Timeliness for dynamic source selection in situated public displays

Abstract

Dynamic sources, which make regularly updated data available for use by other applications, are increasingly a key enabling feature of the web. They are extensively used in all sorts of social media applications where they are re-combined in multiple ways to generate new aggregate services. Public situated displays are an emergent area where dynamic sources can also play a key role in providing situated and frequently updated content. However, the specificities of public displays raise the need for automated selection of the most relevant sources to present. This study addresses relevance from the perspective of timeliness. We propose a timeliness model that supports the most common types of dynamic source. To validate that model, we set an experiment with a public display exhibiting content from dynamic sources and receiving from users feedback on its timeliness. The results from this experiment suggest a reasonable match between our model and the users' perspectives on timeliness. The results also show that the model is able to make comparative calculations of timeliness for different types of dynamic source. These results enable us to conclude that timeliness functions may help to significantly increase the relevance of content automatically selected from dynamic sources.