When deployed in an enterprise, semantic technology can of simplify and rationalize disparate systems and databases. Semantic technology can sit ‘on top of’ data repositories in the enterprise to form a virtual directory and query system; it can be used as either the runtime system for business rules or simply model and prove the consistency of business rules that are run in traditional rules systems (e.g., iLog and TIBCO); and it can be used to model and implement interactions between information workers and the cases with which they interact.
To address the challenges of integrating semantics into the enterprise Thematix architects solutions that derive ultimately from a business and information architecture, and which are realized as running code built by Thematix contractors or the client’s own resources. The process is one of ever-greater specificity and definition, as discussed below.
Business and Information Architecture
The business architecture describes the business capabilities associated with core business capabilities, the critical processes carried out by the capabilities, and the information associated with these processes. The business architecture is a structured electronic document that can be analyzed with the aid of software, and that can be represented diagrammatically for understanding by decision makers and designers.
Many eChannel business processes such as deal development and offer selection will
involve the analysis and manipulation of complex information about deals, inventory and client preferences. Computer support for these processes follows a pattern that is often called “case management”. Case management is characterized by a need to manage and view complex information. Each case will be different from other cases, and often a path to the resolution of the case is unpredictable in advance. Standard process automation is rarely helpful for these processes, but process routing and automated decision making based on case information and business rules typically yield excellent results.
The Business Architecture is analyzed to identify governance responsibilities and rules for the management of eChannel information and applications across multiple geographies, business units, licensees and business partners. The analysis examines the patterns of creation, use and modification of information as it is passed between business capabilities in the course of executing primary business processes. In addition, this analysis produces a business level description of information used by activities within the capability. The description resembles an SOA service interface definition and is often called a Business SOA. It is valuable because it drives the IT SOA, insuring good alignment between the business activities and the IT assets that support them. It allows IT to track business changes via changes in Business SOA, replacing a traditional “one-shot” requirements gathering process.
The Information Architecture models the information at a business level that is created by and used in business processes, such as deal development, marketing, deal deployment, and deal evaluation. The Information Architecture models significant business entities (deal, customer, contract, location…), their important subcomponents and the primary relationships between these entities. It provides the information vocabulary that must be common between business and IT. It supplies guidance for data design, but does not address data schemas or data representations (binary, XML, relational).
Thematix models the Information Architecture with semantic technology (ontologies and rules). Semantic technology allows the expression of more of the meaning in the Information Architecture than alternative models such as Entity-Relationship (ER, EER). For example, semantic technology can assert:
- that a business entity is a subclass of another business entity (e.g. that a corporate contract offer is a kind of rental offer),
- that a business entity is an (essential) component of another business entity (e.g. that the rate is an essential component of a deal, but the discount is an optional component), and
- that a relationship between two entities is a sub-relationship of another relationship (e.g. the auxiliary drivers of a vehicle are a subset of the drivers of the vehicle),
- the business consequences of the different states of an entity (such as a reservation and a contract).
The ontologies and rules developed for the Information Architecture will form the basis for a set of operational ontologies to be developed for support of semantic-aided search, search engine relevance, semantic-aided offer selection and semantic-aided information governance.
The system architecture will be guided by major patterns that have been identified in the Business and Information Architecture. Key patterns are described in the following sections.
Knowledge work/case management
Some business processes are characterized by complex decision making and possible processing paths. High levels of skill and information are needed to carry out these processes that are often called “knowledge work”. A good many of the important eChannel processes would be classified as knowledge work. IT support for these kinds of processes consists of sophisticated information management and decision automation, instead of process automation. Two kinds of information must be managed: information about the case, and information about the processing of the case. The information structures are complex and highly variable, and work best with schema-less information management systems such as triple stores and XML stores.
Semantic Information Model – Ontology
A semantic information model will perform two functions: 1) supply a schema for a set of data, represented as triples with a lot of structural variability, and 2) provide a definition of the items and relationships in the data set for people using the data and for query engines processing the data. At the operational level, the semantic information model (an ontology) will provide terms and patterns for complex data that simplify access to the data. At the architecture level, the semantic information model defines the information needed to carry out decisions and processes that are characteristics of case management.
Master Data Management
One of the essential tasks in automating knowledge work is to provide a unified access to information that resides in different places and exists in many different forms. Example locations include word processing files, images kept in a file system, web page definitions held in a content management system, and tabular data kept in a RDBMS. In a knowledge work process like creation and deployment of a deal, cost and time to deployment are substantially reduced when all of the knowledge workers have shared access to information about the deal and all of the collateral associated with the deal. Many studies of knowledge work processes like these have shown that a large fraction of skilled worker time is spent simply looking for the information needed to make a decision or create a collateral item.
Master data management is an architecture pattern used to combine information held in different locations and with different structures. It has been matured in many initiatives to integrate information in legacy and relational databases. The master data management pattern can be extended to manage any information that can be given a URL by using triple stores and ontologies to provide the schema information needed to find and transform the information. In this way, deal structures can be tied to web page contents, FAQs and search indices that are collateral items associated with the deal.
The system architecture is typically a collaborative work involving client, Thematix and the other partners expected to be involved in a given project. This system architecture should define a new platform that addresses the present set of issues and provides flexibility and agility for the future. Service-oriented Architecture is the design and analysis discipline that can provide these features and should be a key part of the System Architecture.