In multimodal text information extraction, there is inconsistency in the temporal and spatial alignment of multimodal data such as text, audio, and images, which leads to a decrease in the fusion performance of multimodal data and a decrease in the final extracted information coverage and F1 value. To this end, research is conducted on multimodal text information extraction algorithms based on knowledge graphs. Firstly, utilizing Hidden Markov Models and Convolutional Neural Networks to extract multimodal text and image feature information. Based on knowledge information and combined with the Neo4j graph database, a multimodal text knowledge graph is constructed, with entities as nodes, relationships as edges connecting nodes, and entity attributes as additional information for nodes. This intuitive representation makes the connections between complex and difficult to understand multimodal data clear and visible, effectively solving the problem of data alignment. Then, the depth first search algorithm is used to deeply explore the implicit knowledge in the knowledge graph, extract degree centrality, graph density, and betweenness centrality feature items, and construct a multimodal text information extraction feature set. Finally, by establishing an information extraction model based on BiLSTM and training the model using gradient descent, the obtained multimodal text information extraction feature information is input into the trained BiLSTM model to achieve multimodal text information extraction. The experiment shows that the algorithm achieves an F1 value of 0.94 or above when extracting text information from different modalities. Compared with other methods, it has significant advantages in terms of information coverage of the extracted results.
HuDai-wang, JiaoYi-yuan, LiYan-ni. Novel and efficient algorithm for entity relation extraction with the corpus knowledge graph[J]. Journal of Xidian University (Natural Science Edition), 2021, 48(6): 75-83.
LinZe-ning, WangJia-peng, JinLian-wen. Visual information extraction deep learning method: a critical review[J]. Journal of Image and Graphics, 2023, 28(8): 2276-2297.
LiYu-qi, LiLong. Simulation of web repetitive information extraction based on pattern recognition algorithm[J]. Computer Simulation, 2022, 39(3): 439-443.
JiXu-rui, WeiDe-jian, ZhangJun-zhong, et al. Research progress on information extraction methods of Chinese electronic medical records[J]. Computer Engineering and Science, 2024, 46(2): 325-337.
LaiJuan, HongYan-wei. Deep learning network based on rule constraints for text information extraction[J]. Computer Engineering and Design, 2021, 42(12): 3548-3554.
FangPing, XuNing. Graph model summary extraction algorithm based on BERT bidirectional pretraining[J]. Application Research of Computers, 2021, 38(9): 2657-2661.
ZuXian, XieFei, LiuXiao-jian. Keyphrase extraction combining word and document embeddings[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(2): 294-304.
[15]
SmithE, PapadopoulosD, BraschlerM S K. Lillie: information extraction and database integration using linguistics and learning-based algorithms[J]. Information Systems, 2022, 105(3): 101938.
WangXiao-yu, WangFang. Keyword extraction from a paper's abstract based on semantic text graph[J]. Journal of the China Society for Scientific and Technical Information, 2021, 40(8): 854-868.
JingLi, YaoKe. Research on text classification based on knowledge graph and multimodal[J]. Computer Engineering and Applications, 2023, 59(2): 102-109.
ZhaoYi-ming, WuLin-rong, RenXiao-xiao. Chinese text semantic graph construction based on multiple knowledge graphs[J]. Information Science, 2021, 39(4): 23-29.
ZhuBing-bing, LuoFei, LuoYong-jun, et al. An automatic text summarization algorithm based on clause extraction[J]. Journal of East China University of Science and Technology(Natural Science Edition), 2024, 50(1): 114-120.
ZhaoYe-hui, LiuLin, WangHai-long, et al. Survey of knowledge graph recommendation system research[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(4): 771-791.