Knowledge Discovery Mechanism of Heterogeneous Data: Enhance Elderly LOHAS via Planting System

• Main Authors: Hung-ChihHsueh, 薛弘志
• Other Authors: Yau-Hwang Kuo
• Format: Others
• Language: en_US
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/75247234122542830116

碩士 === 國立成功大學 === 醫學資訊研究所 === 101 === The crisis of the new century issues is still around in the aging population. Therefore, the concept of active is extremely important issue aging at this stage which encourages the elders to participate leisure activities and lifelong learning to enhance elder health vitality and life quality. In this issue, we develop a clicks-and-mortar system - Eden Garden which reaches to participate in leisure activities and enhance the mental LOHAS index. According to the elderly usage demand, we propose a novel knowledge discovery mechanism that employs the fuzzy set theory and text mining techniques. It aims to extract implicit knowledge from heterogeneous data and discovers the most suitable patterns to elders for plant cultivation. Unlike traditional approaches focused on textual knowledge extraction, our research is regarded the relations between nominal features and numerical features as analysis units. In this thesis, we use the context semantic analysis method to define the nominal feature category and estimate the cover range of each category via statistical interval estimation method. Finally, we are according to above steps to construct the fuzzy knowledge extraction model. In addition, we also discover the suitable patterns from the extraction results for user. Simultaneously, we further use the user experience feedback of tuning the discovery method to meet the users’ actual needs. In knowledge extraction experiment, we use the expert questionnaires to assess the reasonable of extraction results and use the expert judgment results as answers to calculate the extraction accuracy. The experiment results are shown that the accuracy of extraction method can be reached 85%. In pattern discovery experiment, we use the simulation way to simulate the actual usage status, then according to the user experience feedback to tune the original discovery method. The pattern discovery experiment results are shown that discovery method can meet users’ actual needs after tuning the mechanism via user experience feedback. Finally, we use the LOHAS index scale table to assess the mental index of elders, and observe the change of index after using the system.