Design And Implementation Of Objectionable Mobile Application Monitoring System Based On Textual Features

Because of the lack of effective supervision,many mobile applications contain a lot of objectionable information,such as pornography,violence and reaction,which not only endangers the physical and mental health of adolescents,but also all kinds of untrue propaganda that have an adverse impact on social stability.In objectionable text detection,feature thesaurus is the premise of the filtering effect,but the current lack of objectionable mobile applications feature thesaurus;The common keyword list-based detection methods lack the semantic analysis of the texts and are prone to high false positives.Traditional topic model filtering methods encounter the problem of feature sparseness of short texts.In order to effectively detect objectionable applications,this paper propose a method to monitor the objectionable applications in the application store by analyzing the description information applied in the application store,setting up the monitoring system of the objectionable application based on the textual feature.The contributions of this thesis are as follows:1.By introducing the TF-IDF algorithm,the method of selecting objectionable textual features is improved.The method divides the weight of feature words into sensitivity and importance.By introducing the TF-IDF algorithm,the word frequency of characteristic words is counted,and the sensitivity of characteristic words is calculated.Finally,we can get the weight of sensitive words through the formula.Experiments show that the method proposed in this paper has a better detection effect than the previous feature extraction method.2.Establishing a method based on BTM adaptive topic model.The data sparseness problem of the application description information is solved by introducing the BTM topic model and a criterion is established to select the best K value of the BTM model by introducing the weight of the signature database.3.Establishing an objectionable mobile applications detection engine.By introducing the application store vote mechanism and the application store distribution credibility mechanism design detection engine.The experimental results show that the detection engine can effectively reduce the false alarm rate of the adaptive topic model.4.Designed and implemented of an objectionable mobile application monitoring system.From the aspects of the design idea,outline design,detailed design and effect display of the objectionable mobile application monitoring system,we have made overall planning and detailed description for the monitoring system of the objectionable mobile applications monitoring system.Combined with the list-based filtering method,the system adds a blacklist and whitelist,a manual review mechanism,etc.to enhance the robustness of the system.