A method to detect indicators of practical application of technology

An algorithm that uses multiple types of literature data to determine the indicators for practical application in numeric values

In collaboration with the RIKEN Interdisciplinary Theoretical and Mathematical Sciences Program (iTHEMS), Ricoh has developed an algorithm that detects the indicators of practical application from multiple types of technical literature data.

*

This technology is patent pending.

Background

In recent years, there has been a surge in the development of research methods that utilize data to comprehensively and quantitatively grasp global technological trends. Ricoh, in collaboration with RIKEN, has developed a unique algorithm that visualizes the junctures of technological trends from literature data. This proprietary algorithm can identify keywords that have surged in patents and papers, detecting indicators of emerging technologies. However, it has been challenging to quantitatively determine whether these indicators point to early-stage research or are approaching practical application.

A visualization method to detect indicators of emerging technologies

Solutions

By applying our previously developed algorithm, we have developed a new algorithm for multivariate analysis of keywords commonly appearing in patents and papers. This advancement allows us to numerically determine the presence of practical application indicators.

Technical highlights

This algorithm detects technologies that are transitioning to the commercialization phase after completing a particular stage of academic research at universities and research institutes. It identifies these technologies as having “indicators of practical application.” These indicators are identified in three steps: calculation of the coordinates of keywords commonly appearing in patents and papers, pattern classification using the position of coordinates, and calculation of the length of the link connecting the patent and paper coordinates.

1. Calculation of the coordinates of keywords commonly appearing in patents and papers

For each keyword that commonly appears in patents and papers, the previously developed algorithm calculates each keyword's coordinates (the pattern of increase or decrease in the chronological data). Specifically, for each keyword, it creates a line graph of cumulative frequency in each year of occurrence, and plots the two sets of coordinates obtained by converting the area of the graph and the area of the squared function on planes S and A. While the previous algorithm used a single set of coordinates for either patents or papers, this new algorithm plots two points for each keyword—one for patents and one for papers—connecting them with a line segment (link) to represent the keyword.

Plotting the coordinates for each keyword commonly appearing in patents and papers

2. Pattern classification using the position of coordinates

We capture indicators of practical application by detecting patterns where the number of academic papers, the primary output of academia, has already converged, while patents, the primary output of commercialization, have rapidly increased in recent years.

A model for capturing the indicators of practical application

When plotting the patterns of common appearance of keywords in patents and papers on a diagram, keywords that have increased rapidly in recent years are placed at the left end of the diagram, while keywords that have converged are placed at the top or right end. Indicators of practical application can be captured by extracting keywords whose appearance pattern in patents is placed at the left end and whose appearance pattern in papers is placed at the top or right end.

Keywords that have increased rapidly in recent years are placed at the left end of the diagram, and keywords that have converged are placed at the top or right end

3. Calculation of the length of the link connecting the patent and paper coordinates

To easily identify the patterns in which the patent coordinates are at the left end and the paper coordinates are at the top or right end as shown in section 2 above, the length of the link is utilized.

By calculating the link length for each keyword and examining them in order from the longest to shortest, keywords that show indicators of practical application can be efficiently extracted.

Identification of patterns based on the link length between two sets of coordinates

*

This technology was developed in collaboration with Hideaki Aoyama, Visiting Principal Researcher, and Wataru Soma, Visiting Researcher of the Interdisciplinary Theoretical and Mathematical Sciences Program (iTHEMS) at RIKEN.

Ricoh's vision

By conducting data-driven research using this technology, Ricoh will bring comprehensiveness and speed to its search for new topics in its own research and development. We are using this approach and it will continue to contribute to the identification of emerging research themes and the creation of high-value intellectual property.