For many years, Patent Searchers retrieved patent information using classical hosts (STN, Questel Dialog and ORBIT in the old days). A specific command language was essential for performing searches and good understanding of the command language was a basic requirement. This was very useful for retrieving accurate legal information and finding relevant information for patentability, patent clearance as well as material for oppositions and due diligence. This information was mainly intended to be used by Patent Attorneys, to give a legal opinion on certain projects and/or products.
Over the last years, a new group of clients coming from the “business” has become more and more interested in patent information (as well as trends in patenting). And more specific requests for patent information not only for legal opinions, but also for more business related decisions became a fact. These “new clients” also want to receive a different kind of information out of patent data.
To retrieve this information, new tools are needed to create more landscape like output. On the other hand, the possibility to handle bigger sets of data and perform more statistical analysis is a requirement. New initiatives have been undertaken to satisfy the needs of the searchers and several tools have been developed, like tools for semantic searching.
One promising tool is created by Treparel, which came up with the product called KMX, to handle larger sets of patent data. The basic principle of feeding the system with some relevant documents (which are particular relevant to the topic) makes it possible to retrieve “similar” documents almost immediately and with a high precision.
First trials with the software look very promising, with a relevancy of over 90% for certain topics. Although we are not yet there, we think KMX is a product with future potency and is worth to be further developed and improved. Currently I’m involved in further improving this tool in order to make it useful for handling big sets of patent full-text data.
One of the remaining questions will be whether or not this tool can replace the classical way of searching or need to be seen as an add-on. The “experts” have the last word, of course. At least a good integration with full-text databases is a necessity. Nevertheless, it is clear there is a high need for tools that can handle big datasets, in order to deliver reliable information out of patents.
In a next news item I will further inform you.