[Paper Review] Towards Analytics Aware Ontology Based Access to Static and Streaming Data (Extended Version)

Evgeny Kharlamov, Yannis Kotidis+11 more|arXiv (Cornell University)|Jul 18, 2016

Semantic Web and Ontologies2 references3 citations

TL;DR

This paper proposes an analytics-aware extension to Ontology-Based Data Access (OBDA) that natively supports aggregate functions, correlation operations, and streaming data processing. By integrating analytical functions directly into the ontology and query language, and employing cost-aware query optimization, the approach enables efficient, declarative analytics over heterogeneous static and streaming data—demonstrated with real-world turbine diagnostics at Siemens, reducing query complexity from hundreds to a single ontological query.

ABSTRACT

Real-time analytics that requires integration and aggregation of heterogeneous and distributed streaming and static data is a typical task in many industrial scenarios such as diagnostics of turbines in Siemens. OBDA approach has a great potential to facilitate such tasks; however, it has a number of limitations in dealing with analytics that restrict its use in important industrial applications. Based on our experience with Siemens, we argue that in order to overcome those limitations OBDA should be extended and become analytics, source, and cost aware. In this work we propose such an extension. In particular, we propose an ontology, mapping, and query language for OBDA, where aggregate and other analytical functions are first class citizens. Moreover, we develop query optimisation techniques that allow to efficiently process analytical tasks over static and streaming data. We implement our approach in a system and evaluate our system with Siemens turbine data.

Motivation & Objective

• To address the limitations of traditional OBDA in handling analytical queries involving aggregation, correlation, and streaming data.
• To enable declarative, high-level ontological queries that express complex analytics without requiring low-level data source queries.
• To reduce the time and complexity of diagnostic workflows in industrial applications like turbine monitoring.
• To develop query optimization techniques that leverage source capabilities and minimize data transfer.
• To integrate ontology-based access with modern data stream management systems for real-time analytics.

Proposed method

• Extends the DL-Lite_A^agg ontology language to support aggregate functions (e.g., min, mean) and correlation operations as first-class citizens.
• Introduces a query language that allows analytical functions to be expressed directly in ontological queries, decoupling them from source-specific syntax.
• Designs a query rewriting strategy that pushes analytical operations to data sources where possible, leveraging precomputed aggregates.
• Employs cost-based optimization to select efficient execution plans, minimizing data transfer and computation.
• Integrates with a Data Stream Management System (DSMS) like ExaStream to support continuous, real-time processing of streaming sensor data.
• Adapts materialized window signature techniques from data warehousing for efficient summarization of streaming and historical data.

Experimental results

Research questions

• RQ1 How can OBDA be extended to natively support analytical functions such as aggregation and correlation in ontological queries?
• RQ2 What are the performance implications of pushing analytical operations to data sources in a heterogeneous, distributed environment?
• RQ3 Can a unified ontology-based interface reduce the number of required queries in industrial diagnostics workflows?
• RQ4 How can cost-aware query optimization improve efficiency when processing mixed static and streaming data?
• RQ5 What optimizations are effective for accelerating complex analytics over live and archived data streams?

Key findings

• The proposed approach reduced the number of required queries from hundreds to a single ontological query in Siemens turbine diagnostics.
• The system achieved significant performance improvements by pushing analytical operations to data sources, avoiding full data retrieval and enabling use of precomputed aggregates.
• The integration with ExaStream enabled efficient, continuous processing of streaming sensor data with materialized window summaries.
• Evaluation on real Siemens turbine data demonstrated the feasibility and efficiency of the approach in a large-scale industrial setting.
• The framework supports complex analytics such as Pearson correlation and threshold-based filtering directly in the ontology layer, improving expressiveness and maintainability.
• Preliminary results suggest that approximation algorithms and precomputed materialized structures can further accelerate query execution while maintaining result quality.