This paper introduces a semi-automated technique for classifying text-based close call reports from the GB railway industry. The classification schema uses natural language processing techniques to classify close call reports in accordance with the threat pathways shown on bow-tie diagrams. The method enables categorisation of a very large number of unstructured text documents where safety-related information has not previously been extracted due to the infeasibility of analysis by human readers. The results demonstrate mixed accuracy in the categorisation of close calls, with the highest accuracy being for the threat pathways that are more frequently reported. This work paves the way to machine-assisted analysis of text-based safety and risk databases, and provides a step forward in the introduction of data analytics in the safety and risk domain. Others working in this area have speculated that approaches such as this could be mandatory for safety management in the future.

a b s t r a c tThe GB railways collect about 150,000 text-based records each year on potentially dangerous events and the numbers are on the increase in the Close Call System. The huge volume of text requires considerable human effort to its interpretation. This work focuses on visual text analysis techniques of Close Call records to extract safety lessons more quickly and efficiently. This paper treats basic steps for visual text analysis based on an evaluation test using a pre-constructed test set of 150 Close Call records for ''Trespass", ''Slip/Trip hazards on site" and ''Level crossing". The results demonstrate that visual text analysis can be used to identify the risks in a small-scale test set but differences in language use by different cohorts of people interferes with straightforward risk identification in larger sets. This work paves the way to machine-assisted interpretation of text-based safety records which can speed up risk identification in a large corpus of text. It also demonstrates how new possibilities open up to develop interactive visualisations tools that allow data analysts to use text analysis techniques for risk analysis.

This paper presents the case for IT transformation and big data for safety risk management on the GB railways. This paper explains why the interest in data driven safety solutions is very high in the railways by describing the drivers that shape risk management for the railways. A brief overview of research projects in the Big Data Risk Analysis (BDRA) programme supports the case and helps understand the research agenda for the transformation of safety and risk on the GB railways. The drivers and the projects provide insight in the current research needs for the transformation and explains why safety researchers have to broaden their skill set to include digital skills and potentially even programming. The case for IT transformation of risk management systems is compelling and the paper describes just the tip of the iceberg of opportunities opening up for safety analysis that, after all, depends on data.

This paper proposes a model to assess train passing a red signal without authorization, a SPAD. The approach is based on Big Data techniques so that many types of data may be integrated, or even added at a later date, to get a richer view of these complicated events. The proposed approach integrates multiple data sources using a graph database. A four-steps data modeling approach for safety data model is introduced. The steps are problem formulation, identification of data points, identification of relations and calculation of the safety indicators. A graph database was used to store, manage and query the data, whereas R software was used to automate the data upload and post-process the results. A case study demonstrates how indicators have extracted that warning in the case that the SPAD safety envelope is reduced. The technique is demonstrated with a case study that focuses on the detection of SPADs and safety distances for SPADs. The latter provides indicators for to assess the severity of near-SPAD incidents.

Abstract:Moving away from standard approaches of safety risk analysis to new approaches that incorporate big data analytics brings with it many opportunities to include new sources of data. These data sources could be the numeric data sources that are used for traditional safety analyses, but could also include text-based sources, such as accident reports, or even social media data feeds. This paper describes an automatic text mining approach to obtain information from close call events (accident "near misses") that can be used for safety management decision-making. The results from this work have shown how automated text mining can be used to extract information from big data sources and be used to inform safety decision-making. Further research in this area intends to look at how the techniques that have been proven to date can be improved with the use of machine-learning techniques.