This paper proposes an extensible framework for capturing more data semantics in semistructured data models. The framework is extensible so that it can incorporate the latest advances in diverse domains, from web security and e-commerce to transaction-time databases. The additional semantics for each domain are captured in enriched labels. The new labels are sets of descriptive properties. The properties used as examples in this paper include transaction time, price, security, quality, and valid time. But the properties do not have to be the same for every database or even for every label within a database since this framework permits missing properties. Support for required properties, to model properties such as security, is also built into the framework.
Several new operations are needed to manipulate labels with properties. Match chooses a set of paths from the semistructure that match a user-given path regular expression. Collapse combines the properties in labels along a path to create a new label for the entire path. Slice slices a portion from each label on a path. Finally, Coalesce coalesces a property from a set of edges. These operations are built into the AUCQL query language, an implemented, Lorel-like query language, which is briefly described in this paper.
This work may be extended in a number of directions. Labels can be further extended to include a set of labels. This does not greatly increase the modeling power since multiple descriptions of the same relationship can be split into individual labels on a multitude of edges. However, it is essential to storing coalesced labels, which may be of some convenience to the user.
We also need to research translating meta-data in XML, such as RDF  or P3P , to a set of properties. The translation should be relatively straightforward since there is a clear mapping between paths in an XML data-set and properties: each path maps to a property, the labels along the path collapse to the property's name, while the terminal value of the path is the property's value.
Finally, and perhaps most importantly, the impact of our framework on path indexes must be addressed. We expect that a spatial or (bi)-temporal index can be generalized to index paths through properties in labels, and we plan to investigate this issue in the future.