Advanced Keyword Searching

When searching for patents, it is important to pick the best keywords that will select for the patents you want to find.  The terms found in these patents may not be obvious, so you'll have to go through an iterative process to identify key terms found in patents you want to isolate.  For these reasons, AcclaimIP includes several powerful tools to help you analyze patent keywords.  These tools are discussed later in this chapter.

What do patents read on?

You'll find that the vast majority of patents don't read on devices or products such as endoscopes, mobile phones, laser printers, or light bulbs, per se.  In general, patents also don't even read on components of devices such as touch screens, microprocessors, cameras, and fusers.  Instead, they tend to read on features of components of devices (for example a mobile phone is a device, a screen would be a component, and maybe a new scratch resistance feature of the glass that makes up the screen would be the feature that the patent was written about).  This distinction is important because you will likely get better, more targeted results if you focus on a single feature at a time.

A good search strategy will focus keywords at the feature level, or possibly the component level, and not on the device level.  AcclaimIP provides a multi-tiered matrix query feature which will help you roll-up your feature and component level queries to the device, or even product class, level.


Steps for Identifying Relevant Keywords, Strings, and Concepts

The steps for identifying keywords that will help find the patents you are looking for include:

1.  Understand what you are trying to find.  

If you are searching for mobile phone patents, you'll find hundreds of thousands of them.  The better approach is to find patents that relate to a specific component of the device such as the touch screen, battery, or antenna.  Certainly the terms you might use to find battery patents are going to be different than the terms you will use to find touch screen patents.  As mentioned above, you may want to search for specific features of an antenna, such as loop antennas.

2.  Survey patents for terms actually found in patents.

You don't need a committee of experts or a thesaurus to identify terms that will help find the patents you are searching for.  The patents themselves will provide the clues and terms you need to find related patents.

AcclaimIP offers three different tools, discussed in the next three chapters, for surveying keywords and concepts found in patents and creating queries from the terms you find.  Each tool uses a unique approach to help you in your search.  The tools are:

  • Keyword Tool:  Uses a simple term counter to identify terms and strings that occur in high frequency in a patent.
  • QueryFlow Tool: Uses a more sophisticated approach and identifies "important" terms in a single patent or a group of patents.  Important terms occur at a relatively high frequency in a patent, but a relatively low frequency in the corpus at large.  That is, they are generally more rare in patents in general (for instance the word "the" would occur all the time, but a specific technical term might only be in a handful of patents and so would be counted as a more important term).
  • Clustering Tool: Clustering is a technique to identify key terms and concepts from a large set of up to 1,000 patents at a time.

 3.  Run your queries using each of the tools.  I generally use the Boolean "OR" operator and combine the entire list of terms I discovered during my keyword survey step.

One of the most underutilized concepts in patent searching is the use of the Boolean “OR” operator.  Most people understand the basic idea behind the Boolean OR.  OR expands your results.  “Salamanders OR Newts” finds patents with either term, and is broader than searching for only “Salamanders” or “Newts,” or requiring both terms be in the document by using the Boolean AND.  “Salamanders AND Newts” returns the fewest results.

Due to its expanding qualities, many patent searchers shy away from using too many ORs in their search queries, but I think it is the BEST way to do keyword searching.  For example, if one patent professional uses "vehicle" for car, and one uses "automobile" for car, using the Boolean OR in searches like this is going to give you the largest number of relevant documents.  You can always then pare down the results using filters and facets.

The next three chapters we cover using each of these advanced tools to both identify keywords, strings, and concepts, and to generate queries from the terms you identify.


Add your comment

E-Mail me when someone replies to this comment