Data and Text Mining for the Detection of Fraud in Public Contracts: A Case Study of Ecuador's Official Public Procurement System

Corruption is present in different forms and typologies, directly affecting the execution of both public and private contracts. The doctoral thesis aims to establish a methodology to prevent and detect corruption automatically in public procurement. By using machine learning techniques and Natural Language Processing (NLP), algorithms for detecting and predicting favouritism and oligopoly are developed. In addition to detecting corruption and its types in the Ecuadorian Public Procurement System (SERCOP) and also visualising the results in an appropriate way, in order to detect and prevent future acts of corruption. In order to analyse the feasibility of the study, a mapping and systematic literature review was carried out, allowing the hypothesis and the methodology to be followed in order to execute and evaluate the developed algorithms. Finally, the detection of favouritism based on process qualification parameters and types of contracting is tested.