Back to results list
Please use this identifier to cite or link to this item:
|Title:||Computational organizational narrative generation for decision support learning||Authors:||Yeung, Chui Ling||Advisors:||Cheung, Benny (ISE)
Lee, W. B. (ISE)
Tsui, Eric (ISE)
Knowledge management -- Data processing.
|Issue Date:||2015||Publisher:||The Hong Kong Polytechnic University||Abstract:||The ability of knowledge workers to make decisions is one of the most important capabilities for the survival of an organization. Learning by trial and error may produce incorrect decisions that can lead to serious outcomes and the lessons learnt would be expensive. This is particularly true for some high-risk industries where any wrong decision may lead to industrial accidents which cause injuries or fatalities. Researchers have developed various decision support systems (DSS) in different domains over the years.However, the current DSSs only aid knowledge workers in decision support by showing different statistics. The information is too abstract and cannot motivate and educate workers to learn deeply how to make decisions. Narratives providing detailed information regarding events are well recognized as a powerful technique to conduct knowledge transfer, knowledge sharing and knowledge retention. It not only helps decision makers to establish associations between the events and their resolutions during the decision-making processes, but also provides a useful means of motivating and teaching novices how to observe the environment and make decisions. However, narratives are costly as traditional approaches in narrative construction are labour-intensive, expensive and time-consuming. In some high-risk industries, a new narrative is generated after an accident or the occurrence of a tragedy. The opportunity cost of generating a new narrative for a lesson learnt is relatively high as compared with the traditional approach. In this research project, a methodology of computational organizational narrative generation (CONG) was established to generate narratives in a semi-automatic manner. A prototype of CONG named computational organizational narrative generation system (CONGS) was built. Several intelligent computational algorithms were used to support the automation of the narrative generation.
In order to evaluate the capability of CONGS, a study was carried out in the construction industry. In the quantitative evaluation,CONGS had better performance when compared with existing tools. The precision and recall rates of CONGS were around 75% while that of the existing tools were around 50%. In the qualitative evaluation, construction experts were invited to evaluate the performance of CONGS. The questionnaire results indicate that the experts are satisfied with the performance of the systems. Of the respondents,75% agreed that the new narratives generated by CONGS can be classified as a new narrative for training and learning. An experiment was designed to measure the learning outcome gained by the participants through reading the narratives generated by CONGS. The results indicate that the generated narratives can facilitate readers to learn the important information in the narratives. These narratives can act as an alternative source for training or educating novices to learn how to make decisions. Organizational narratives can be systematically analyzed and retained with the successful development of CONGS. New narratives can be constructed in a semi-automatic and dynamic way. This not only allows knowledge workers to analyze a problem before it actually occurs, but it also helps knowledge workers to explore possible scenarios from new narratives and learn new problem solving methods so as to improve the quality of their decisions.
|Description:||PolyU Library Call No.: [THS] LG51 .H577P ISE 2015 Yeung
xix, 255 pages :color illustrations
|URI:||http://hdl.handle.net/10397/40916||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b28391792_link.htm||For PolyU Users||208 B||HTML||View/Open|
|b28391792_ira.pdf||For All Users (Non-printable)||8.35 MB||Adobe PDF||View/Open|
Citations as of Dec 16, 2018
Citations as of Dec 16, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.