Capturing Software Requirements Modeling from Ontology Tree

Yaqing Liu, Jinghuan Guo -, Yong LiuInternational Journal of u- and e- Service Science and Technology 2015 · DOI

TL;DR

This paper introduces an "Ontology Tree" approach to software requirements modeling. By decomposing large domain ontologies into a hierarchical tree structure, the authors aim to improve the clarity, development, and maintainability of software models, especially for complex business domains.

What Problem Does This Paper Address?

Software requirements modeling is a crucial but challenging phase in software engineering. For large and complex business domains, developers often lack sufficient knowledge, leading to incomplete or poorly structured domain models. While using domain ontologies has helped capture more comprehensive requirements, existing methods often result in tightly coupled models that are difficult to maintain. The paper addresses this by proposing a novel way to represent and utilize domain ontologies to create more manageable and maintainable software models.

Methods

The core of the paper's contribution lies in its two-step framework:

  1. Ontology Decomposition into an Ontology Tree: The authors propose decomposing a large, monolithic domain ontology into a hierarchical "Ontology Tree." This is achieved by first representing the ontology using formal concept analysis and concept lattices. They then employ a decomposition algorithm that leverages semantic similarity between attributes to cluster them, effectively breaking down the ontology into smaller, more manageable sub-ontologies. These sub-ontologies are then organized into a tree structure.
  2. Requirements Model Generation from Ontology Tree: Once the ontology is structured as an Ontology Tree, the requirements model is generated from this tree, rather than the original, large ontology. The mapping rules are straightforward:
  3. Classes in the ontology map to classes in the domain model.
  4. Instances map to entities.
  5. Properties map to properties.
  6. "Synonym of" relationships between classes map to inheritance between classes.

The paper also introduces algorithms for attribute clustering (Algorithm 1: ACSS) and formal context decomposition (Algorithm 2: FCD) to facilitate this process.

Datasets

The paper uses a "European Cities Context" as a running example to illustrate its concepts. This context is presented in tabular form (Table 1, Table 2, Table 3.1, 3.2, 3.3) and is used to demonstrate the decomposition of formal contexts and the generation of concept lattices. No specific large-scale datasets were used for training or evaluation in the traditional machine learning sense.

Key Insights and Ideas

  • Decomposition is Key to Manageability: The fundamental insight is that large, flat ontologies are problematic for software engineering. Breaking them down into a hierarchical "Ontology Tree" structure directly addresses the issue of complexity and coupling.
  • Ontology Tree vs. Flat Ontology: Representing the domain knowledge as a tree, rather than a single, large ontology, makes it easier to navigate, understand, and modify specific parts of the model. This is particularly beneficial for large domains.
  • Semantic Similarity for Decomposition: Using semantic similarity of attributes to cluster them is a clever way to group related concepts, which naturally leads to a more meaningful decomposition of the ontology.
  • Improved Maintainability: By having a structured ontology tree, changes to requirements can be more easily located within specific sub-ontologies, making the software development and maintenance process more efficient and less error-prone.

Key Results

The paper presents a case study of applying their approach to a restaurant management information system for a hotel.

  • Ontology Simplification: An initial hotel ontology with 78 classes and 128 dependence relations was simplified to 21 sub-ontologies and 110 dependence relations.
  • Improved Usability for Stakeholders: The top-level ontology, now with only 15 classes, is more accessible to top managers, while the more detailed bottom ontologies are suitable for hotel staff.
  • Enhanced Maintainability: The authors claim that their approach makes it easier to locate parts of the model that need changing and to evaluate the scope of influence for those changes.

Limitations and Future Work

  • Limited Empirical Validation: While a case study is presented, the paper could benefit from more extensive empirical validation across a wider range of domains and larger-scale projects to definitively prove the superiority of their approach.
  • Algorithm Complexity: The paper describes algorithms for decomposition and clustering, but a detailed analysis of their computational complexity and scalability for extremely large ontologies is not provided.
  • Future Work: The authors suggest that further research could explore more sophisticated methods for ontology decomposition and the generation of requirement models from the ontology tree. They also imply that the mapping rules could be further refined.

Summary updated

BibTeX
@article{liu2015capturing,
  author = {Yaqing Liu and Jinghuan Guo - and Yong Liu},
  title = {{Capturing Software Requirements Modeling from Ontology Tree}},
  journal = {International Journal of u- and e- Service Science and Technology},
  year = {2015},
  doi = {10.14257/ijunesst.2015.8.11.01},
}
Figure