Methods for Assignee Normalization

Normalization is the process of efficiently organizing data in to eliminate redundancies.  In patent analysis, Assignee Normalization is a process of preparing clean and accurate Assignee names from the raw data that exists on the published record.  When assignees are not present or organized properly, the accuracy of the analysis is greatly hampered and the results cannot be relied upon. 
Let’s summarize the different challenges users face with respect to Assignees when working with patent data. These are: 
  • Unknown Assignees where there isn’t a Assignee name in the publication
  • Unclean Assignee Names (Misspelt or subsidiaries with different company endings)
  • Mergers and Acquisitions of companies
  • Multiple Assignees (or Inventor names appearing in the Assignee field)
The methods used to tackle each of the above challenges are different and there isn’t one size fits all approach possible. In this blog, we will take a look some of these methods.
Unknown Assignees
There are different methods used to locate probable assignees for records that don’t have one. This aspect is seen in US patent applications that do not have an assignee name until right before grant (mostly because companies don’t want to reveal their identities until the last moment). A user can use the following methods for locating probable Assignee for such records:
  • Locate assignee from INPADOC family information – The INPADOC database provides information of corresponding patent applications in different countries and these may include the assignee name.
  • Locate assignee using Inventor matching – The same inventors may have appeared in patents that have an assignee name. Further if the attorney too is same and/or the filing is around the same time, then one can assume with confidence that the Assignee for the unknown record will be the same.
  • Locate assignee from US Assignments Database (Only for US records) –  US applications may have already had an assignment event at the PTO which would be available from the Assignments database and this would make it easy to lookup the Assignee.
Finally if none of the above work, the user can either manually provide an Assignee name to the record or instruct the software to use the inventor name as Assignee name. Patent iNSIGHT Pro has automated tools for each of the above method that leverage the above logic used and provide Assignee suggestions to the user.
Unclean Assignee Names
Auto cleanup is used to combine large group of assignee and to create small groups which can be used for further analysis. This process is faster and mostly precise as compared to manual process. This activity can be performed by fuzzy matching, thesaurus matching and regular expression based pattern matching. 
Unclean assignee names may be in the form of misspellings of assignee names or subtle differences in naming of a company, occurrences of duplicate entries, no unique assignee records.
Fuzzy algorithm based merging of names might not be effective in cases where names are very short of when names phonetically sound similar (Short Chinese names). For this, we can use manual cleanup, wherein you can create groups manually. In Patent iNSIGHT Pro, a list of all unique assignees present in the patent data set is provided and then the user can choose any one of the above method to merge name. As seen in the figure below, it is evident that the selected set of assignees is the same organization.
Mergers and Acquisitions
In case of US records, for companies to be able to take action on any patent, they must report their ownership to the USPTO as per section 37 CFR 3.73. This information is publicly available in the US Assignment database and is highly valuable resource to track ownership. So if a user wants to find updated ownership information such as change of name, mergers, execution date, assignments database helps in tracking. Patent iNSIGHT Pro integrates with the latest Assignment data and gives current owner suggestions for all US patents and applications.
Multiple Assignees
Instead of having multiple assignee names, for simpler and precise analysis, a single assignee per document is useful. In many cases especially for WO applications, inventor names that as present as co-applicants may show up in the Assignee field and so it’s important to be able to remove these. Patent iNSIGHT Pro has automated tools to identify inventor names in the Assignee field and remove them. The alternative for the user is to manually go through the names which take a lot of time for large datasets.