Application of Text Mining for Explicit Knowledge Retrieval at Kenya Coastal Development Project

Abstract

This study applied and evaluated a Text Mining model and tools in the retrieval of explicit knowledge at the Kenya Coastal Development Project (KCDP). The study identified textmining techniques that could be used to develop and evaluate a text-mining model for the task, complemented with system analysis of the existing systems through surveys of 52 purposively sampled relevant staff of the KCDP using a questionnaire and focus group discussions (FGD) to collect data to establish the current situation at the KCDP in terms of records and knowledge management systems in place and how text mining may be used to retrieve knowledge from the existing systems. Text data were also collected from the project and related websites, using various Python programming language libraries including Python Request 2.22 and Beautiful Soup 3, and summarized using algorithms that included Luhnsummarizer, Lsansummarizer, Lexranksummarizer and Edmondsummarizer. Topic modelling was also performed with the text data using Latent Dirichlet Allocation (LDA) topic-modelling algorithm, and the model was evaluated to establish its performance. It was concluded that text mining and analysis could be used to analyze explicit knowledge from both structured and unstructured electronic data at the KCDP using the model. The study recommended that more research be done in the development and evaluation of proposed models and text analysis tools and code libraries should be developed to support other languages than English, such as Kiswahili