Large-Scale Corpus Retrieval Algorithm Based on Graph Representation Learning

The human society has entered the era of information explosion, and it is more difficult to obtain the desired information from the mass information. In large-scale corpus research, fast and efficient retrieval algorithms are very important for deeper knowledge discovery and language research. Up to now, there is still no efficient data model to represent the corpus, and extracting information is very time-consuming. In order to solve these problems, this paper conducts research based on graph representation learning, captures the time sequence information of dynamic networks, mines the high-level structural semantic information hidden by dynamic networks, and effectively mines the information contained in network data. This paper mainly studies the algorithms of graph representation learning, including matrix analysis algorithm, random walk algorithm and deep learning algorithm, and constructs a large-scale corpus retrieval algorithm based on graph representation learning. The superiority of the algorithm studied in this paper is verified by comparing the Wikipedia corpus with four baseline models, such as TF-IDF, DE-BERT, Node2vec and DeepWalk.