This week, Machine Translation said, | learning recommendation system and chat robot the latest research progress of science and technology – Sohu introduction this period a total of four PaperWeekly share arXiv recently released the high quality paper, including Machine Translation, said the study, recommendation system and chat robot. Artificial intelligence and related research with each passing day, this article will take you to understand what the four research directions are the latest developments. Four paper respectively is: 1, A General Framework for Content-enhanced Network Representation Learning, 2016.102, Collaborative Recurrent Autoencoder: Recommend while Learning to Fill in the Blanks, 2016.113, Dual Learning for Machine Translation, 2016.114, Two are Better than One: An Ensemble of Retrieval- and Generation-Based Dialog Systems A General Framework for Content-enhanced 2016.10 Network Representation Learning author Xiaofei Sun, Jiang Guo, Xiao Ding and Ting Liu for Social Computing and Information Center Retrieval Harbin Institute, of Technology, China network representation content-enhanced the keywords, arXiv problem using network structure features and text features to learning network The embedding model of the node in the network in general this paper thinking is clear, the method of learning to a large extent reference to the word2vec method. For a node V, the node will be connected to the V as a positive example, do not want to connect to the node as a negative example. So how to integrate into the content? The virtual content node C is set in the network, and the c_v content of the V node is described as a positive example, and the other as a negative c_v. At the same time, we consider the similarity of the network and the similarity of the text, so that the vector of V is close to the positive case. The total optimization function is shown as follows相关的主题文章: