[tweetmeme source=”Intellogist” only_single=false]
The Intellogist team often gets questions about how to perform comprehensive searches on all patents owned by a particular company. We often refer these folks to our wiki Best Practices article on Assignee Searching, because we think it’s a good overview of the challenges inherent in this type of searching, and the strategies that can be used to overcome them. But one particular point we’d like to illustrate today is how both the corporate tree and browse index features in your patent search system can help you get better recall on your assignee searches.
One major obstacle to searching corporate names is the parent-subsidiary relationship. You may be interested in all patent documents owned by a certain company; in that case, you will first have to identify the names of any subsidiaries owned by the company. Several patent search systems on the market today attempt to help their users investigate this problem by loading corporate tree databases into their products as a type of reference tool; these databases simply show parent-subsidiary relationships. Examples include the corporate lookup functions in Thomson Innovation, TotalPatent, and the new Questel product orbit.com. It’s worth noting that each of these vendors has loaded a different corporate database into their tool; for example, Thomson Reuters uses a database produced by 1790 Analytics, while TotalPatent has loaded a LexisNexis database called Corporate Affiliations. It’s possible that two different corporate lookup products could give users different results, depending on how these databases are produced and when they are updated. It’s also worth noting that it’s unclear how much international coverage these databases offer, as detailed documentation on how they are produced is not available.
Another major obstacle to searching on names of any type is that they can be written multiple ways. Name variants can be introduced into the patent corpus through spelling or typographical errors, abbreviations, even problems introduced when transliterating names from one alphabet to another (e.g. from Chinese characters to Latin alphabet characters). Some patent search products offer a “browse index” feature that will allow you to see erroneous or unusual entries that may exist in the patent data. This feature was common in the older command line search products such as Dialog Classic and STN, but it is also available in newer web-based products such as PatBase, QPAT, and orbit.com. One downside is that these tools show you related listings alphabetically, so if you suspect that a variation may be introduced on the first few letters of a company name (e.g. in Chinese transliteration, Tsinghua vs. Qinghua), you may want to try some alternate initial spellings when using the browse tool.
Both of these challenges relate to the information retrieval goal of doing a complete search within patent databases – but what if the patent databases don’t actually contain the current assignee information due to a re-assignment? For more useful information about assignee searching (including a discussion of re-assignment challenges, and information on databases that predict patent assignees), please check out our Best Practices wiki article on Assignment Searching. If you have more tips for assignment searching, please consider registering on Intellogist (it’s free!) and editing this wiki article to add your contributions!
This post was contributed by Intellogist team member Kristin Whitman.