Financial Network Representation Learning for Time Series Investment Behavior

Stock market is one of the important parts of modern financial networks. The mass data of stock transaction contain valuable but sparsely distributing information. How to implement effective data analysis and excavate commercial value on financial data have been appealing to the whole industry and academia. The traditional research on stock network mainly starts from the correlation of stock prices to build a trading network and conduct research, ignoring the role and value of equity investment information. Financial network representation learning for time series investment behavior, combined with the special nature of financial transactions in the stock market on the basis of the traditional network learning based on random walk, a high efficiency financial network random walk model is proposed, mapping the timing concept into the node interaction tour mechanism while taking into account the investment behavior, from which the embedded vector is more able to support downstream machine learning tasks such as funding community discovery.