Entity–attribute–value Model - History of EAV Database Systems

History of EAV Database Systems

EAV, as a general-purpose means of knowledge representation, originated with the concept of "association lists" (attribute-value pairs). Commonly used today, these were first introduced in the language LISP. Attribute-value pairs are widely used for diverse applications, such as configuration files (using a simple syntax like attribute = value). An example of non-database use of EAV is in UIMA (Uniform Information Management Architecture), a standard now managed by the Apache Foundation and employed in areas such as Natural Language Processing (UIMA). Software that analyses text typically marks up ("annotates") a segment: the example provided in the UIMA tutorial is a program that performs Named Entity Recognition on a document, annotating the text segment "President Bush" with the annotation-attribute-value triple (Person, Full_Name, "George W. Bush"). Such annotations may be stored in a database table.

While EAV does not have a direct connection to AV-pairs, Stead and Hammond appear to be the first to have conceived of their use for persistent storage of arbitrarily complex data. The first medical record systems to employ EAV were the Regenstrief electronic medical record (the effort led by Clement MacDonald), William Stead and Ed Hammond's TMR (The Medical Record) system and the HELP Clinical Data Repository (CDR) created by Homer Warner's group at LDS Hospital, Salt Lake City, Utah. (The Regenstrief system actually used a Patient-Attribute-Timestamp-Value design: the use of the timestamp supported retrieval of values for a given patient/attribute in chronological order.) All these systems, developed in the 1970s, were released before commercial systems based on E.F. Codd's relational database model were available, though HELP was much later ported to a relational architecture and commercialized by the 3M corporation. (Note that while Codd's landmark paper was published in 1970, its heavily mathematical tone had the unfortunate effect of diminishing its accessibility among non-computer-science types and consequently delaying the model's acceptance in IT and software-vendor circles. The value of the subsequent contribution of Christopher J. Date, Codd's colleague at IBM, in translating these ideas into accessible language, accompanied by simple examples that illustrated their power, cannot be overestimated.)

A group at the Columbia-Presbyterian Medical Center were the first to use a relational database engine as the foundation of an EAV system.

The open-source TrialDB clinical study data management system of Nadkarni et al. was the first to use multiple EAV tables, one for each DBMS data type.

The EAV/CR framework, designed primarily by Luis Marenco and Prakash Nadkarni, overlaid the principles of object orientation onto EAV; it built on Tom Slezak's object table approach (described earlier in the "Entity" section). SenseLab, a publicly accessible neuroscience database, is built with the EAV/CR framework. Additionally, there are numerous commercial applications that use aspects of EAV internally including Oracle Designer (applied to ER modeling), Kalido (applied to data warehousing and master data management), and Lazysoft Sentences (applied to custom software development). An EAV system that does not sit on top of a tabular structure but instead directly on a BTree is InfinityDB, which eliminates the need for one table per value data type.