A new research project at FIZ Karlsruhe – Leibniz Institute for Information Infrastructure allows a faster and easier access to innovation trends in future technologies through trend mining in patent information. FIZ Karlsruhe has developed efficient analysis tools for trend research with patent information which draw 'landscapes' of the international research.
According to Sabine Brünger-Weilandt, President and CEO of FIZ Karlsruhe, the new research project 'Trend Mining for the Sciences' seeks to strengthen the know-how within the area of data- and especially textual analysis on a consequential level, for supporting science and economy even better when it comes to company-critical business decisions.
It has become more and more important for industry and science to ascertain new trends and developments in research as early as possible in order to stay ahead in an increasingly tough international competition. The new research project Trend Mining for the Sciences is funded by the Leibniz Association on basis of the joint initiative for research and innovation in the course of the line of funding for especially innovative projects. It will be financed for three years. The methods of text mining are to be applied to the challenging text type of 'patent texts' for the first time.
Patent information reflects the innovation capabilities and productive efficiency of companies, regions, industry sectors and entire national economies. The results of technical and natural scientific research are often exclusively published in the form of a patent. Experts estimate that 70 – 90 percent of all the technical knowledge ever published is documented exclusively in patent publications. Hence, patent publications are an extremely valuable source of information for researchers in science and business.
Within the recently started Leibniz-Project the involved researchers of FIZ Karlsruhe and the University of Hildesheim aim to develop new software tools for being able to conclude innovation trends from the enriched and edited patent documents by analysing full texts, patent classifications, bibliographic and other information. The difficulty mainly comes from the patent terminology and the linguistic structures which differ from the everyday speech. The goal of the research project is to develop prototype software which can be used easily by researchers, information professionals and patent experts for identifying innovation trends.
This software shall be able to make trends visible based on an analysis of different text features in the patent publications by combining the results with further information such as scientific publications, with the aim to show innovation trends more simply, faster and comprehensively than before. The high reliability of the search results and the information lead are crucial advantages. The use of this new software will make competition relevant and business critical decisions more certain.
While the term Data Mining has become more known in public recently the experts of FIZ Karlsruhe have worked to develop special tools to extract information from a huge amount of data already for almost 30 years. Trend analysis from patent information is already possible. However, the method available is very time consuming and only meaningful if a trend can be numerically identified on the basis of increasing patent publications. Moreover, only by a careful processing of patent information the information contained in the data about technology and markets becomes specifically and appropriately analysable. Content, system and tools have to be perfectly synchronised. The online service STN® International, operated by FIZ Karlsruhe, plays a central role in this context. The online service provides a unique combination of premium databases. The content of STN International currently includes about 150 databases and 1.5 billion documents. It covers patent specifications and metadata, like abstracts of scientific publications, facts like chemical structures and reactions as well as biosequence data.
For the analysis of the retrieved patent information special visualisation tools like STN AnaVist are needed. STN AnaVist allows a large number of hits (out of thousands) to be visually displayed as "patent landscapes". These types of landscapes are formed by merging documents with similar content. With the help of numerous analysis features, these landscapes can then be interactively evaluated.
