為了在同儕網路中要達成搜尋有效性及準確性，因此發展了社群式同儕網路這類新興的技術。從目前相關的社群式同儕網路研究中，本研究發現這些社群式同儕網路的確改善非結構式同儕網路搜尋效率低落的問題。然而，從過去研究者提升非結構式同儕網路搜尋效率的角度來觀察社群式同儕網路，本研究發現有3點因素是影響社群式同儕網路的搜尋效率。1.記錄正確回應節點問題，2.支援語意搜尋問題，3. 維護節點列表問題。因此，本研究發展出一種「社群關聯式同儕網路」，它是利用海伯法(Hebbian Rule)來設計社群式關聯程度機制，這個社群式關聯程度機制就是讓節點與節點之間所形成的人際互動都有權重值。它的重要特色除了達到讓正確回應問題的節點能夠獲得與詢問節點較高的權重值外，還增加了機器學習的能力，讓每一個節點可以在搜尋過後會調整其權重值以增進搜尋效率。 In order to improve search performance and accuracy, social-like P2P Networks are developed in last years. Our research discover that these methods in social-like P2P Networks can improve search performance in unstructured P2P Networks. However, We find there are three factors that can influence search performance in social-like P2P Networks. First, how to record the peers which have positive response. Second, how to calculate semantic similarity for searching. And third, how to maintain the peer profile. We use Hebbian rule to design the mechanism for calculate the associated weights of peers when they have social interactions, called ‘Associated social-like P2P Networks’. The distinguishing features for improving search performance are the peers with correct responses have higher weights, and adjust the weights by the ability of learning after searching.
