PatentBrain.2
From Intellogist
| If you found this page through a web search, we invite you to visit our Main Page to see what Intellogist is all about. This Community Report exists for our users to share information and experiences about patent search resources. Registered users can add, edit, or delete material on this page. The information contained on this page is the result of community collaboration and is not vetted by individual experts or fact-checkers. For in-depth reviews of other patent search resources, please read our Intellogist Reports. For information about editing Community reports, please see our Help pages. If you'd like more information about why Intellogist did not write about this search tool, please see our Criteria for Intellogist Articles. |
PatentBrain.2 is a Hybrid of the Machine and Human Intelligence, a patent search and analysis engine combining the superior analytical capabilities of human mind and the processing capacity of a computer. The machine algorithm extracts and processes patent classification numbers that humans (patent examiners) assign to each patent.
Why use class/subclass as the organizing factor?
When the patent examiner analyzes the patent application he determines to which classes/subclasses the patent belongs. PatentBrain.2 uses patent classification numbers as "concepts". This "human involvement" at the early classification stage eliminates many typical semantic errors in pure language-based search engines.
This classification is used by the "Find Similar" algorithm to find similarities and logical connections between patents. It allows the searcher to quickly focus on the most relevant patents. More work is done now to explore other definitions of "similar" and incorporate them into the algorithm.
The "Find similar" algorithm is the main reason for using the PatentBrain search engine. Once you find the patent or patents of your interest select them by checking the square near the patent number, then click on the "Find similar" button. It'll instantly re-arrange the patents according to the similarity to the patent that you have selected.
What are the columns on the right side of search results?
Each column corresponds to one USPTO or IPC class/subclass. The number in each row-column, patent-subclass intersection = (the number of subclasses in a patent) + (the number of patents in the subclass in the search results table). This is to place a "value" on the particular intersection. It is presumed that the subclasses with most patents in them, and patents with most subclasses are more "valuable".
What are the colors and the frames around the cell? The value-dependent cell colors are to give the table a map-like appearance, as in the patent landscape. If the intersection has a outline frame, then it means that the patent in that row belongs to the subclass in that column.
Patent Landscaping, Patentability, and the Freedom To Operate
The patent landscape that you see on the right side of the results table reveals the patenting activities in a given field. The high value numbers on the map indicate more developed areas, the blanks and the low values show less developed areas of technology. More information can be found here.
Coverage: This engine covers full content of USPTO and esp@cenet online databases. Your query goes there directly. The engine processes the results according to the PatentBrain.2 algorithm.
Why two search pages, doesn’t the world patents page cover also US patents?
The US patents page allows longer search strings and more sophisticated queries.
Search Pages:
Be patient, the query and analysis may take up to five minutes!
Here are step by step instructions on how to most efficiently use our algorithmic patent search and analysis engine:
1. First run a query using your keywords. The US and the World patents search pages each have different syntax rules. (See "Help")
2. Then get a quick and broad view: scan your mouse over patent numbers and you'll instantly see the pop-up of title and the abstract of each.
3. Dig deeper: click your mouse on the patent of your interest and get the full text.
4. Then select the patent of your interest (check mark it), and hit the "Find Similar" button. This brings to the top the most similar patents to the one you are currently focused on.
5. The process is iterative. If you find another similar patent, you can then select both relevant patents and run "Find Similar" again. Thus a gradient of similarity is established in the search results table.
6. Remove unwanted patents: Select patents at the bottom of the hit list. Click "remove patents".
7. Click "Save Query". This saves it in your local computer. You can reload it the next time and use the "Find Similar" again.
8. Different saved queries can be merged by loading them, then saving the resulting table.
You have now created a conceptually interlinked database of patents in your field that is
- viewable,
- sortable,
- possible to maintain and keep updated.
Recent News, September 2009
PatentBrain.2 may offer class definitions, and permit saved file conversion for local use in up-coming versions. Intellogist member Lipa Roitman consistently returns email so users should register and try this web tool for creating patent landscapes and finding similar patents.
