Article Preview
Top1. Introduction
A smart enterprise is a new management model and organizational form that integrates the new generation of information technology such as big data analysis, Internet of Things, cloud computing, and mobile Internet. It can not only promote the transformation and upgrading of the internal production relations of the enterprise, but also complete the "Internet +" social productivity. The harmonious docking of the company can further release the innovative vitality of the company’s employees and provide the company with the source of sustainable development.
Smart enterprise is a human-machine collaborative enterprise that is built on a data-driven basis and presents the characteristics of AI as a whole. Smart enterprises are required to strengthen the construction of the Internet, deepen big data mining, promote management reform and innovation, and deeply integrate advanced information technology, industrial technology and management technology from the perspective of the enterprise as a whole, so as to realize the digital perception and network transmission of all elements of the enterprise , big data processing and intelligent application, so that the enterprise presents a flexible organizational form and a new management model with automatic risk identification, intelligent decision-making management, and self-correcting and upgrading. Safety production is an important part of the development of smart enterprises. With the initial formation of big data for safety production, emerging technologies such as big data and AI will inevitably become the key to promoting the further development of modern safety management or breaking through the bottleneck. However, the current safety production field is still in the exploratory stage in terms of safety big data collection, storage, and analysis methods. The data often presents the characteristics of scale, source diversity, and low density of valuable information. Security information processing capabilities are difficult to perform the analysis of massive security or accident data, and the valuable information contained in the data still needs to be further explored and utilized, so as to provide effective suggestions for accident prevention. The continuous improvement of the level of enterprise informatization has provided unprecedented perspectives and information perception capabilities for safety management. At the same time, it has accumulated a large amount of safety information data, which has brought huge "dividends" to safety production supervision, and is more efficient and in-depth prevention. The accident provided an opportunity.
However, despite the fact that enterprises have accumulated a large amount of accident data resources for production safety, the corresponding lack of production safety information processing capabilities has led to the ineffective use of the value of massive data, which in turn restricts the in-depth study of accidents, and enterprise safety managers cannot learn from historical accidents. In time to learn from experience and effective prevention, leading to the repeated occurrence of similar accidents. With the rapid expansion of information resources, production safety data also tends to be diversified and complicated. Compare with traditional structured data, unstructured data such as text, audio, pictures and videos contain more safety accident information, gradually become the main form of recording accidents. However, the amount of these accident data is huge and the structure is complex. At present, its utilization rate is relatively low, and there is still a huge space for mining and utilization. Therefore, how to use AI to efficiently and fully tapping the potential value of the massive accident data to prevent work safety accidents have become a hot topic of current research. However, the current production safety information processing capabilities of enterprises cannot match the increasing amount and complexity of safety information data, traditional data analysis methods can no longer effectively process this information. However, with the rise of AI technologies such as machine learning and data mining, it provides a way to realize deeper mining and analysis of massive accident data.
The main contributions to this paper is We construct a mining process for the cause of safety production accidents based on the Latent Dirichlet Allocation (LDA)topic model.The rest paper is structured as follows. Section 2 introduces related work about data mining and machine learning technologies to classify accident data. In Section 3, text-based accident data mining process is analyzed. The experimental results are reported in Section 4, and make suggestions for smart enterprises in safety production accident prevention, finally Section 5 concludes this paper.