Debbie Wilson
Senior Geographic Information Architect
Ordnance Survey
United Kingdom

Abstract
Developing New Data Products and Content – Gaining Efficiencies and Consistency

Ordnance Survey is under increasing pressure to develop and deliver a wider range of derived contextual and analytic data products from our large scale content with improved consistency and currency. GenIE (Generalisation Information Engine) is an innovative new system that shall automatically generate derived and generalised content driven by changes our core large scale content. To ensure improved consistency of our small scale contextual and analytical content we have developed a single consolidated logical model that defines the features and properties that are applicable to all resolutions and applications. Profiling is used to specify which features and properties are applicable to a specific generalised content store. By using profiling, instead of re-defining the model for each content store it ensures that entities are defined consistently and we can re-use existing configurations reducing development timescales. To implement the different derived content profiles we have adopted a model driven implementation approach to automatically:
• Generate schema conformance tests (Cucumber) from the logical model to support Test Driven Development (TDD) using Enterprise Architect Documentation Templates • Generate profile schema for different content stores using ShapeChange • Transform (flatten) the schema to reflect implementation constraints using ShapeChange • Generate database schema using ShapeChange, Enterprise Architect documentation templates and ETL tools • Generate data specification documentation for end users using Enterprise Architect documentation templates
By adopting a model driven approach this has delivered the following benefits: 1) Significantly reduces the time taken to implement a content store for a new product (> 2 weeks) 2) Implementation schema can now be developed by a wider range of developers (software engineers, product development consultants) reducing reliance on specialist data modellers 3) Ensures consistency between the various implementation schema (GML, DDL, ArcGIS) as they are no longer manually created by different developers.