Traditional Approach
A typical Integration approach has been to map two interfaces. The interfaces can be system-level, system to datastore, and system to service (asynchronous - protocol agnostic).
Inference-based Mapping - Semantic Services
Semantic web services are web services which have metadata about the service definition. The service request and response fields are described in the metadata. The field information can be used to infer about the services. Inference-based mapping is an approach which can map two interfaces given the metadata about the services. The relationships across the fields can be described using rules, ontological concepts and terms. Service definitions or schemas which describe the interfaces can evolve. Automated Discovery Selection, Composition and web-based execution of services
Semantically Enabled SOA (SESA)
WSMO Web service Modelling ontology has ontology and Rule language for semantic web with support for WSMO. Semantic Execution environments have the capability in deploying and executing semantic web services.
WSML Top Down Process Design Bottom Up
- web service description and annotation.
- Semantic BPM Semantic Business Process Analysis
- Semantic Business Process Modelling
- Semantic Business Process Design
- Semantic Business Process Execution
- Code Generation based Mapping- Standard-based Integration Templates for Interface
Mapping Out-of-the-box connectors can be the templates which evolved from inference engines for mapping two standard interfaces. The templates are stored in a repository for standard interfaces with version information. In Memory Inference In scenarios of translation between two languages, the traditional method is to automate the translation word by word and workflow will be triggered for review. In the new approach, in memory changes during workflow are stored. In memory, inference will help speed translation and bring down the language translation's manual review.
Recommendations
In the data integrations, the fields which describe the customer transaction can be used for loyalty, offers and recommendations. In e-commerce (B2C, B2B, C2C), system interfaces can be integrated using a new approach of inferring the data and using them as recommendations. ETL Extraction of the data and transforming them for loading the data repository is the traditional approach. The new approach evolving is to store the transformations evolved from the inference engines for mapping two interfaces. The transformations are stored in the repository. Transformations can be clustering the data into groups, correlated data transformations, rules, logic and sorting/filtering. Data Relationships The Discovery of structured and unstructured data relationships need to be identified and stored in the repository. The data can be graphs, spatial, text, sensors and location.
Applications
- Intelligence
- Bio Informatics
- Health Care informatics
- Finance
- Media
Top comments (0)