This research is devoted to the problem of search and selection of educational resources. The main objective is to create a recommendation system for optimizing an individual educational trajectory using machine learning methods.
- Choose the most appropriate method for text classification.
- Prepare a dataset for analysis.
- Build a mathematical model and train it on the prepared data set.
- Develop an interface.
During the study of machine learning methods, it was found out that the most suitable method is an artificial neural network. To train it, a dataset was prepared, which consists of the texts of publications from the groups of the social network “VKontakte". The texts were divided into categories describing their educational subject. The TensorFlow library was chosen to build a mathematical model as a widely used platform in machine learning. After studying the documentation, a neural network was built and trained to classify text. A program (chatbot) that interacts with the API (application programming interface) VKontakte was developed to let users conveniently interact with the classifier. Thus, the model can be used through a dialog in the social network. The program analyzes the user's current subscriptions and suggests several groups that are suitable for their estimated interests.
In conclusion, it should be noted that the objectives set at the beginning of the research were accomplished, particularly:
- A suitable method was chosen - a neural network.
- A dataset which consists of the texts of publications has been prepared.
- A neural network that classifies text into specific categories has been built.
- An interface in the form of a VKontakte chatbot has been developed.