Abstract:
Technology investment decisions are vital for a country or an organization to set project
priorities to uplift sustainability. Predicting of the future direction of technological
advances can help investment and technology choice decisions. At present, there is a
high dependency on expert opinions in technology trend predictions. As such, a datadriven
approach is needed within the technological trend prediction process to support
technology investment decisions. Many studies have been conducted on technology trend
prediction using different methods and techniques. Patent analysis has been identified as
an effective trend analysis method as a patent is rich with technology-based information.
This research aims to investigate blockchain technology as a case study to explore
its trends by applying text mining and machine learning techniques. Blockchain is
potentially a key technology in a new technological paradigm of increasing automation
and integrating physical and virtual worlds. This research focuses on two unsupervised
approaches: clustering and topic modeling. Initially, patent data are divided into several
timeframes according to the publication time. In the clustering approach, keywords
are extracted from patents by text mining, and the keywords with similar semantics
are grouped together to form clusters. The Latent Dirichlet Allocation (LDA) model
extracts topics from the patent data in the topic modeling approach. Then, the process
of identifying emerging technology areas is satisfied by the clustering and topic modeling
results using data visualization techniques like scatter plots and distribution plots. The
results from both approaches were usually complimentary. The study found that Distributed
Ledger Technology for Cryptocurrency Transactions and Smart Contracts are
the emerging technologies in the blockchain domain. It is depicted that the cryptography
technology area has evolved to ensure the data security of transactions.
