A Semantic Retrieval Algorithm for Enterprise Archives Based on Bert-BiGRU-CRF-Harris Hawk Model

Abstract

Abstract The extracted enterprise archive text is depicted through relationship extraction and semantic analysis to simplify the archive retrieval efficiency and reduce communication costs. As a foundational work for newcomers, it begins with a comprehensive review of the latest developments in semantic retrieval technologies, offering a rapid acquaintance with contemporary research trends. A novel algorithm is designed to enhance the precision and accuracy of document retrieval in a corporate context by leveraging semantic representation. The core of the proposed methodology is an integration of the BERT base model with a BiGRU layer and the innovative application of the Harris hawk optimization algorithm. The BERT + BiGRU combination is utilized to conduct the semantic matching task within enterprise archives, while the Harris hawk optimization refines high-level semantic vector representations.For empirical validation, the study introduces an enterprise archive retrieval dataset and compared to Bert,Bert_BIGRU_CRF,Bert_BIGRU,BIGRU algorithm, the results evidently support the superiority of the BERT + BiGRU + CRF + Harris Hawk model, showing remarkable performance advancements over conventional algorithms.