Advanced Data and Knowledge Management Systems

Información del proyecto

ADKMS

Identificador del acuerdo de subvención: 1498

Proyecto cerrado

Fecha de inicio 1 Diciembre 1984

Fecha de finalización 1 Diciembre 1989

Financiado con arreglo a

• European programme (EEC) for research and development in information technologies (ESPRIT), 1984-1988

Coste total

Sin datos

Aportación de la UE

Sin datos

Coordinado por

Siemens Nixdorf Informationssysteme AG
Germany

Objetivo

The objective of ADKMS was to develop a knowledge-based system with an inferential capability for the intelligent and efficient management of large databases, suitable for both naive and domain-expert users. Access to the system was to be provided by a Natural Language (NL) interface.
The task for Phase 1 was to develop the NL handlers, and to integrate a relational database management system with a knowledge-based system.
The tasks for Phase 2 were the improvement, integration and evaluation of the Phase 1 results in an industrial application domain.
The advanced data and knowledge management system (ADKMS) project covers knowledge representation, knowledge base development, data and knowledge base coupling, natural language analysis and generation, and intelligent access to databases. The main components of ADKMS are:
the knowledge representation system BACK, a hybrid reasoning system supporting complex representation of a domain terminology and database access via a powerful logic oriented interface language;
the knowledge base development environment BIT, providing a range of integrated graphics tools for interactive acquisition, browsing, navigating and validating large knowledge bases;
the natural language access system, consisting of 3 levels: tactical, for language parsing and generation (NUGGET); strategic, for generating structured query language (SQL) queries and the formal representation of the answers; and a data level, consisting of the database (using ORACLE) and the additional knowledge.

The objective was to develop a knowledge based system with an inferential capability for the intelligent and efficient management of large databases, suitable for both naive and domain expert users. Access to the system was provided by a natural language (NL) interface. A functional layered architecture was designed. Prototypes of natural language interfaces for German and Italian were constructed. A prototype of a hybrid knowledge representation and inferencing system (KRS) was mapped onto the database in a transparent fashion, through a Prolog structured query language (SQL) interface. A prototype of a database extension module was constructed, and coupled to a relational database management system (RDBMS). The whole project was subsequently evaluated and the definition of requirements according to a real application proposed. A common semantic representation was specified for Italian and German. A new version of KRS was provided, including a new query language to fulfil industrial demands and to interface DB4. The extended database query language was installed on a Nixdorf Targon 35. The NL handlers, the knowledge representation system and the extended RDBMS were state of the art products.
The major results of Phase 1 were:
-the design of a functional layered architecture for an ADKMS
-prototypes of natural language interfaces for German and Italian from Nixdorf and Olivetti
-a prototype of BACK, a "hybrid" knowledge representation and inferencing system (KRS) from the Technical University of Berlin, and a partial reimplementation from Nixdorf
-a mapping of the KRS onto the database in a transparent fashion, through a Prolog/SQL interface
-a prototype of a database extension module, and its coupling to an RDBMS
-evaluation of the system in field and laboratory trials.
Phase II started resulted in:
-the evaluation of the whole project and the definition of requirements according to a real application proposed by Datamont
-the specification of a common semantic representation for Italian and German
-a new version of BACK-System, including a new query language to fulfil industrial demands and interfacing DB4
-the porting of the extended database query language onto a Nixdorf Targon 35.
The NL handlers, the BACK knowledge representation system and the extended RDBMS were state-of-the-art products.
Exploitation
The industrial prototypes provided a foundation for intelligent database management systems with greater functionality and non-restricted natural language handlers, allowing sophisticated AI applications to very large data and knowledge bases. These systems should combine the advantages of expert systems and database management systems.
The results have been incorporated into the ESPRIT II project 5210, AIMS.

Ámbito científico (EuroSciVoc)

• ciencias naturales › informática y ciencias de la información › ingeniería del conocimiento
• ciencias naturales › informática y ciencias de la información › inteligencia artificial › sistemas expertos
• ciencias naturales › informática y ciencias de la información › base de datos › base de datos relacional

Programa(s)

• FP1-ESPRIT 1 - European programme (EEC) for research and development in information technologies (ESPRIT), 1984-1988

Tema(s)

Datos no disponibles

Convocatoria de propuestas

Datos no disponibles

Régimen de financiación

Datos no disponibles

Coordinador

Siemens Nixdorf Informationssysteme AG

• Aportación de la UE: Sin datos
• Dirección: Pontanusstraße 55 ; 33102 Paderborn ; Alemania
• Coste total: Sin datos

Participantes (3)

Bull SA

Francia

• Aportación de la UE: Sin datos
• Dirección: 68 route de Versailles ; 78430 Louveciennes
• Coste total: Sin datos

Ingegneria C. Olivetti and C. SpA

Italia

• Aportación de la UE: Sin datos
• Dirección: Via Palestro 30 ; 56100 Pisa
• Coste total: Sin datos

TECHNISCHE UNIVERSITÄT BERLIN

Alemania

• Aportación de la UE: Sin datos
• Dirección: FRANKLINSTRAßE ; 1000 BERLIN
• Coste total: Sin datos