Gellish database

A Gellish database is a semantic database that uses the formal language Gellish as their data model and for their content.

Universal data structure

Gellish databases are semantic databases that all have the same universally applicable data structure, the Gellish Expression Format. That data structure is suitable to contain any fact that is expressed in a Gellish language variant, such as Gellish English or Gellish Dutch, which are formalized versions of natural English and natural Dutch. This means that the structure of a Gellish database does not need to be extended when the scope of the database increases. The Gellish Expression Format is based on an extended version of the object-relationship-object (ORO) principles, that are also applied in RDF and OWL. Those principles state that every atomic fact is expressed as one or more binary relations. Higher order relations are converted into collections of binary relations. A Gellish database consists of one or more Naming Tables and one or more Fact Tables. A Naming Table and a Fact Table can also be combined in a Message Table, which is intended especially for data exchange between systems. Each row in a Fact Table contains one main fact and a large number of auxiliary facts. The auxiliary facts provide additional information about the main facts. For example, their status, author, creation date, validity context, language, etc. A Naming Table contains the terms, names, codes, abbreviations and synonyms by which the concepts and individual objects are named, possibly in multiple languages. They also contain unique identifiers for the named concept and things.

The Gellish Expression Format was originally defined as an implementation method for implementations of the generic data models ISO 10303-221 and ISO 15926-2, both of which International Standards are based on the same ORO principles. A Gellish database requires that each fact is expressed by one or more relations and that each relation is explicitly classified by a relation type (= fact type). The Gellish language requires that the relation type shall be one of the Gellish standard relation types. It also requires that each individual thing is explicitly classified by a kind of thing, which kind shall be selected from a formal dictionary that defines the concepts and terminology (vocabulary) of the Gellish variant, such as ISO 15926-4 or from the Gellish English Dictionary-Taxonomy with possible proprietary extensions. In conventional databases the relation types and the classification of individual things are usually implied by the database structure, which makes them limited and not extensible. Standard fact types (relation types) in a Gellish Database can be chosen from one of the above ISO standards or from the Gellish English Dictionary-Taxonomy.

Tables

Each Gellish Database consists of one or more tables in Gellish Expression Format or equivalent formats, such as extended RDF formats. Each of those tables uses basically the same format and is application system independent. This is different from conventional databases in which all database tables are different and that usually have proprietary data structures. Each of the tables in a Gellish Database shall contain at least columns for the obligatory expression components, such as a left hand object, a kinds of relation and a right hand object. Apart from that core, the tables may also have one or more columns for the optional contextual facts that provide the contexts that are required for the correct interpretation of the expressions. Preferred collections of contextual facts are defined in standard subsets. The expression components and contextual facts are defined in specification document 'Gellish Syntax and Contextual Facts - Definition of Universal Semantic Databases and Data Exchange Messages'.

The content of Gellish Database tables can be created conform ISO 10303-221 or ISO 15926 or can be compliant with the dictionary-taxonomy-ontology that defines the formal Gellish language and the vocabulary of its variants such as Formalized English, Formalized Dutch or any other formalized natural language. The standardized tables, in combination with the formal Gellish language, enables to combine an arbitrary number of Gellish tables into one Database. Furthermore, such a database might be centralized, but can also be a distributed database. This also enables to combine the results of a query to various independent data stores, which then act as a distributed database.

A Gellish Database may be implemented in various formats. It can be in the form of an SQL database, or in RDF/XML, or even in XLS (a format of Excel spreadsheet tables).

Limitations of conventional databases

Conventional databases typically consist of many tables, each of which is composed of a number of columns. The definition of those tables and columns determine the storage capabilities of the database, whereas the relations between the columns define the kinds of facts that can be stored in such a database. Those columns and relations determine the database structure that defines the expression capabilities of the database. Similar rules apply for the structure of data exchange files and thus for the information that is exchanged in electronic data files. This conventional database technology has some major constraints:

When data was not covered during the database design and thus is not included in the data model, then such data cannot be stored in the database nor exchanged via such a data file structure. Different databases have different data structures, which causes that data in one database cannot be integrated with data from other databases nor exchanged between databases without dedicated data conversion. A database modification or extension requires redesign of the database structure, modification of software and data conversion, which makes it a relatively complicated and costly exercise.

Another characteristic of conventional databases is that there are hardy international standards available or used for the content of the databases, being the data that is entered by its users. This typically means that local conventions are applied to limit the diversity of data that may be entered in those databases. As local conventions usually differ from other local conventions this has as disadvantage that data that are entered in one database cannot be compared or integrated with data in other databases, even if those database structures are the same and even if the application domain of the databases is the same. For example, within a company there may be various implementations of the same system in various sites for the storage of data about equipment, whereas for example the performance data about the same type of equipment still cannot be compared with the performance data in another location, because the equipment types have different names and the properties are also different.

