Electoral Recommender System for Indonesian Regional People's Representative Councils (DPRD) Using Knowledge-based and collaborative filtering approach.

Abstract

General Election is the most prominent political activity in the election of people representatives in democratic countries. In the Information era, the veracity and volume of available information opportunities to inform the decision-making of election candidates. This research outlines the development of a Business Analytics Electoral Recommender System. The system aims to provide recommendations for political candidates based on various factors such as demographics, political preferences, and socio-economic status. We also propose an election candidate's recommender system using a hybrid of collaborative filtering (CF) and a knowledge-based (KB) model. The KB approach used a web crawler to scrape and process relevant information on various internet sources and social media to nominate the representative candidates based on the voter's preferences. We used a hybrid approach of The Cross Industry Standard Process for Data Mining (CRISP-DM) to the web scraping process of unstructured and semi-structured static and dynamic information of representative candidates. The recommendation uses the CF approach based on criteria for informed electoral decision-making. The results from user acceptance test showed the score are 86% for ease of use and 78.5% for the perceived usefulness of the proposed system's key features. The average percentage of behavioral intentions, and attitudes towards the actual use and utilization of the system, was 88.5%, which concludes that the respondents were generally satisfied with the proposed system.