[Paper Review] Towards Connected Enterprises: The Business Network System

Daniel Ritter|arXiv (Cornell University)|Dec 29, 2013

Service-Oriented Architecture and Web Services12 references3 citations

TL;DR

This paper presents the Business Network System (BNS), a declarative, scalable framework that automatically discovers, reconstructs, and manages enterprise Business Networks (BNs) from heterogeneous, fragmented data across on-premise and cloud systems. By leveraging Datalog-based inference, a unified network model, and configurable domain-specific rules, BNS enables semi-automated reconstruction of interrelated business process, integration, and social perspectives, achieving conformance checking and real-time query support for enterprise network visibility.

ABSTRACT

The discovery, representation and reconstruction of Business Networks (BN) from Network Mining (NM) raw data is a difficult problem for enterprises. This is due to huge amounts of complex business processes within and across enterprise boundaries, heterogeneous technology stacks, and fragmented data. To remain competitive, visibility into the enterprise and partner networks on different, interrelated abstraction levels is desirable. We present a novel data discovery, mining and network inference system, called Business Network System (BNS), that reconstructs the BN--integration and business process networks--from raw data, hidden in the enterprises' landscapes. BNS provides a new, declarative foundation for gathering information, defining a network model, inferring the network and check its conformance to the real-world "as-is" network. The paper covers both the foundation and the key features of BNS, including its underlying technologies, its overall system architecture, and its most interesting capabilities.

Motivation & Objective

• To address the challenge of limited visibility into complex, inter-enterprise business networks due to fragmented, heterogeneous data across enterprise systems.
• To enable semi-automated discovery and reconstruction of 'as-is' business networks from raw network mining data across technical, business, and social perspectives.
• To provide a declarative, extensible system architecture that supports model-driven inference, conformance checking, and scalable querying without code changes.
• To unify disparate data sources—such as ESB configurations, SRM data, and integration artifacts—into a coherent, analyzable business network model.
• To support enterprise-wide, cross-tenant network management through a cloud-based, extensible system accessible to non-technical stakeholders.

Proposed method

• The system uses a two-model architecture: an inference model for data discovery and transformation, and a Business Network (BN) model for representing the final network structure.
• Raw data from enterprise landscapes is ingested as Datalog facts and processed through Datalog rules to infer network entities (e.g., systems, message flows) and relationships (e.g., ParticipantLink).
• A declarative domain-specific language (DSL) enables domain experts to define and configure inference rules without code changes, supporting flexible, extensible network modeling.
• Conformance checks are applied to ensure data integrity before loading into the knowledge base, using techniques from model checking and data quality assessment.
• The BN model is built on BPMN 2.0 collaboration-conversation semantics, with NetworkSpaces defining distinct perspectives (e.g., business process, integration) for multi-view analysis.
• A client API and generic model interpretation layer enable scalable remote query, traversal, and indexing of the reconstructed network for visualization and operational use.

Experimental results

Research questions

• RQ1 How can a system automatically reconstruct enterprise Business Networks from fragmented, heterogeneous data across on-premise and cloud systems?
• RQ2 What declarative modeling approach enables non-technical users to define and extend network inference logic without code modification?
• RQ3 How can a unified network model support multiple perspectives (e.g., business process, integration, social) while maintaining consistency and scalability?
• RQ4 What mechanisms ensure data conformance and integrity when integrating raw data from diverse enterprise systems into a single network model?
• RQ5 How can the system support continuous, scalable inference and real-time querying across enterprise and multi-tenant environments?

Key findings

• The Business Network System (BNS) successfully enables semi-automated discovery and reconstruction of enterprise Business Networks from heterogeneous data sources, including ESB configurations, SRM data, and integration artifacts.
• The use of Datalog-based inference with declarative rule configuration allows domain experts to define and adapt network inference logic without code changes, enhancing system extensibility.
• The system achieves cross-domain network reconstruction by mapping diverse domain-specific artifacts to a common inference model and a unified BN model, enabling multi-perspective analysis.
• Conformance checks are applied to raw data before ingestion, ensuring data quality and consistency with real-world 'as-is' network states, as validated by prior work on data quality and model conformance.
• The architecture supports scalable query and traversal via a client API and generic model interpretation, enabling real-time access to reconstructed networks for operational use.
• The system's modular design decouples discovery and inference from access, allowing independent evolution of the inference model and BN model, reducing maintenance overhead.