Characteristics

A Gellish database does not have the semantic limitations that conventional databases have, because of the flexible and open Gellish language and because of its standard universal data structure (syntax), which is simple, computer and human interpretable. A Gellish database consists of one or more database tables, each of which has the same table structure (column definitions). Gellish databases are application-independent due to their tables being standardized and universally applicable. A standardized Gellish database table is universally applicable because it enables the application of the following two fundamental principles:

Explicit classification of individual things or explicit specialization of classes, with an unlimited number of classes in a dictionary. The Gellish database table enables to store any kind of object; because any individual object can be introduced by specification of an explicit classification relation between the object and a class, whereas classes (kinds of objects or concepts) can be selected from the very large number of classes that are already defined in the Gellish English Dictionary and if the proper class is not available it can be added by specification of a subtype-supertype relation with a direct supertype of the new class. This is fundamentally different from conventional databases that predefine the object types (classes) about which information can be stored by defining a limited number of entity types and attribute types in a fixed data model.

Explicit classification of relations (facts), by an extensible unlimited number of standardized relation types. The Gellish database table enables to store any kind of fact about any kind of object, because any fact is expressed by a relation, whereas those relations are explicitly classified by relation types that can be selected from the standardized relation types that are defined in the Gellish Dictionary or by relation types that are added to the dictionary as proprietary extensions. This is fundamentally different from conventional databases that predefine a fixed and limited number of relation types between the columns in the database tables (whereas unfortunately those relation types are usually defined only in an implicit way). As a consequence, a Gellish database does not need to be modified or extended when the scope of an application changes and facts from different Gellish databases can be merged and integrated whenever required without a need for a conversion exercise. Furthermore the content of a Gellish Database uses a common Gellish Dictionary for all its data, including for example, equipment types, property types, document types, activity types, etc.

Gellish Expressions in a Gellish Database

A Gellish Database is contains one or more expressions that are compliant with the definition of one of the Gellish language variants, such as Formalized English, and use the concepts and vocabulary that is provided in its dictionary or its proprietary extension. The core of a Gellish Database or Message table consists of three columns, just as is the case in RDF/Notation 3. Each row with those three columns in such a table expresses a main (binary) fact. For example, the fact that the Eiffel tower is located in Paris can be expressed as follows:

Left hand object	Relation type	Right hand object
The Eiffel tower	is located in	Paris
The Eiffel tower	is classified as a	tower
Paris	is classified as a	city

The left hand objects and the right hand objects may either be selected from the Gellish English dictionary or may be new proprietary objects that are introduced by defining them on separate lines. If such a new object is an individual thing, then it shall be defined by a classification relation with a class, as is done in the above table and if the new object is a class, then it shall be defined on a separate line by a specialization relation with their direct supertype. The relation types (such as 'is located in' and 'is classified as a') shall be selected from the Gellish English dictionary, otherwise the expression cannot be called standard Gellish, but becomes a proprietary extension of Gellish English.

Multi-language support

Furthermore, a Gellish database structure supports the simultaneous use of multiple languages. This is enabled because a Gellish database table contains a separate column for the language in which a fact is expressed (see the example table below). Thus a Gellish database supports the use of various natural language specific versions of Gellish. In principle, there is a Gellish variant language for each natural language, depending on the availability of a translation of the Gellish concepts. For example, the Gellish English Dictionary defines Gellish English, and contains partial translations to Gellish Deutsch (German) and Gellish Nederlands (Dutch). International terminology (such as most units of measure and mathematical concepts) is included as International Gellish.

Unique identifiers, homonyms, synonyms and automatic translation

A Gellish database uses a unique identifier for each thing, irrespective whether it is a user object, a concept from the Gellish dictionary, a fact or a relation type. The following Gellish database table is an extended version of the above example and includes the language in which the fact is expressed as well as the identifiers of the objects.

Language	UID of left hand object	Name of left hand object	UID of fact	UID of relation type	Name of relation type	UID of right hand object	Name of right hand object
English	1	The Eiffel tower	101	5138	is located in	2700887	Paris
English	1	The Eiffel tower	102	1225	is classified as a	40903	tower
Dutch	1	De Eiffel toren	103	4691	is a translation of	1	The Eiffel tower

The unique identifiers enable the use of synonyms and homonyms and enable that a computer can automatically translate a Gellish expression in a certain language into a Gellish expression in another language. This is because the meaning of a Gellish expression is captured as a relation between the unique identifiers, so that the meaning is language-independent. This adds automatic translation capabilities to Gellish expressions, because a Gellish message can be created e.g. in Gellish English whereas computer software can present it in another Gellish variant, such as Gellish Dutch if a dictionary or a translation is available, such as on the third line in the above table.

Contextual facts

A full Gellish database table has a number of additional columns that enable the expression of contextual facts or data about the main facts. For example, columns for:

• a textual definition of the left hand object
• the context in which a fact is valid
• a unit of measure with its UID
• the status of the fact (accepted, proposed, deleted, replaced, etc.)
• the originator of the fact
• the date of creation of the fact

External links

• http://www.gellish.net/
• Gellish Syntax and Contextual Facts - Definition of Universal Semantic Databases and Data Exchange Messages
• http://sourceforge.net/project/showfiles.php?group_id=28353
• Example of a road in Gellish
• 'Semantic Modeling in Formal English', by Andries van Renssen, ISBN 978-1-304-51359-5.