Facing the profound changes in the media environment in the new media era, exploring the construction of an autonomous knowledge system of communication studies is of fundamental and directional significance to its disciplinary development. Under the guidance of the overall framework of knowledge mapping, a set of autonomous knowledge systems for communication science is being constructed. A feedback mechanism is introduced to the preprocessed communication studies corpus for ontology construction. In terms of data layer filling, a joint information extraction model based on Bert-BiLSTM-CRF-Att is proposed, after which the fusion and storage of the knowledge graph are realized through semantic similarity and Neo4j to complete the construction of the knowledge graph of the autonomous knowledge system of communication studies. Conducting experiments on the joint information extraction model, the model has a high recognition speed, and the F1 values of entity recognition and relationship extraction are 77.77% and 88.47%, respectively, which are better than the comparison model, and the experimental results reflect that the joint information extraction model has a better recognition efficiency and recognition accuracy, which proves the effectiveness of the information extraction model of this topic in the construction of the autonomous knowledge system of communication science. This paper’s construction of an autonomous communication knowledge system based on the knowledge graph can enhance the efficiency of communication knowledge acquisition and provide a new direction for the integration of communication knowledge.