A Novel Method for Predicting Technology Trends Based on Processing Multiple Data Sources

Main Article Content

Thanh Viet Nguyen
Alla Kravets


In order to gain competing capability in conditions of quickly scientific changes, it is crucial to track the evolution of existing technologies and to explore promising and emerging technologies. Moreover, numerous previous studies showed that sudden changes in R&D and patents are actually correlated with great variations in the market profit of firms. For this reason, if stock prices of an enterprise keep uptrend, then the technologies developed by considering one will be likely to become promising innovations in future. In this paper, we proposed a method to predict technology trends based on processing multiple data sources by mining Web news, forecasting stock price trends of high-tech companies, and patent clustering analysis. Different from other studies, our proposed method promotes an idea of predicting technology trends by forecasting stock price trend using univariate and multivariate data preparation approaches, with the utilization of Bayesian optimization for exploring best hyperparameters of machine/deep learning models, also a new method for patent analysis. Besides, a program system was created for analyzing word burst detection, predicting the time series of stock prices, and analyzing patent documents. After collecting patents of Samsung Electronics Co Ltd, as a case study, clustering analysis is implemented on extracted noun phrases to explore technology trends developed by the company. These technology trends have recently been confirmed by domain experts in their corresponding published articles. The obtained forecast precision is about 93.8%, which proves that the proposed method gains positive reliability.


Download data is not yet available.

Article Details

How to Cite
Nguyen, T. V., & Kravets, A. (2023). A Novel Method for Predicting Technology Trends Based on Processing Multiple Data Sources. Advances in Systems Science and Applications, 23(1), 69-90. https://doi.org/10.25728/assa.2023.23.01.1251