Mobility, Context-Awareness and Social Choice

Mastering Context and let Citizens have their Say…

In this axis, our intention in 2015 was to pursue the work we had initiated in 2005 on the collection and exploitation of context information in applications running on mobile devices equipped with several sensors (including GPS, camera, etc.), through two major notions: context-awareness and trajectories.

Context-awareness here refers to the capacity of a system or application to acquire (mainly through sensors) data it transforms into context information, seen as a selected set of data that the system or the application needs in order to adapt and improve its behaviour from the user’s point of view. Focusing on context-awareness in mobile applications, we have investigated two research directions.

First, in B. Aydin’s PhD thesis, we have addressed the challenge of retrieving and providing information about the surroundings to a mobile user using her/his Smartphone while walking, by coupling new technologies such as Augmented Reality (AR), location-based services and information provided by the LOD Cloud. This thesis results in an architecture called ARCAMA-3D.

Second, in G. Ben Nejma’s PhD thesis we have defined the notion of spontaneous community as a group of persons who are put in relation and interact using their smartphones in circumstantial, accidental, incidental or fortuitous situations. The challenge here was to design and build dynamically reconfigurable applications dedicated to spontaneous communities in a ubiquitous environment. The result is a dynamically reconfigurable architecture, called Taldea, for joining existing or creating new spontaneous communities.

Modelling trajectories, whether they refer to mobile objects or to individuals, is essential to better understand, ease, control or survey their moves or their behaviour. In Tales Paves Noguira’s PhD thesis a framework for automatic annotation of semantic trajectories has been proposed. Trajectories of mobile objects are semantically augmented with information about the environment coming from OpenStreetMap.

We have also paid attention to more abstract trajectories through the notion of life trajectory in D. Noel’s PhD thesis. This concept is broadly used by sociologists, urban planners, etc., who put in perspective the information on individuals to better understand their choices. We have designed an architecture, which embeds an ontology design pattern and a model of life events explanatory factors and makes it possible to immerse life trajectories in the Semantic Web and to analyse them using SPARQL queries. Behind these two research axes (context-awareness and trajectories), a question that naturally arose from our research is how to better orient a (mobile) user in her/his choice, which led to a third research direction.

A. Sales Fonteles’ PhD thesis is a first illustration. Here, a generic architecture for Spatial Crowdsourcing Platforms (like Mechanical Turk), called Genius-C, is proposed. It embeds an algorithm that solves what we have defined as and proved NP-hard: the Trajectory Recommendation Problem (TRP), which consists, for a given mobile user willing to spend time accomplishing tasks through such a platform, in finding the best sequence of tasks to be followed, while respecting both the spatiotemporal constraints associated with the tasks and the personal constraints expressed by the user.

A second illustration is provided by D. Nurbakova’s PhD. thesis which has focused on the recommendation of activity sequences during distributed events such as conventions, festivals, cruise trips, that have become very popular in recent years, attracting hundreds or thousands of participants. Here, the problem of recommendation of activities sequences has been defined, and the types of influence that have an impact on the estimation of the user’s interest in activities have been studied. The proposed approach, called Anastasia, solves this problem by: 1) estimating the user’s interest, 2) using sequence learning techniques, 3) using discrete optimisation to build an itinerary that takes into account spatiotemporal constraints. Finally, beyond this notion of better choices for mobile users, the natural generalization is how to make better decisions for a society of agents.

This research question naturally fits in the context of mobility and context-awareness in two ways. First, the aggregation of knowledge produced by individual users into a collective knowledge base raises the question of how to best aggregate this knowledge, and second, as stated above, the need for crowdsourced knowledge urges for efficient and fair procedures to allocate the individual tasks to the users. The domain that investigates these questions is Computational Social Choice. This domain is well represented in the team, as witnessed by the ongoing ANR Project Cocorico-Codec, the voting experiment Voter Autrement, and the works on fair division of indivisible goods.