Integrating XML Data Sources using RDF/S Schemas: The ICS-FORTH Semantic Web Integration Middleware (SWIM)

Ioanna Koffina, Giorgos Serfiotis, Vassilis Christophides, Val Tannen, Alin Deutsch

TL;DR

This paper introduces SWIM, a middleware that bridges the gap between the Semantic Web's RDF/S schemas and the vast amount of data residing in Deep Web sources, particularly relational databases and XML. It proposes a novel Datalog-based framework for mapping these disparate data sources to a unified RDF/S view, enabling efficient query reformulation and optimization.

What Problem Does This Paper Address?

The Semantic Web (SW) promises to make information on the web more accessible and understandable by using standardized ontologies (like RDF/S). However, a significant challenge is integrating this with the enormous amount of data stored in "Deep Web" sources, which are often accessed through web forms and stored in relational databases or XML. These sources have their own formats and query languages (SQL, XPath, XQuery), creating a heterogeneity conflict. The paper addresses the need for a system that can seamlessly publish legacy data as RDF/S instances, allowing users to query this data using SW languages while the system handles the translation to the underlying source formats.

Methods

The core of SWIM's approach lies in its formal framework for mapping specification. Instead of directly mapping RDF/S to XML or RDB, the authors propose representing both data models as first-order logic predicates. This allows them to leverage existing techniques from relational query optimization.

Key methodological aspects include:

  • GLAV Approach: SWIM employs a hybrid Global-Local As View (GLAV) approach. This means mappings can define global schema elements in terms of local sources (GAV) and local sources in terms of the global schema (LAV), offering greater flexibility than pure GAV or LAV.
  • Linear Datalog: Mappings are specified using Linear Datalog (without recursion). This framework is chosen for its expressiveness and tractability. The head of the Datalog rules represents RDF/S views (using RVL), and the body uses XPath atoms to query XML sources.
  • Query Reformulation and Optimization: SWIM can reformulate RQL queries (used for RDF/S) into queries understandable by XML sources. This process involves "chasing" and "backchasing" steps, guided by the defined mappings and constraints. The goal is to produce minimized and optimized queries, eliminating redundancies.
  • Handling Heterogeneity: The system is designed to handle syntactic, structural, and semantic conflicts between XML and RDF/S schemas, such as differences in naming conventions or class hierarchies.

Datasets

Not discussed in the paper. The paper focuses on the methodology and framework for integrating arbitrary XML and RDB sources with RDF/S schemas.

Key Insights and Ideas

  • Unified Logic Framework: Representing both RDF/S and XML data models within a first-order logic framework is a powerful abstraction that simplifies the integration problem and allows reuse of established database techniques.
  • Datalog for Mappings: Using Linear Datalog for mapping specification provides a declarative and expressive way to define the relationships between different data models, bridging the gap between high-level SW queries and low-level source queries.
  • GLAV as a Sweet Spot: The hybrid GLAV approach offers a good balance between the ease of evolution of LAV and the query rewriting power of GAV.
  • Importance of Optimization: Reformulating queries is only half the battle; optimizing these reformulated queries to be minimal and efficient is crucial for practical performance, especially when dealing with large Deep Web sources.

Key Results

The paper presents a conceptual framework and architecture for SWIM. While specific quantitative results are not detailed in this extended abstract, the key qualitative findings are:

  • SWIM can establish mapping rules between XML and RDF/S.
  • It can reformulate RDF/S queries (like RQL) into XML queries (like XQuery).
  • The system aims to provide sound and complete query reformulation, ensuring all correct results are returned.
  • The optimization process aims to simplify queries and eliminate redundancies.

Limitations and Future Work

  • Complexity of Mappings: The paper acknowledges that while more expressive mappings can handle complex scenarios, they also make query reformulation harder. The authors aim to prove soundness and completeness of their reformulation algorithm to manage this complexity.
  • Performance: While optimization is a focus, the paper doesn't present empirical performance benchmarks. The efficiency of the chasing and backchasing algorithms on large-scale data would be a critical area for future work.
  • Support for RDB Sources: The abstract mentions SWIM's capability to handle RDB data, but the detailed examples and discussion primarily focus on XML. Further elaboration on RDB integration would be beneficial.
  • Schema Evolution: The paper touches upon the ease of evolution with LAV, but a detailed discussion on how SWIM handles schema changes in the underlying sources or the RDF/S schema would be valuable.

Summary updated

Abstract

Semantic Web (SW) technology aims to facilitate the integration of legacy data sources spread worldwide. Despite the plethora of SW languages e.g., RDF/S, OWL recently proposed for supporting large scale information interoperation, the vast majority of legacy sources still rely on relational databases RDB published on the Web or corporate intranets as virtual XML. In this paper, we advocate a Datalog framework for mediating high level queries to relational and or XML sources using community ontologies expressed in a SW language such as RDF/S. We describe the architecture and the reasoning services of our SW integration middleware, called SWIM, and we present the main design choices and techniques for supporting powerful mappings between different data models, as well as reformulation and optimization of queries expressed against mediation schemas and views.
BibTeX
@inproceedings{koffina2005integrating,
  author = {Ioanna Koffina and Giorgos Serfiotis and Vassilis Christophides and Val Tannen and Alin Deutsch},
  title = {{Integrating XML Data Sources using RDF/S Schemas: The ICS-FORTH Semantic Web Integration Middleware (SWIM)}},
  booktitle = {Semantic Interoperability and Integration},
  year = {2005},
  url = {https://openreview.net/forum?id=KuvAc163jK},
}
Figure