In many scenarios, information required for a certain task is spread among a variety of systems and web services. Examples are headhunters and HR officers who spend much time gathering information about the skills and qualification of a candidate for a job. Similarly, criminal investigators have a hard time collecting all information about a suspected person or organisation, or about an illegal product offer, and identifying the connections between these (e.g. who is trading what).
A lot of the required information is available on the Web, e.g., in social networks, as well as in private databases, but it is scattered over multiple different sites and systems. The sources of this information change frequently, and because of legal restrictions with regard to data protection and privacy, it is not allowed and desirable to store results of such research unconditionally. This requires on-demand search and information integration. Still, for each search result its provenance has to be transparent, i.e. at what source it has been found. Different sources provide their information in different data formats and schemas and have different accessibility restrictions; this heterogeneity of data needs to be accommodated.
Despite the complexity of the data, the search user interface has to be learnable for its users, who are experts in HR or criminal investigation but should not have to be expert data engineers.