Supporting search and query in a hypermedia network is an important issue which should be addressed in the hypermedia systems. Structured hypermedia documents contain well-formed logical structures which can be exploited by structure queries. By using the powerful structure query facilities, users of the hypermedia systems can easily locate any interested portion of the hypermedia documents and obtain greater benefits than by just moving around the hypertext network following hyperlinks. In this paper, we provide approaches related to the storage and retrieval of structured hypermedia documents based on the database paradigm in order to support structure queries. We define a hypermedia document model to represent logical document structures. Based on the model, we present a structure query language which has a powerful addressing capability. We illustrate an efficient index structure for structured hypermedia documents using structure-bearing element identifiers. We also provide a vers...

this document also contains fact data of different sorts (e.g. dates, names of persons and institutions). So users may want to ask for this information, too. That is, besides documents, they want to search for other types of objects, too. There may be two different goals behind such a need:

A number of software toolkits exist for the development of text storage and retrieval systems (TSARS). This study compares three data models applicable to such toolkits and discusses the suitability of one of them as the basis of a toolkit unifying all three data models. The three data models are: (1) the text model, also known as the inverted file approach, (2) the hypertext model, and (3) the relational model. In the design of the relational model changeability was a key consideration, but more often it is sacrificed to save development resources or improve performance. As it is not uncommon to see successful TSARS exist for 15-20 years and be subject to manifold changes during their lifetime, it is the relational model which is considered for use in the unified toolkit. It seems as if the relational model can be enhanced to incorporate the text model and the hypertext model.