tweens [
ITTuB is a project that, by carefully analysing the characteristics of tweens,
seeks to investigate how to design an interactive recommender system of videos
for them. The research around ITTuB is novel from two convergent viewpoints.
The design of the interaction of the system with tweens will itself be novel
because it will explore innovative interaction modalities based on playful tangible
objects (brie y, tangibles) for tweens. Fig. 1 shows an example of a playful
tangible progression board for 8{10 year tweens, using di erent micro-electronics
components for interacting with children [
The paper is organised as follows. It starts with related work and background in the two areas converging into ITTuB: recommender system design and interaction design and children. The paper then moves on explaining the main ideas of ITTuB. It does so by giving the main context scenarios of the project and its research goals. 2
Traditional video recommender systems rely mainly on the assumption that
users' preferences may related to the semantic features of videos (e.g. [
In ITTuB, we make a di erent assumption. We still believe that traditional video attributes are important to be used in information retrieval applications, where the goal is to provide a way of indexing multimedia content so that users can explicitly (i.e., manually) query that content at the semantic level. However, we wish to investigate if this assumption holds also for recommender systems which aim at automatically nding contents that the user prefers, without the user querying the system.
Therefore, our goal is to leverage the stylistic features automatically extracted
from video content and complement them with the semantic features so as to
deliver video recommendations. For this purpose, we explore and evaluate a
recent approach for movie recommendations that integrates traditional semantic
attributes, such as genre, director and cast, with stylistic features, such as
lighting, colors, background, and movements. Such aesthetic features are likley to
matter for recommending videos to tweens, who are 8{12 year olds with speci c
characteristics [
More in general, tweens tend to enjoy systems which are highly interactive.
They are growing up expecting things to give them immediate feedback about
their choices; compared to other generations, they are more skilled about
technologies, and have di ering abilities to express their ideas and to follow
structured tasks [
Hereby, we give preliminary scenarios concerning the intended usage of the ITTuB recommender system, and the research goals of ITTuB presented below.
SCENARIO 1: Amirali wants quality video recommendations. Amirali is a 12-year old boy, who often uses his IPad tablet for accessing videos over the Internet and posting his game play sessions. However, the recommended videos do not often satisfy his expectations. \If I watch a cartoon, it does not mean that I only want to watch cartoons all my life!"|thinks Amirali. Yet, that is what happens when he uses video recommender applications on his IPad. SCENARIO 2: Lucrezia rates her preferred videos.
Lucrezia is an 8-year old child, addicted to musical movies; she is utterly in love with Frozen! She watches them online, but all existing recommender systems annoy her with their continuous boring questions. One day, Emily, her literature teacher, shows her the ITTuB app and the tangible toolkit of ITTuB. She is reasonable, but cautious, about how her pupils uses apps, and has done a great job of educating them about the people they might encounter online. Emily has chosen the ITTuB app because it guarantees ethical treatment of users data. Lucrezia opens it because she is attracted by the tangible toolkit that comes with the app. The app challenges Lucrezia to play with her preferred videos so as to give her better recommendations. The girl is immediately engaged and gives the app info on her preferred videos by using tangibles of the toolkit. The app recommender system learns about her choices and uses them to ne tune its recommendation mechanism.
The main goals of the research are related to the scenarios. The most ambitious goal is to nurture children's knowledge of features of videos, recommended to them (e.g., colour, light, and motion). Thus, the system should be meant as an educational tool, which encourages children to re ect about what they watch. While the children may not be interested in understanding techniques, certain explanations with regard to the emotional aspects of the videos formulated by visual features, can be bene cial.
The other main goal is to investigate novel interaction approaches for
children's video recommendations, which can foster children's curiosity to explore
videos and their features. Being targeted primarily to younger children, the
system should be playful, e.g., gami ed, and mix tangible physical objects and
traditional screen-based interfaces for videos, like in [
The paper presents the ITTuB project, which explores how to design an interactive recommender system for tweens. It presents the background and related work on which ITTuB is based. In particular, the paper explains tweens' main characteristics and why they matter for the design of the ITTuB system. Then the paper advances ideas for the design of the system: they aim to combine a novel recommendation mechanism, which exploits sytlistic features of videos, with novel interaction modalities, which mix the virtual worlds of videos and the physical world of tangible objects for tweens.