In our previous post about patent data quality, “5 Consequences of Patent Data Errors: The Applicant Name Field“, we explored the impact of data inaccuracies in one of the most problematic fields for patent information users: the applicant name field. Common patent data errors or inaccuracies in this field can result in various negative outcomes from wrongful litigation to incorrect competitive benchmarking, to misguided strategic decisions – all of which end up costing organizations time and money.
While the errors in the above mentioned article were identified by patent authorities and may be widely known, we’re revealing 6 additional common patent data errors that our own research team have identified as sources of inaccurate data. These have been identified based on our long running partnership with our customers, industry partners and government organizations.
1. Corporate name changes
It is common for companies to change their names due to various reasons like restructuring, change of ownership, rebranding, etc. For example, Panasonic was previously named Matsushita Electric before they changed their name in 2008. To put this in perspective, more than 10,000 patent families would have been missing from Panasonic’s portfolio that were assigned to Matsushita Electric at the time of the name-change, had they not been reassigned properly in the database.
2. Ongoing mergers and acquisitions
After corporate mergers and acquisitions (M&A), the resulting combined entity becomes the commercial owner of the combined patent portfolio. Due to a variety of reasons, a merger or an acquisition does not always result in a direct aggregation of the original portfolios.
Our research team closely follows all announced M&A proceedings until the deal goes through and is finalized or is called off, to ensure these changes are updated to the portfolios and corporate trees in our database. The team tracked over 700 M&A activities that were announced between the years 2000 and 2015 alone, among 185 small and medium enterprises (SMEs) from the US.
3. Mergers and acquisitions that happened in the past
Our data researchers take special care during data harmonization to ensure that all historical M&A activities that the firms have gone through are also taken into consideration when assigning patent families to their current ultimate owners. In order to ensure reliability, all historical M&A activities need to be accounted for and extensive checks and corrections also need to be performed to make sure these transactions are reflected in the current portfolio of patents assigned to each company.
4. Reassignment of patents
Patents are only assigned correctly when historic and current owners are identifiable. When patent sales or trades take place, the status of the patents involved is updated with the respective authorities. This data sheds light on who the patents belonged to and who have they been reassigned to; and this needs to be carefully studied in order to ensure any reassignments are updated in the database.
Back in 2016, the mobile phone maker Xiaomi bought close to 1500 patents from Microsoft that covered technologies like communications, video and cloud. If this patent sale is not reflected in the database, Microsoft will still be shown as the owner of these patents. Imagine the kind of consequences if these common patent data errors are not corrected before used in an analysis or to make decisions.
5. Firms with identical or generic names
Common data patent errors, which affect accurate assignee information, are caused by companies having identical generic names and/or patent offices translating the original name into a different language. This could lead to wrongful attribution of a patent to a completely different portfolio than the one it was supposed to be assigned to.
Take a look at two universities – one in China and one in the U.S who share the exact same name: Northwestern University. When a database includes only machine translated information, it opens up the possibility for patents owned by either of these universities to be incorrectly assigned to the other one. There is no relationship between the two, combining patents owned by each respectively would be a massive oversight. Hence it is imperative that this data is checked for inconsistencies by an experienced human researcher. Chinese corporations often include their company’s location in their corporate names, which leads to additional company names that are similar.
6. Translation and transliteration mistakes at patent offices
Our research team comes across numerous or common patent data errors and mistakes that stem from patent offices haphazardly adapting or translating original company names to their local languages. In most cases these translations are mere phonetic equivalents of the original name in the local language. In such cases, the researcher leans on his/ her personal judgement and experience to figure out the actual name of the company in order to assign the patent to the correct owner.
Spotting trivial errors can be like searching for a needle in a haystack. To ensure state-of-the-art data quality, LexisNexis® has a highly-skilled team of experts focusing entirely and only on this task. The LexisNexis® PatentSight® Data Harmonization team members come from diverse backgrounds, with varying expertise in many areas of study, technological fields, and possess varied language skills. Our expert data researchers with keen eyes, years of experience and immense patience catch such mistakes that would otherwise be overlooked by a system that uses only AI or ML tools for this purpose.
Learn more about PatentSight®.
Excellent data quality is the foundation of reliable analyses. Learn how PatentSight enhances patent data here.
Access articles from our patent data quality series:
About the author: Carsten Guderian
Carsten is a Senior Project Leader and has a background in Economics and Business Administration, particularly innovation management and patent analytics. He has been affiliated with PatentSight since 2012.
