Please use this identifier to cite or link to this item:
Title: Computational narrative simulation for organizational learning
Authors: Wang, Wai-ming
Degree: Ph.D.
Issue Date: 2009
Abstract: Organizational learning (OL) is one of the most important capabilities for the survival of an organization. Organizations need to have the dynamic capabilities to create, capture, harvest, share, replenish and apply their knowledge. Knowledge comes from experience and resides in the narratives of how people deal with real-life problems, so the acquisition of knowledge must be interwoven with the process of applying it. Knowledge workers need to have exposure to problems in order to achieve OL. However, it is difficult to acquire knowledge from narratives and difficult to share it efficiently and effectively. Traditional methods are shown to be ineffective in fulfilling the requirements of the important knowledge processes and in supporting the factors that facilitate OL. In the present study, a Computational Narrative Simulation (CNS) approach is presented and a Computational Narrative Simulation System (CNSS) has been designed and built to overcome the limitations in the existing methods. By incorporating the technologies of knowledge-based systems (KBS) and artificial intelligence (AI), four computational intelligent algorithms have been developed for supporting the automation of the narrative simulation. These algorithms are: Fuzzy-Associated Concept Mapping (FACM), a narrative prediction and construction algorithm, Hybrid Case-based Reasoning (HCBR), and Self-Associated Concept Mapping (SACM). The FACM was developed in order to collect unstructured narrative knowledge and convert it into structured knowledge represented in such a way that it can be processed easily by computers. It supports the simulation designer in managing a massive quantity of narrative data collected from the workers. The narrative prediction and construction algorithm is used for selecting relevant narratives for the construction of narrative simulations. Narratives that refer to situations that will probably occur in the near future are selected for constructing the beginning of the narrative of a narrative simulation. The HCBR algorithm has been developed for deducing the decision points, questions and answer choices in the narrative simulation. A SACM algorithm has been developed which serves as an inference engine for associating the decision points with the narrative segments in the narrative simulation. To evaluate the performance of these algorithms, a series of quantitative experiments has been carried out. The value and capability of these algorithms are realized by using various public databases and industrial data. The results are also benchmarked with well known algorithms and commercial software (e.g. ANNIE). It is interesting to note that the FACM increases the recall rate (about 40% to 50% improvement) and maintains a high precision rate (over 80%) when compared with the baseline algorithm. There is an over 70% accuracy of the narrative prediction and construction algorithm in the prediction of propositions when the time interval is set to 3 months or above. The accuracy is good, even though the accuracy of the method is calculated on a basis on strict restrictions and strict conditions. The accuracy of HCBR is higher than that of the well known algorithms known as the Case-based Reasoning (CBR) algorithm (about 12% to 22% improvement) and the Rule-based Reasoning (RBR) algorithm (about 3% to 47% improvement). For testing the SACM, two experiments were carried out. The first experiment shows that the accuracy of SACM is compatible with the accuracy of human beings. According to the correlation analysis, SACM has a high correlation (0.90) when tested on laymen of the tested domain when compared with CBR (0.69). The knowledge inference of SACM has the advantages of higher laymen learning capability, smaller data size and faster speed, than CBR. The results of the second experiment show that the SACM and the CBR have a similar average accuracy and the accuracy of SACM increases significantly when the number of learning cases increases. Clearly, the SACM is useful and capable of simulating human learning activities. It is an important base for the development of the CNSS. In order to evaluate the performance of the CNSS as a whole in real-life implementation, a prototype system has been built and trial implemented for case management in a social service organization (i.e. Baptist Oi Kwan Social Service (BOKSS)) organization. A survey was distributed to the case workers for measuring the performance of the narrative that was constructed by the CNSS. The results show that the participants generally agreed that the narrative was informative, realistic and authentic. The participants were able to relate personally to the narrative and they said that they had learnt something new from the narrative. They agreed that the length of narrative was appropriate, the narrative was easy to understand and the questions were easy to answer. On the whole, the participants commented that the constructed narrative was useful. A controlled experiment was also carried out for measuring the learning outcome created by the system. From the results, the average mark of the experimental group were 17% higher than that of the control group, which indicated that the system was able to improve their work. On the whole, a novel computational narrative simulation (CNS) method has been established, which addresses the deficiencies and limitations of traditional narrative simulation approaches for achieving organizational learning (OL). It integrates a number of technologies. Such integration of technologies is considered as novel. Four computational intelligent algorithms have been developed for the realization of the CNS. Fuzzy associated concept mapping (FACM) has been developed for automatically mining concept maps from natural language texts by the novel integration of natural language processing (NLP) with concept mapping. Narrative prediction and construction has been developed based on the technologies of KBS, computational linguistics, and a forecasting method for automating the narrative construction. Hybrid case-based reasoning (HCBR) has been built by integrating case-based reasoning (CBR) and rule-based reasoning (RBR) technologies. Self-associated concept mapping (SACM) extends the use of concept mapping by proposing the idea of self-construction and automatic problem solving. A Constrained Fuzzy Spreading Activation (CFSA) model has been developed for supporting rapid and automatic decisions inference in SACM. The development of the four algorithms not only contributes to the advancement of the computational intelligent technologies in the field of the present study, but also provides an important means for supporting the building the computational narrative simulation system (CNSS). With the successful development of the CNS methodology and hence the CNSS, an organization is capable of managing narrative knowledge in a systemic manner. The CNS makes efficient and effective use of organizational narrative knowledge and narrative simulation for achieving OL. The CNSS provides an important means for automatic integration of narratives with simulation. As a result, the time for construction of a narrative simulation can be dramatically reduced, and the learners can test their actions and consequences in a more cost effective, faster, appropriate, flexible, and ethical manner. The learners can recognize the choices, decisions, and experience that lead to the consequences of the decisions that have been made. With the successful implementation of the CNS, the capabilities of an organization for knowledge creation, knowledge sharing, knowledge storage, and knowledge application can be significantly enhanced.
Subjects: Organizational learning.
Knowledge management -- Data processing.
Hong Kong Polytechnic University -- Dissertations.
Pages: xvi, 216, 24 leaves : ill. ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of May 28, 2023

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.