Analyzing movement of users in indoor settings introduces new challenges, which are not addressed by the existing research in the area. The challenges stem from the fact that we need to capture and analyze massive users’ trajectories that take place indoors, but also to provide and link them with enriched information, such as additional data on user’s activities, captured via a mobile device. We consider herein an application context in museums, where visitors are offered aid through devices (e.g. audio guides, mobile applications). The data collected can be used to allow museums to learn more about their visitors and their visiting trajectories and behavior patterns.
So the main challenges of the thesis are related to:
• modeling of users’ trajectoriesin indoor environments (geometric level), enriched
with information about the visitors and their activities (e.g. use of related multimedia,
consultation of texts or other descriptive works, etc.) (symbolic and semantic level),
• development of analytical methods to analyze these enriched trajectories that would also scale to handle the increased size of the data. These methods might extend the search techniques of spatiotemporal data for the detection of recurrent movement
patterns to indoor environments,
• interpretation of test results in a museum setting, in order to better understand visitor’s
• suggest in real time paths or points to visit based on the analysis of user’s behavior.
The work builds on a collaboration with the Louvre, which will provide the data, support and validation of the use cases’ analysis and interpretation.
The thesis takes place within the “Trajectories” (French: “Trajectoires”) project, funded by the Heritage Science Foundation (LabEx Patrima) carried by the ETIS laboratory at the University of Cergy-Pontoise (UCP), in collaboration with the DAVID laboratory at the University of the Versailles Saint-Quentin-en-Yvelines (UVSQ), the AGORA laboratory (UCP) and the Louvre Museum. The thesis will start in October 2016 will be co-directed by Dimitris Kotzinos (ETIS) and Karine Zeitouni (DAVID), in collaboration with all project partners.
We distinguish in the literature three types of moving objects trajectory models: geometric, formed by positions specified as coordinates; symbolic, formed by positions specified as reference to an object in the environment known in advance, and semantic, enriched with information describing their context / domain . We are interested in modeling users’ trajectories combining symbolic and semantic approaches in the museum context. A particular challenge is the design of several models with semantic aspects (location, type of work) and several levels of granularity in both space (work, room, wing) or meaning (painting, Italian painting, Italian painting XIII fifteenth century), allowing analysis at several levels.
Behavioral analysis methods based on trajectories  have been proposed for geometric or symbolic trajectories , but the integration of semantic dimensions related to the visitor, to his actions, the environment remains a challenge. To this is added the aspect of processing massive data sets, which means that we need to develop trajectories’ mining algorithms that scale well with the number of trajectories and the size and type of desired patterns.
In a closed environment (indoor), modeling and analysis are made more difficult by the absence of reference maps/points and the unavailability of space models. In the state of the art, only some mining approaches of frequent patterns  have been proposed. Moreover, the difficulty of having a precise location relative to points of interest and navigation constraints make the development of models in these environments more complex . The developed model should be able to represent the uncertainty of the indoor location and should be generalizable to different contexts related to the accuracy of the tracking device and the semantic model.
• Develop a model of users’ trajectories in the museum, at the symbolic and semantic level. Consider several semantic facets and multiple levels of granularity in this model. Consider the uncertainty of indoor location while modeling the geometries of the trajectories.
• Develop methods of analysis for this type of trajectories, focusing on the search for recurring patterns. Develop methods to enrich the position with several semantic dimensions, e.g. the characteristics of places and visitors, visitor actions, etc. Develop methods of mass analysis that scale well and their efficiency can be validated experimentally. Model the different types of the visitors’ profiles, define measures and develop interpretations for deviations for these types of profiles.
• Explore the extension of these models and methods to the analysis of the trajectories in real time, to dynamically suggest a suite of courses, business objectives, content consulting, etc., depending on the visitors’ behavior(s).
For more information click "Further official information" below.