The use of text mining to improve knowledge discovery in a project environment
Abstract
This thesis investigates how project managers can use knowledge discovery from text on the available project data to improve future projects using a petro-chemical company in South Africa as a case. This concept has been explored and developed in the analysis of scientific databases. Currently, project managers analyse project reports by manually reading through each report and attempting to decipher trends and patterns. The study employed an action research method. Empirical data was collected using interviews on project managers to establish the use of text data mining on the available data and implementing the action plan by analysing the data using the chosen text data mining technique. Other parts of the interviews entailed evaluating with the project managers whether the technique extracts the anticipated knowledge efficiently in a user-friendly format. The findings of the study confirm that the kind of information collected by different projects differ, but in general it includes the project scope, people as well as the terms of reference.
Information collected from projects should be used to learn and predict the future of future projects. Multiple participants concurred that information collected to learn and predict the future may certainly improve future projects. Project managers confirmed that lessons learned should be captured in a database where specific information can be retrieved as and when required. The data analysis process should be measured where lessons learned should show trends automatically and provide useful graphical reports. A well-structured database which can search or retrieve data from various projects to make sense of the lessons learned is ideal. The more predictable the database can be the better and the ability to convert learnings to easily extractable knowledge is important. A system that can show patterns or trends of critical information from the projects is needed.
The findings from this study confirm that there is significant amount of information gathered and created during the lifecycle of each project. In addition, most of the information is essential for the success of current projects and future projects. It is important to know the historic project performance for insights such as cost estimation, pitfalls from previous projects as well as scheduling. The project managers indicated that it is critical to have a centralized repository where all the projects’ information is stored and accessible. A centralized repository would be instrumental in correlating data from similar projects.
Lessons learned reports are critical to the function of project managers in order to avoid re-learning a lesson that has been learnt in another project. It was also confirmed that project managers need a better way of analysis which summarizes the key aspects of the lessons learned as well as best practices. The key insights that were found to be important for project managers include summarization of the numerous reports into a digestible format. Furthermore, the study confirms that text mining could be used to improve knowledge discovery in the project environment. The study therefore recommends the use of Python text mining tool to discover knowledge from past projects. Finally, study proposed an 11th project management area called Project Knowledge Management. The application of Project Knowledge Management contributes the success of projects in project environments.