Trigger-action programming is an emerging paradigm for personalizing the behaviour of Internet of Th ings applications and services. In this area some recent research ef orts have been dedicated to designing environments able to support people without programming experience in specifying the desired personalization rules. Lit le at ention has been paid to how to provide intelligent support in identifying relevant rules. In this position paper we discuss the possible approaches to making such personalization platforms able to exploit the data collected on the behaviour of the users as well as the surrounding devices and things.
•Human-centered computing~Human computer interaction (HCI) • Computer systems organization~External interfaces for robotics • Soft ware and its engineering~Soft ware notations and tools 1 INTRODUCTION
hT e Internet of Th ings (IoT) is the network of objects in our daily life (such as lights, refrigerators, car components, medical devices, dog collars, etc.) that can send or receive information with other devices on the network. Th ese objects include sensors and actuators of various kinds and can interact with each other, with human beings and with the environment to exchange data in order to react to realworld events, trigger actions and activate services. Th ey are increasingly used in many sectors: home, retail, industry,
Copyright © 2019 for the individual papers by the papers' authors. Copying permitted for private and academic purposes. This volume is published and copyrighted by its editors. agriculture, and so forth. Consequently, we use our applications more and more in dynamic contexts in terms of services, devices, objects and people where many events can occur.
Users' activities can vary quickly and some of them
cannot even be anticipated when designing interactive
applications, but are only discovered during their actual
use. It may be very difi cult for developers to predict all
possible context-dependent scenarios, because there may
be unforeseen (at design time) requirements that need to be
supported when the application is actually used. Indeed, we
are all dif erent from each other in terms of culture,
abilities, interests, goals. Th e increasing pervasiveness of
such technologies implies that they are more sensitive than
ever to such dif erences. Th us, end-user development (EUD)
approaches [
Trigger-action programming [
An example personalization platform is reported in [
hT e platform can be applied in several domains exploiting IoT technologies (e.g. smart retail, industry 4.0, home, health, etc.). For example, we are organising trials of this type of personalization platform for the PETAL project in the Ambient Assisted Living domain. In this context, the target users are elderly with mild cognitive impairments and their formal and informal caregivers. Th e older adults have a tablet for their interactions with applications supporting cognitive stimulation exercises and providing information useful for their daily life. Th ey also wear a smartwatch able to connect with Bluetooth and Wifi at the same time. Th is device is exploited to detect personal data (e.g. step counters and heart rate), and user’s indoor position (with the support of Bluetooth proximity beacons), these sensed data are sent to the context manager via Wifi connection. Other sensors include beacons with accelerometers to detect use of doors, windows, cupboards, objects, and sensors to detect motion, light, indoor and outdoor temperature, humidity. In addition, a number of lights that can be controlled in terms of colour, intensity, temperature, duration will be used as well for providing stimuli in order to help the older adults to bet er organise their daily activities. Th us, the lights can set scenes with activation and relaxation ef ects depending on the period of the day and the state of the user.
The massive amount of the data generated by this type of personalization platform can be processed in a way to add intelligence to the platform. Machine Learning (ML) algorithms can be used to achieve this goal. In this case, ML algorithms should be exploited to build knowledge by processing the available data with the goal to allow the platform to take dynamic real time initiatives based on what has been learnt from the historical data.
In order to introduce in this type of personalization platforms the ability to exploit the data related to the behaviour of users, devices and objects with the goal to automatically identify relevant personalization rules, there are two possible types of relevant data (see Figure 1): Data concerning the user behaviour (such as movements, interactions with objects and applications, …), and contextual conditions (such as time, weather, light, …); Data concerning the personalization rules that have been created and executed beforehand. Behavioural and contextual data In this type of platform, a machine learning approach should be able to identify in which contextual situations it is preferred that some actions occur. Based on the historical data collected by the context manager it is possible to train a model, which is then used to identify some rules. This type of approach can be effective when the possible events and actions are limited. For example, an intelligent thermostat is able to detect the room temperature at which the heating system should be turned on based on the user’s previous choices. Thus, in this case the data are when the user turned on the thermostat and the temperature at that time. The consequent action in this case is to turn the thermostat on. A more general solution is rather difficult because it implies the ability to monitor in a continuous and reliable way many possible relevant contextual aspects and actions performed (corresponding to changes of the appliances status or commands sent to applications). Thus, for example, if the goal is to personalize the use of the lights, then we should have tools able to monitor various aspects (time, user position, weather, …) and actual lights uses in order to identify possible routines and preferences in associating them.
One further issue is how to exploit the rules identified
through machine learning. Indeed, previous studies [
Rules‐related data
In this case it can be possible to design a personalization
rule recommendation system based on rules previously
provided according to two possible approaches:
collaborative or content-based filtering. In collaborative
filtering the basic idea is that if user A has specified a rule,
which is also indicated by user B, then there is a good
probability that other rules specified by user B can still be
relevant for user A. Thus, it is sufficient to monitor the rules
indicated by various users and then apply this approach to
identify what rules to suggest. In the case of content-based
filtering it is possible to consider various types of
recommendations. Some can be obtained through
generalization of the content of some part of the existing
rules. Thus, for example if there is a rule that says that
when the user enters the bedroom then the lights should be
on, one possible suggestion obtained through
generalization can be that when the user enters any room,
then the lights should be on. Another type of
recommendation can be obtained by trigger refinement,
which means narrowing the situations when the trigger
should be fired by adding conditions that make the rules
more suitable to meet more precisely specified needs.
Other recommendations can be based on the actual users’
behaviour and their preferences. This means exploiting a
specific case of context-aware recommendation systems
[
Trigger-action programming captures various aspects that characterize the emerging need to support personalization of IoT environments. EUD approaches can empower even people without programming experiences to customize such applications. Although intelligent support can be useful in the perspective to augment these EUD approaches based on the analysis of the data concerning the actual user behaviour, it needs to be carefully designed in order to obtain meaningful results.
In this position paper we provide some indications about what possible approaches to data collection and analysis can help us to obtain mixed-initiative, usable platforms for personalization of daily IoT environments in such a way as to enhance the user experience.