Data Collection and Classification for Dataset creation with Multi agent system improved by Deep Learning

Abstract

Due to the expanding growth of information in the entire world, humans create computers to handle this huge quantities of information because of that computers become more integrated into our day-to-day lives and technology continues to advance and became more complicated to deal with the development of information, but in the last years new generation of computers appear which is the 5 th generation of computers that need data to work so we fall in a new problem which is the absence of organized data (dataset), to solve this problem we propose an approach to create a dataset with Multi Agent System (MAS) by collecting data from the web and classify it with an supreme deep learning model. In addiction to that we use our model to analyze opinions about a subject. We test our model with both of types pictures and text, for pictures we create a collection of 2578 that was perfectly labeled, and for the text we creat e a dataset about a subject of 100000 text and classify it using our Deep Learning model and with this dataset we trained a model to classify text about specific subject we observe that this last model can classify text for this subject perfectly better than the first model. To present our work, we create an application, which offers users the possibility to collect and classify data, the users can also check opinions of others in real time or for date interval.