Using metadata to help the integration of several multi-sources set of updates

Today, spatial data are increasingly available on the web and users can update their datasets more easily. Different sets of updates result from diverse sources are furnished to the user, each containing updates acquired in different manners, with different quality and at different times. A special context where the data and updates could come from different sources is a military mission. Indeed, the actors are distributed between different sites and one particularity is that they can be either a producer or a user of the data. They have their own dataset and can update them in several ways but must regularly supply their evolutions to the others actors in order to guarantee the success of the mission. Therefore, each actor receives many heterogeneous sets of updates and must integrate them in their own dataset in accordance with their needs. In this context, the user receives several set of heterogeneous updates which can have different quality, which can contain errors due to the manner they were acquired and they have to integrate them in their personal dataset. Thus, all the evolutions are not necessarily interesting for the user, and conversely one set of updates may not cover all the user needs. These heterogeneous sets of updates could also be concurrent each others and be concurrent with the user dataset. In this context, how can a user efficiently update his spatial dataset with some evolutions which are not necessarily pertinent and probably concurrent? This is the essential question to answer if we want to improve the update of spatial data by different sets of evolutions which are coming from multiple sites. In this paper, we will study the main problem arising when we integrate concurrent and heterogeneous updates and we will propose a process which helps the user to integrate efficiency multi-source updates into his dataset. This process comprises several steps : Firstly, we classify the evolutions to remove the heterogeneity, secondly we take into account the user needs and exclude the non pertinent data, thirdly we check the concurrency control between all the updates, and finally we reconcile the data if a conflict was detected. This process uses metadata to choose the “best” evolution to be integrated in the dataset. The metadata used are structured in accordance with the ISO 19115 standard specifications. ha l-0 04 39 74 8, v er si on 1 8 D ec 2 00 9 Author manuscript, published in "International Conference of the Global Spatial Data Infrastructure (GSDI), Santiago, Chili, 06/11/2006-10/11/2006, Santiago : Chile (2006)"