The Model Observatory offers two services. First and foremost, it is an index of all known and validated repositories of Software Engineering models. The idea is to popularize the existing resources so that they are better utilized, and hopefully, more people are inspired to contribute, too. For this purpose, we collect as much meta-data as we can and store them locally so that an advanced search can be executed quickly and comprehensively. As a side effect, we hope to be able to standardize the inconsistent terminology to some degree.
Second, the Model Observatory contributes a model repository of its own, populated by the models we have collected over time. As far as we can tell, it contains more models than existing repositories, those models are larger individually than what is offered elsewhere, and the information provided is much richer than what is found in existing repositories: all of them come as an XMI-file, generated reports in PDF and HTML, and most of them come with a human-written report detailing the why and how of the models. Moreover, the models the Model Observatory provides do not come as isolated items, but with documented relationships, e.g., precedence or alternative.
The Model Observatory is currently in the build-up; a first version is to be released in October 2014, officially introducing its first increment (the FMI Model Index) at MODELS 2014.
In the first stage, the Model Observatory only collects links to and summary data about existing repositories. This stage is currently under review (September 2014), and will be released at the MODELS conference in Valence. Usage is free and does not require a registration.
In the second stage, meta-data from the models found online will be collected and fed into a model discovery service we will host at the Model Observatory. In this stage, we open up the Model Observatory for foreign contributions, i.e., reports of further repositories.
In the third stage, we will add our own models. Their metadata will be stored in the same way as that of other repositories, but the actuall models are stored physically by Zenodo. Zenodo will also provide DOIs.
In the fourth and last stage, a makeover of the whole web site will integrate the various contents seamlessly so that a search for a model or set of models with a particular profile will Furthermore, a search facility based on the Visual Model Query Language (VMQL) will be integrated to allow search-by-example. This will not just improve existing use cases, but enable new ones such as model reuse. Finally, a model procurement pipeline will be set up that harvests models from publicly available sources such as pictures and convert them into semantic models automatically by the Gourmand tool.