Evaluating the Effectiveness of Spam Message Classification and Detection based on Deep Bi-GRU Method

More and more people in the modern information era rely on SMS because of its low cost, great mobility, and ease of usage. However, as SMS has gotten more popular, spam messages have emerged as a serious problem, with detrimental effects on not only people's daily lives but also public safety and social harmony. In light of this, research into the creation of technology for the intelligent classification of spam communications has assumed a higher priority, and spam message filtering has become an urgent issue demanding quick attention. The first three parts of the proposed method are preprocessing, feature selection, and model training. Chi-square and document frequency are employed for preprocessing. This study employs a genetic algorithm to select features. The models are then trained after feature selection with D-Bi-GRU. The proposed approach surpasses the two most common alternatives, GRU and Bi-GRU.