網絡是一種數據表示形式，目前已被廣泛應用在多個領域中，例如在社群網絡中，我們將節點視為個人或群體，節點之間的邊則稱作鏈結，而鏈結預測的核心概念是藉由分析網絡中節點之間的交互作用，來推斷節點之間是否存在新關係，或是挖掘網絡中的隱藏鏈結，而有效的網絡分析能夠使我們對數據背後的內容有更深入的了解。 目前鏈結預測已被廣泛的應用在社群網絡、電子商務、生物資訊等各個領域，透過鏈結預測，可以幫助研究人員了解網絡的樣貌，並從中挖掘訊息來間接的反映現實生活中的情形。在鏈結預測中，透過網絡嵌入的方式，能夠將網絡中的節點訊息投射到低維向量空間中，並有效的保留網絡結構。在本論文中，我們將採用三種網絡嵌入的方式，分別是：Matrix Factorization based methods、Random walk based methods 以及Deep learning based methods ，每一種網絡嵌入法都有各自的優缺點，因此我們提出一個集成學習模型來保留每一網絡嵌入的特性，透過不同的網絡嵌入學習節點表示。我們在五個資料集上進行實驗，結果顯示利用多個網絡嵌入表示法的學習，透過多個不同的分類器進行訓練，最後以深度神經網絡作為最後的結果預測，能有效提升鏈結預測的準確率。;Network is a form of data representation, and it has been widely used in many fields. For example, in social networks, we regard nodes as individuals or groups, and the edges between nodes are called links, which means the interaction of the people. By analyzing the interaction of the nodes, we could learn more information on the relationship of the network. The core idea of link prediction is to predict whether there is a new relationship between the pair of nodes or to discover the hidden links in the network. Nowadays, link prediction has been used in social networks, e-commerce, biological information, and other fields. Moreover, researchers use graph embedding for link prediction, which effectively preserves the network structure and converts the node information into the low-dimensional vector space. In this study, we use three graph embedding methods: Matrix Factorization based methods, Random walk based methods, and Deep learning based methods. Each method has its own strength and weaknesses, so we propose an ensemble model to combine these graph embedding to a new representation for each node. The new representations will be regarded as the input of our link prediction model. The performance evaluations are conducted on multiple datasets. Experimental results show that using multiple graph embedding for representations can effectively improve the performance of link prediction.
