Short Papers
Agricultural Information Systems, Association Rule Mining, Recommender Systems
The aim of this work was to design, develop and deploy a web recommender system based on association rules to automatically make recommendations of agricultural content, according to the profile of a user community. The data used in this study were extracted from a database of accesses to the site of Embrapa Information Agency. The user visits were stored in a structure of access lists, and from such lists, association rules between pages were generated. The set of rules led to a knowledge base that was used to make content recommendations to users. The system was evaluated using a metric called bounce rate, so that by means of statistical tests it was possible to evaluate the impact of these recommendations. The results revealed that through recommendations, users find relevant information associated with their visits and increase their time spent on the site. The system was validated for the sugarcane crop, but it can be easily extended to provide recommendations regarding other crops.
Flavio Barros, Leandro Oliveira, Stanley Oliveira