Akademik Veri Yönetim Sistemi
Journal of Scientific Reports-B, sa.010, ss.1-12, 2024 (Hakemli Dergi)
In today's world, making millions of data understandable has become
important. To take faster steps in criminal matters, especially by using these
data, data analysis should be done quickly. In this context, sentiment analysis
performed with the natural language processing (NLP) method of artificial
intelligence enables the elimination of possible loss of life and property. In
addition, by listening to all radio frequencies at the same time in possible
terror areas, the attacks of terror organizations can be analyzed with natural
language processing methods, so that the attack can be prevented before it
takes place. In this study, natural language processing methods of artificial
intelligence were used in the analysis of text, audio, and image data in the virtual
environment for the detection of terror threat elements. In this way, it is
aimed to ensure the healthy intervention of law enforcement officers and the
security of life by analyzing the talks of terror elements in terror zones. For
this purpose, an 85% accuracy rate was reached with the word/sentence vector
creation method GloVe in the first model created with the Spark NLP library on
textual data. In addition, a 74% accuracy rate was achieved with the LSTM
method on audio data, while a 71% accuracy rate was achieved with the GRU
method on visual data.