A Topic-Based Patent Analysis for Mapping Technological Trends in the 5G Industry
by Nicolas Sacchetti
At the 4POINT0 Conference on Policies, Processes, and Practices for the Performance of Innovation Ecosystems (P4IE 2022), Davide Pulizzotto, along with his colleague Mehdi Raiem and in collaboration with Anas Ramdani and Catherine Beaudry, presented their research.
Pulizzotto and Raiem contend that 5G technology is set to assume a dominant role in the communications market. Beyond shaping communication’s future, it is expected to drive progress across various other sectors, potentially revolutionizing entire industries and giving rise to new markets such as autonomous vehicles, smart urban and industrial infrastructures, and interconnected medical devices.
Patents as Signals
The World Intellectual Property Organization (WIPO) states that patents cover nearly 90% to 95% of the world’s inventions. The researches highlight that since patents are public documents, data on applications or granted patents can be accessed swiftly. Consequently, the research team has employed patents as indicators of technological trends, developing pathways to monitor innovation continuously. Patents, therefore, act as a measure of a company’s R&D investment strategy.
Measuring R&D Strategy
The study aimed to resolve two principal questions:
- 1. What are the pivotal elements of 5G development, and how can these be discerned through a bottom-up approach solely using patent data?
- 2. Who are the primary entities driving the development of 5G technology?
Unstructured Textual Data
The focus of the study was on the unstructured textual data within patents, encompassing the abstract, invention description, and drawings description, among others. Natural Language Processing (NLP) and text mining techniques were employed to distill textual features from these patent documents.
« Pulizzotto et al. (2022) assert that textual descriptions and linguistic patterns in patent documents efficiently yield technical knowledge, providing more robust indicators of technology trends than bibliographic information. »
Pulizzotto and Raiem draw attention to the bipolar configuration of the topic network map. Among the top 20 entities in their dataset, there is a pronounced dichotomy between China and the USA. Chinese companies and universities cluster around certain topics in the top left, while US companies group in the bottom right.
Chinese entities show more interest in communication terminals, signal detection, and network security. Conversely, American entities focus on user equipment and computer programs, aligning more closely with 5G commercialization. Central topics that appear across companies signal the core of the competitive 5G landscape.
This paper provides a comprehensive analysis of 5G technological development, highlighting significant differences between Chinese and American entities.
Methodologically, its primary contribution lies in applying a semi-supervised machine learning approach for document filtering, as opposed to manual methods.
For managers, this research serves as a tool to gauge competitors’ technological positioning and investment strategies, which is critical for market success.
Furthermore, it offers governments and universities insights into 5G technological advancements, aiding the adjustment of policies and financial support for innovation.
The research team plans to utilize Deep Learning to develop a model specialized in identifying 5G technology within patents and scientific papers.
Ce contenu a été mis à jour le 2023-11-08 à 11 h 29 min.