Construction Safety Risk Management Based On Natural Language Processing

The construction industry occupies an important position in the Chinese economy.However,due to its own complexity and backward management level,construction safety accidents frequently occur in construction projects.Therefore,improving the technical level of construction safety management is an urgent problem to be solved.At present,the field of construction safety risk management mainly relies on manual reading of construction reports for construction safety risk prediction,and the reuse rate of existing safety accidents case and technical specifications is low.Based on this,this paper combines natural language processing,case-based reasoning techniques and construction safety knowledge,establishes construction safety risk prediction and case-based reasoning models,automatically identifies hidden safety hazards in construction reports,and find similar cases for reference of accident experience to put forward solutions for new risks,and to realize knowledge reuse.This article first combs the application status of construction safety risk prediction,accident case reasoning and natural language processing in the engineering field,analyzes the shortcomings of construction safety risk management at this stage,and points out the use of natural language processing and case reasoning technology to achieve feasible automated risk prediction and case reuse.Secondly,this paper collects 2050 accident cases and related construction safety regulations on government websites to establish a case library,uses natural language processing technology to carry out corresponding pre-processing work,and uses the Albert-Bi LSTM-CRF model to realize construction named entity recognition.Then,based on relevant laws and regulations and construction safety knowledge,construction safety risk rules are established,and if-then is used to match semantic rules to realize the safety risk warning of construction reports.At the same time,to improve the efficiency and performance of construction safety risk processing,this paper uses the TF-IDF algorithm to calculate word feature values,proposes a word meaning extension based on the Word2 vec model,and realizes the retrieval and experience reuse of similar accident cases through space vector model and improved cosine similarity formula.Finally,the new risks and corresponding treatment measures found in the construction report are stored in the original case database as new case samples for the convenience of the next risk inquiry.This paper introduces natural language processing and case-based reasoning into the research of construction safety risk management,which not only realizes the safety risk prediction of automated construction reports,but also realizes the reuse and sharing of case knowledge,which provides a reference for the development of construction risk management towards intelligence.