Exploring a Modelling Method with Semantic Link Network and Resource Space Model

  • Muhammad Adnan Rafi

    Student thesis: Doctoral ThesisDoctor of Philosophy

    Abstract

    To model the complex reality, it is necessary to develop a powerful semantic model. A rational approach is to integrate a relational view and a multi-dimensional view of reality. The Semantic Link Network (SLN) is a semantic model based on a relational view and the Resource Space Model (RSM) is a multi-dimensional view for managing, sharing and specifying versatile resources with a universal resource observation.

    The motivation of this research consists of four aspects: (1) verify the roles of Semantic Link Network and the Resource Space Model in effectively managing various types of resources, (2) demonstrate the advantages of the Resource Space Model and Semantic Link Network, (3) uncover the rules through applications, and (4) generalize a methodology for modelling complex reality and managing various resources.

    The main contribution of this work consists of the following aspects:

    1. A new text summarization method is proposed by segmenting a document into clauses based on semantic discourse relations and ranking and extracting the informative clauses according to their relations and roles. The Resource Space Model benefits from using semantic link network, ranking techniques and language characteristics. Compared with other summarization approaches, the proposed approach based on semantic relations achieves a higher recall score. Three implications are obtained from this research.

    2. An SLN-based model for recommending research collaboration is proposed by extracting a semantic link network of different types of semantic nodes and different types of semantic links from scientific publications. Experiments on three data sets of scientific publications show that the model achieves a good performance in predicting future collaborators. This research further unveils that different semantic links play different roles in representing texts.

    3. A multi-dimensional method for managing software engineering processes is developed. Software engineering processes are mapped into multiple dimensions for supporting analysis, development and maintenance of software systems. It can be used to uniformly classify and manage software methods and models through multiple dimensions so that software systems can be developed with appropriate methods. Interfaces for visualizing Resource Space Model are developed to support the proposed method by keeping the consistency among interface, the structure of model and faceted navigation.
    Date of AwardNov 2022
    Original languageEnglish
    SupervisorHai Zhuge (Supervisor) & Xiaorui Jiang (Supervisor)

    Keywords

    • Resources Space Model
    • Semantic Link Network
    • Resource Management
    • Classification Dimensions
    • Software Engineering Models
    • Probabilistic Graphical Model
    • Discourse Structure
    • Text Summarization

    Cite this

    '