Patent database searching is an important aspect of any technical innovation and development today. It is estimated that over a million patents are being published annually, of which more than 90% are not available in any other form of publication1. The information contained within the patent documents is highly technical in nature and that is also of great relevance business and commerce.

Patent searching is used in a wide variety of situations. It ranges from legal related searches such as invalidity or clearance searches to competitive intelligence searches such as competitor analysis, to even technical searches such as a technology landscape. The ones requiring patent searches range accordingly from in-house intellectual property (IP) and legal counsels, to technology managers and business leaders. As can be understood readily, the search results form the basis of several critical decisions.

Patent searching is an art that requires both technical knowledge as well as ability to read and understand patent language in line with the laws in place. It is now well accepted that it is a specialized skill, and a professional from the respective technical field needs to be trained and groomed especially for this task2.

Patent searching is typically conducted in a suitable database. A typical search situation involves obtaining a search request from a particular source (or may be generated from self-interest usually by experienced searchers and/or technologists). The search request may comprise some details on the aspects, which may be technical or a non-technical description of a product or process. From the search request, a set of concepts are identified, which then form the basis for coming up with synonyms for each concepts. The combination of concepts and synonyms is the keywords, which is the basic building block for putting together search strings.

It must be noted here that sometimes search requests may also include searching for particular assignees or inventors. In this case also, there is requirement for putting together the keywords to build a search string. This includes adequately capturing the spelling variations of the names provided as well as the various name changes and acquisition related or assignment changes to encompass all the related patents.

There are several databases in which patent searching may be done. Many databases such as USPTO, Google Patents, Free Patents Online, esp@cenet etc. are freely available, while other databases such as ThomsonInnovation, Orbit, Patbase etc. are available on the basis of a subscription. Each of these databases varies in terms of at least one of the following: Data Coverage, Search Engine, Interface.

Some of the databases, especially the paid ones, attempt to make life easier for a patent searcher by building a query from a given text by automatically identifying the concepts, building a keyword list, identifying synonyms, generating a search string and running the search3.

Patent searches can also be achieved through the use of classification codes. Briefly, all patent documents are assigned one or more classification codes by the examining authority based on the technical features of the invention. Thus, a search for a particular aspect can be achieved by looking at all the patents under a particular classification code. However, this suffers from the drawback of those inventive features that may be used in many different situations, thus requiring the identification of several classification codes and searching each of them for relevant patents.

Another useful technique employed by patent searchers includes citation searches. In this type of search, all the patents cited in a particular patent (backward citations) as well as all those patents citing the identified patent (forward citations) are used to further identify more relevant patents. This method has been found to be useful in speeding up the process. Advanced citations analysis may also be used to focus in on the relevant patents faster4 [Amberscope].

In a typical situation, a professional searcher uses a combination of all three methods in conducting a search. However, it has been observed that while classification code and citation search are useful, neither of them can be depended on exclusively for a search, while a keyword based search alone can be used with a certain degree of confidence [Personal experience, as well as discussion with other professional searchers]. Thus, a keyword based search is an essential weapon in a professional searcher’s arsenal.

The keywords based search string is generally run on at least two databases to ensure a comprehensive search is conducted. This is necessary because of the way different databases organize the data and display the results. For example, USPTO has all the US patents and patent publications in its database while Orbit organizes the data into FamPat and PlusPat, wherein the family members are arranged differently [Questel manual ref]. Similarly, the same search on Orbit and Patbase will deliver different results set, the extent of difference depends on the nature of search, the number of foreign patents in the family members of the results hit, and a number of other factors. This variation in the results set is made use of by a professional searcher to arrive as close to a comprehensive search as is possible within a given time frame.

Indeed it has been noted by several experts that searching in multiple databases is necessary to ensure comprehensiveness of the search results, or at least to ensure that the major relevant ones are not missed during a search5.

1 See for example article by Anthony Trippe, http://www.patinformatics.com/revisiting-an-old-standard-80-of-technical-information-is-found-only-in-patents/ published on 21 April 2014 2 The Importance of a Professional Patent Search by Daniel Meyers published on 12 February 2015 3 esp@cenet and Google patents also allow natural language based text searching, but esp@cenet poses a limit on the number of terms used 4 www.amberscope.com 5 For example, in page 130 chapter Patent Searching ain’t what it used to be, by Stuart Kaback notes “Throughout the years, I have seen how often an effective patent search required the use of multiple databases.” in Chemical Information: Information in Chemistry, Pharmacology and Patents. Ed. Harry Collier 1989]. In page 32 in chapter Introduction to Patent Searching by D. Alberts et al. note that “Whenever possible multiple databases and systems should be used since each system provides unique features, different coverage and different indexing policies.” in Current Challenges in Patent Information Retrieval ed. By Mihai Lupu et al. 2011]. From http://landon-ip.com/PatentSearches.aspx “…we strategically use multiple search engines, databases…”