Extractive convolutional adversarial networks for network embedding

A Scholar Disambiguation Method Based on Heterogeneous Relation-Fusion and Attribute Enhancement

“…The name disambiguation method based on network representation learning is to use the relation of cooperative, citational and other relational networks in the literature to mine scholars' features [12]. The existing network representation models mainly focus on topology structure while ignore the information from node attributes which is potentially valuable to network embedding [3].…”

“…is proposed to differentiate different types of feature relations; and 5) for the problem of feature information missing, the SPCNE model and attribute enhanced representation model are fused into a new disambiguation network model called HRFAENE (Heterogeneous Relation Fusion and Attribute Enhanced Network Embedding Model). The HRFAENE model integrates both the document attribute features [3] and the feature network structure information (multiple relations) ( Figure 2) which are very important in further improving the accuracy, uses weak features as node attributes in strong feature networks, and iteratively learns network structure information and node attribute information with an aim to better identify the disambiguation entity. The experimental results showed that the HRFAENE model has much better disambiguation accuracy than existing models and good stability, indicating that the proposed model in this paper can be used for effectively disambiguating duplicate names in reality and therefore greatly improving the accuracy of information retrieval.…”