SecureCrowd: Fraud Detection and Trust Analysis in Crowdfunding

Abstract: Crowdfunding has emerged as an accessible means of raising financial assistance for medical needs, education, disaster response, social welfare, and community initiatives. Despite its benefits, the open and decentralized nature of online fundraising can expose donors to fraudulent campaigns, inaccurate information, inadequate verification, and uncertainty regarding campaign credibility. This paper presents SecureCrowd, a web-based crowdfunding platform designed to support fraud identification and trust-oriented campaign verification through machine learning and administrative assessment. The proposed system enables campaign creators to submit fundraising requests together with relevant supporting documents, while administrators evaluate campaigns using machine- learning predictions and documentary evidence before approval. Logistic Regression, Decision Tree, and Random Forest classifiers are employed, and their individual predictions are combined through majority voting to obtain the final classification. Campaign characteristics such as fundraising category, intended purpose, required amount, beneficiary information, urgency, previous campaign activity, description-related attributes, contact information, and supporting-document details are considered during the analysis. SecureCrowd is developed using Python and Django, with SQLite for data storage and HTML, CSS, JavaScript, Scikit-learn, and Pandas for application and data-processing functions. The platform provides role-based access, campaign management, document handling, donation tracking, verification workflows, and fundraising-progress visualization. Functional testing indicates successful operation of the major authentication, campaign, verification, donation, database, and dashboard functions. The proposed platform demonstrates the potential of combining machine-learning-based analysis with human administrative verification to promote greater transparency and confidence in online crowdfunding.

Keywords: crowdfunding, fraud detection, trust analysis, machine learning, campaign verification, Django, donation management