View article

The digital computer has greatly increased our capacity for storing and accessing information. The internet, and particularly the World Wide Web, have vastly increased the amount of information available to us. Information retrieval and hypermedia research has greatly reduced the time and effort required to fulfill a searcher's information need; however, problems still remain. To access multimedia information, text metadata must usually be assigned to the multimedia objects, requiring an (often prohibitively) large amount of time and effort. Alternatively, some systems use low-level features of the media to allow a searcher to find objects similar to a query object. Such features cannot always identify two media objects depicting or representing the same real-world concept; in the case of images, different camera angles and lighting conditions mean that semantically similar images are visually very different. Additionally, semantic relationships between objects may not be expressed. This thesis presents a way of addressing these problems by drawing on the field of semiology, in which a symbolic object has two parts: That which is being represented (in the plane of content), and the symbolic object doing the representing (in the plane of expression). A semantic layer is introduced holding real-world concepts, and connections to the multimedia objects that represent them. Knowledge about these concepts can be introduced by connecting them with semantic relationships. A prototype multimedia information system incorporating a semantic layer feature, the multimedia thesaurus, is introduced. The construction and use of a museum application is …