Conference material: "Proceedings of the International Conference on Computer Graphics and Vision “Graphicon” (19-21 September 2022, Ryazan)"
Authors:Charninå M.M.
Prediction and Visualization of Trending Research Topics in Social and Cognitive Robotics
Abstract:
This paper presents the results of a method designed to realize visualization of long-term prediction of trending research topics in the field of social and cognitive robotics. Meaningful topics were identified among the words included in the titles of scientific articles. The longevity of the citation trend growth was the target for the machine learning algorithm CatBoost. We conducted experiments on a dataset including 5 million scientific publications to demonstrate the effectiveness of the proposed model. The accuracy rate of 5-year forecasts for a number of experiments was about 60%. Trending topics are built from trending keywords located closely in the semantic vector space. The following trending topics in the field of social and cognitive robotics have been identified: recognition, deep learning, engagement, disorder, conversation, cognitive computing, attention, robotic platform. Trending keywords and topics are visualized on a semantic map built using the t- SNE method. Visualization helps to see the Big Picture, identify promising directions, understand trending topics and reveals related keywords.
Keywords:
Cognitive robotics, social robotics, visualization, long-term prediction, trending research topics, proven forecast accuracy, CatBoost, scientific bibliography, Big Data