| Summary: | 碩士 === 輔仁大學 === 資訊管理學系 === 93 === Knowledge has become the most important production element in the era of knowledge economy. Knowledge contains two parts - explicit knowledge and implicit one. If and only if we understand the two parts of knowledge, we say we understand knowledge.
As the progress of information technology, virtual community in the Internet becomes the main platform to share knowledge. However, because of the characters of the post in the virtual community, the contented-based recommendation system does not fit. Moreover, collaborative recommendation system gets the problem called “ratings sparsity”. In the other way, the current recommendation systems do not consider the social relationship which is an important issue when people share knowledge.
This thesis implemented 6 recommendation modules based on 6 measures which are used to estimate the social relationships between two members in a forum – a kind of virtual community in the Internet. When some member A creates a new topic, the recommendation modules will recommend people who are willing to discuss with A.
This thesis used the data of a virtual community to understand the forecasting ability of the 6 recommendation modules based on social relationships. The experiment result shows that the greatest forecasting ability of recommendation module is based on the “mostpost” social relationship measure. In addition, computing relationship in the light of some specific members, not all of members, can increase the forecasting ability of recommendation modules, no matter based on what kind of measures.
|
|---|
