Using text mining and machine learning to leverage unstructured data and detect EWS: A case study of a construction project

Using text mining and machine learning to leverage unstructured data and detect EWS: A case study of a construction project

Abstract- Knowledge Management became the focus of scientific study during the second half of the 20th century. During this time, researchers discovered knowledge resource importance to business organizations. Contrary to early expectations of enhanced management of documents, Management techniques and systems applied in the construction industry fail to deliver the desired performance [1]. Recent research utilizes document content analysis to improve categorization of documents and support retrieval functions. Document text analysis can be performed efficiently using natural language processing. Since project professionals are poor at detecting early warning signs (EWS), identifying the barriers for this cause is critical [17]. Project assessment is useful in identifying EWS associated with the project formalities. This article delivers an unparalleled way to improve the organization of information in organizations and access to inter-organizational systems. The basis is on automated classification of project documents in the construction and in-line with their related project components. Machine learning methods were used for this purpose.
Keywords: Machine learning, early   warning signs, Construction project   management,   unstructured             information, Project assessment

I.     Introduction

The complexity of construction projects makes them prone to failures. Therefore, the introduction of new procurement methods means that many contractors have been forced to rethink their approach to the way risks are treated within their projects and organizations [16]. Project management methodologies can only minimize the risk of failure but cannot guarantee successful completion. However, early prediction of future project trajectory can provide sufficient early warnings and enough time to respond in case significant deviations from the plans are predicted [10]. Such capability requires rapid assessment of project documentation and reports and the ability to infer potential future failures from unstructured information, analysis of emotion and sentiment in the writing style of those reports. Such capability is currently not routinely available.
The research seeks to develop a predictive early warning methodology for project failure prediction by analysing unstructured project documentation such as project reports [15]. Machine learning will be employed to extract from such sources actionable information to compare against project plans and key performance indicators. The research adheres to the developing a project progress assessment methodology encompassing factors affecting project performance. In addition, develop a method and algorithmic approach for the analysis of unstructured project documentation and extraction of key actionable information to allow inference of actual project progress rapidly [9]. The algorithm accompanies the development of a prototype tool tuned against a series of case studies from the literature. Finally it concludes by conducting an empirical study demonstrating the approach.

II.     Why construction projects?

The construction industry is characterized by constant changes. As such, document classification requirements and needs becomes paramount.  In achieving this classification, consideration issues taken into account include:
 • Construction projects are unequalled [5]. Design specification documents contain the plans, formulas, characteristics and implementation plans for construction. The information is both graphic and textual and is communicated to all the stakeholders in the project. Such availability makes construction projects unique. Taking this consideration into account makes an emphasis on these projects relevant to leveraging text mining.
• Dynamic processes are adopted in the construction industry [6]. The design, construction and maintenance is a cycle subject to change over time. There are different concurrent variables that affect the implementation of development projects. Therefore, constant communication must be implemented between the different stakeholders to keep the project phases in sync. The monitoring and tracking poses a challenge which consequently makes construction projects a key area of interest. 
• Construction projects are structurally organized [3]. Each structure is assigned a project team composed of contractors, owners, designers and representatives. The independence of the structures makes construction projects an interesting area of study.
• Despite being structurally organized and independent, there is increased collaboration between the projects that involve exchange of information and data to streamline the performance of the system. Through the collaboration, the realization of differences in size and IT capability emerges. The differences act as a foundation for the design of a classification construction management system.

III.     Early warning signs

Warning signs are prevalent in construction projects. From definition, they are observational signs that form the basis of   proof to the existence of some incipient positive or negative issue [17]. EWS characterize future developments [2]. According to Ansoff’s 1975 [3], there are two available options to a firm that considers preparing against a strategic weak signal surprise. The first option involves a crisis management strategy [9]. The approach ensures that in the event of weak signal communication detection, the activities of the firm are not negatively influenced to a large extent. The second is a mitigation approach where the problem is pre-determined and mitigated to reduce chances of strategic surprises [9]. Attention must be paid to manage both approach to guarantee their success. According to Loosemore, there are three crisis types in a construction project.
1.      Creeping crisis- a type of crisis that is just perceive and not addressed until the effect of the crisis occurs
2.      Sudden crisis- discovered as crisis that occur without prior warning.
3.      Periodic crisis- they occur in cycles some of which are consistent while others are not.

To curb and minimize the effects of crises, contingency plans should be adopted. Implementing such strategies accompanied by a team of professionally trained project management, the influence and impact of the crisis is drastically reduced [15].


An amazing way to arrange the unstructured data in the symmetry is through text mining and students can check online assignment help to get quality work. Thank you for sharing this technique with us as the data is in huge and huge amount sometimes it becomes difficult to handle and manage that data. I hope this will work better.