GOOD: Genetic ordering of documents

Scientific research has become a critical activity for discussion of ideas and dissemination of knowledge in academic communities. To execute this activity, researchers use recommendations systems aiming at retrieving relevant information. Despite the great successful of these kind of software, users demand more refined searches and tasks that involve a strong semantic analysis. For example, recent studies have focused on mechanisms to organize Web news logically, measuring the consistency between text fragments, improving scalability the recommendation systems, enhancing the search space and improving of bibliography. We present GOOD, an original approach to permute scientific articles in order of generating a coherent and comprehensive sequence of articles from an initial article, chosen by a user. Our mechanism uses a genetic algorithm that searches for patterns of consistency between documents. The goal is to generate one or more sequences of articles for facilitating a better progression of the reading and to aid the user in arranging of a substantial amount of documents. We conducted experiments to determine the most appropriate configuration of coherence criteria and genetic operators. Finally, we tested the effectiveness of good compared to sequences generated in random manner and ones organized by date. Our tool proved to be the best in 80% of the results.