Building a Connected Web Observatory Architecture and Challenges Thanassis Tiropanis 1 , Xing Wang 1 , Ramine Tinati 1 , and Wendy Hall 1 University of Southampton Abstract. An increasing number of datasets and analytics resources have been made available on the Web. Often, those resources are hosted on portals that are related to dedicated topics (e.g. online social network analytics) or dedicated communities (e.g. Web Science researchers). A Web Observatory portal brings together communities to contribute or engage with datasets and analytic (or visualisation) applications. In ad- dition, it provides links to dataset and analytic resources that are poten- tially hosted in remote locations. This paper presents the architecture of a Web Observatory at the University of Southampton, which aims to foster engagement with analytics and applications and, at the same time, support linking to remote resources in other locations, potentially hosted in other Web Observatories. We report our experiences in build- ing the ”Connected Web Observatory” and discuss open challenges both in research and standardisation. 1 Building a Web Observatory Web Observatories aim to engage user communities with dataset and analytic resources via dedicated portals. They also aim to link to resources hosted in remote locations in other Web Observatories towards the vision of a global Web of Observatories[1]. Observatories in this sense include the following types of resources: – Portals; engagement portals that bring together communities engaging with analytic projects. They enable those communities to create or share datasets and analytic applications and engage in discourse. – Datasets; those provide quantitative or qualitative data, real-time data, mul- timedia content, open data, archives, e-Science resources. – Applications; analytic applications or visualisations that relate to a specific topic or to a community. – Tools; analytic resources that can support the development of datasets or of analytic applications such as visualisations. They can include harvesters when it comes to the development of datasets or data mining software when it comes to the development of analytics applications. We envisage a Web Observatory as a socio-technical artefact. On the tech- nological side it comprises at least a portal, which provides a list of datasets,