The graph data model (GDM) is so called because its basic schema, consisting of record types and binary link types, constitutes a directed graph. GDM combines features and inherits advantages of many of the earlier models. As in network and hierarchical systems, navigation and efficiency are provided for. GDM also, like the relational model, offers flexibility, set-at-a-time queries, and the possibility of complex relationships; in fact, GDM’s Data Manipulation Language, allowing transitive closure and numerical quantification, goes beyond the expressive power of relationally complete systems. Finally, one can translate the high-level semantics found in semantic models, especially the popular entity-relationship model, into GDM.
The author began her work on this subject as a Ph.D. student at the University of Texas in 1979. Since then she has overseen two implementations of GDM in the DBMS laboratory of Ricoh’s Software Research Center, of which she is now general manager. In fact, one of the remarkable aspects of this book is the breadth of its development of GDM, ranging from an axiomatic exposition to a treatment of such practical ideas as access path strategies and a cost model for query evaluation.
Given this wide coverage, various groups will find this book of substantial interest. Theoreticians and members of the database research community will appreciate the formal exposition, while practitioners will be interested in issues of implementation.
Seven chapters and two appendices make up this monograph:
Introduction
Graph Data Model
Data Definition Language
Data Manipulation Language
Implementation of GDL [Graph Data Language]
A Comparison of Access Path Strategies
Conclusions
Appendix A is “A BNF Grammar of GDL,” and Appendix B gives the “Specification of Run-time System.” The author develops all of these topics carefully and formally, even the implementation issues of chapters 5 and 6 and appendix B. The metric for the cost of query evaluation algorithms presented in chapter 6 is Yao’s model [1]. The depth and breadth of Kunii’s treatment are appropriately thorough throughout, but I would have appreciated a detailed discussion of GDM as a data model against the background of general models described in the classic book by Tsichritzis and Lochovsky [2]. The index is good, the bibliography is excellent, and the foreword is by Gio Wiederhold.
