Tabulator Redux: Writing Into the Semantic Web

Figure 1. The Tabulator. The first frame shows the Tabulator with an RDF source, the Open Linked Data Project open. The second frame shows information within that source expanded, the third frame shows another source within that source expanded, and finally, the last frame shows that the label of that source has been edited from "Music and artist data interlinked" to "Music and artist data linked on the Semantic Web" ABSTRACT A first category of Semantic Web browsers were designed to present a given dataset (an RDF graph) for perusal, in various forms. These include mSpace, Exhibit, and to a certain extent Haystack. A second category tackled mechanisms and display issues around linked data gathered on the fly. These include Tabulator, Oink, Disco, Open Link Software's Data Browser, and Object Browser. The challenge of once that data is gathered, how might it be edited, extended and annotated has so far been left largely unaddressed. This is not surprising: there are a number of steep challenges for determining how to support editing information in the open web of linked data. These include the representation of both the web of documents and the web of things, and the relationships between them; ensuring the user is aware of and has control over the social context such as licensing and privacy of data being entered, and, on a web in which anyone can say anything about anything, helping the user intuitively select the things which they actually wish to see in a given situation. There is also the view update problem: the difficulty of reflecting user edits back through functions used to map web data to a screen presentation. In the latest version of the Tabulator project, described in this paper we have focused on providing the write side of the readable/writable web. Our approach has been to allow modification and addition of information naturally within the browsing interface, and to relay changes to the server triple by triple for least possible brittleness (there is no explicit 'save' operation). Challenges which remain include the propagation of changes by collaborators back to the interface to create a shared editing system. To support writing across (semantic) Web resources, our work has contributed several technologies, including a HTTP/SPARQL/Update-based protocol between an editor (or other system) and incrementally editable resources stored in an open source, world-writable 'data wiki'. This begins enabling the writable Semantic Web.