Patent databases continue to proliferate on the internet. The most recent addition is Patents.com, which describes itself as “one of the most comprehensive free patent search services on the web.” I get a little annoyed when I see claims such as this, especially when the database provider doesn’t state the contents and dates of coverage. Having a great search engine doesn’t mean much if the underlying data is incomplete. So I figured that it was prime time to update my comparison of patent database search engines.
When I did this last in February 2007 I compared Delpion, FreePatentsOnline, Google Patents, Patent Lens, Patent Monkey and the USPTO website. This time I compared five patent databases: Google Patents, FreePatentsOnline, Patents.com, Patents Lens and the USPTO patent database. Patent Monkey was absorbed by Patents.com in 2007. I decided to drop Delphion because it offers access only to US issued patents. Google Patents has added data for US published applications but it’s unclear how frequently the data is updated. The other three non-USPTO databases are updated weekly.
All test searches were restricted to US patents issued from 1976 forward in order to allow for comparisons between databases. (Some databases include non-US patents and/or patents prior to 1976.) My search categories were inventor name, assignee name, keyword in title, and current US patent classification.
In the inventor search category, most of the databases came very close or slightly exceeded the results from the USPTO database. The only exception was Google Patents, which failed miserably at finding patents by inventor name. Perhaps Google has disabled their inventor name search.
Company name search results were also consistent across most of the databases. Again, the exception was Google Patents which found significantly fewer patents for two of the four companies searched. Patent Lens performed well except that it found approximately 200 fewer patents assigned to Bombardier than the other databases (except for Google). This might be explained by the fact that Bombardier’s patent portfolio includes more than 200 design patents which are not covered in Patent Lens. FreePatentsOnline did poorly with “Queen’s University”, possibly because the apostrophe threw off the search engine.
The title keyword search produced varied results. Patent Lens, Patents.com and FPO did very well in three of the four searches. Google Patents found significantly fewer patents in all four searches. In a couple of searches FPO actually retrieved more patents that the USPTO database. But this might be because FPO searches withdrawn patents, which are not included in the USPTO database.
The best scoring database in the current US patent classification search was Patents.com, which produced identical or very close results to the USPTO database. FreePatentsOnline also did well, except when asked to retrieve patents in 623/1.1. The decimal subclass might have confused FPO’s search engine. Google Patents and Patent Lens did poorly. Patent Lens does not index USPC or IPC codes, so the only way to search them is to do a crude keyword search against the front page. The results are next to useless.
Table 4. USPC Search Comparison
Overall, the USPTO database search engine produced the most reliable search results. Patents.com performs well for USPC and inventor name searches. Patent Lens is good for inventor, assignee and keyword searches but not USPC searches. FPO is almost as good as the USPTO database but punctuation may throw off search results. I would not recommend Google Patents for anything except retrieving PDF copies of known patents. Its search results are simply too unpredictable and incomplete.
The USPTO does score poorly in the patent document image category because its images are stored in TIFF format rather than PDF. Users must install a TIFF viewer in their browser in order to view patent documents from the USPTO website and then the documents must be viewed or printed one page at a time. The other four databases in this review support mutli-page PDF downloads.