Trajectory ontology inference considering domain and temporal dimensions—Application to marine mammals

Highlights

• Modeling approach: from raw data to semantic trajectory modeling based on an ontological approach.

• Trajectory ontology inference: how domain and temporal rules are modeled and computed.

• Application domain trajectory model: how the approach can be applied to a specific domain.

• Design and implementation with the focus on inference system.

• Experiments and results.

Abstract

Capture devices rise large scale trajectory data from moving objects. These devices use different technologies like global navigation satellite system (GNSS), wireless communication, radio-frequency identification (RFID), and other sensors. Huge trajectory data are available today. In this paper, we use an ontological data modeling approach to build a trajectory ontology from such large data. To accomplish reasoning over trajectories, the ontology must consider mobile object, domain and other knowledge. In our approach, we suggest expressing this knowledge in the form of rules. To annotate data with these rules, we need an inference mechanism over trajectory ontology. Experiments over our model using domain and temporal rules address an inference computation complexity. This complexity has two important factors: time computations and space storage. In order to reduce the inference complexity, we proposed optimizations, such as domain constraints and temporal neighbor refinements. In this paper, we define a refinement specifically for the application domain. Then, we evaluate our contribution over real trajectory data. Finally, the results show the positive impact of the last refinement on reducing the complexity of the inference mechanism. This refinement reduces half of the time computation and then allows considering larger data sets.

Introduction

Advances in information and communication technologies have encouraged collecting spatial, temporal and spatio-temporal data of moving objects  [1]. The raw data captured, commonly called trajectories, traces moving objects from a departure point to a destination point as sequences of data (sample points captured, time of the capture). Raw trajectories do not contain goals of traveling nor activities accomplished by the moving object. Large data sets need to be analyzed and modeled to tackle user’s requirements. To answer user’s queries we also need to take into account the domain knowledge.

This paper deals with marine mammals tracking applications, namely seal trajectories. Trajectory data are captured by sensors included in a tag glued to the fur of the animal behind the head. The captured trajectories consist of spatial, temporal and spatio-temporal data. Trajectories data can also contain some meta-data. These data sets are organized into sequences. Every sequence, mapped to a temporal interval, characterizes a defined state of the animal. In our application, we consider three main states of a seal: hauling out, diving and cruising. Every state is related to a seal’s activity. For example, a foraging activity occurs during the state diving.

Our goal is to enrich trajectory data with semantics to extract more knowledge. In our previous work  [2], we tackled trajectory data connected to other temporal and spatial sources of information. We directly computed the inference over these data. The experimental results addressed time computation and space storage problems of the ontology inference. Then, we proposed some solutions to reduce the inference complexity by defining time restrictions in  [3] and inference passes refinements in  [4]. These later studies focus mainly on the term of time computation.

In the present paper, we continue studying the ontological inference complexity, specially in terms of inference space storage complexity. We propose two-tier inference filters on trajectory data. In other words, two distinct operations are performed to enhance the inference: primary and secondary filter operations. The primary filter is applied to the captured data with the consideration of domain constraints. The primary filter allows fast selection of the analyzed data to pass along to the secondary filter. The latter computes the inference over the data output of the primary filter.

This paper is organized as follows. Section  2 summarizes recent work related to trajectory data modeling using ontology approach and some introduced solutions to tackle the problem of the inference complexity using data filtering. Section  3 illustrates an overview of the used domain data model. This trajectory ontology contains temporal concepts mapped to W3C OWL-Time ontology  [5] in Section  4. Sections  5 Trajectory ontology inference, 6 Trajectory ontology inference using domain rules details the trajectory ontology inference and the integrated knowledge. In Section  7, we implement the trajectory ontology, the domain ontology rules and the temporal rules. Section  8 addresses the complexity of the ontology inference over the domain and temporal rules. Section  9 introduces an application domain inference refinement. Section  10 evaluates the ontology inference over the proposed refinement. Finally, Section  11 concludes this paper and presents some prospects.