Classification Searching with Keywords
The best searches will use both classification searches AND keyword searches together. Let's examine how you might approach a patent search.
Let's say we want to find patents related to location-based monitoring and messaging. The invention in question uses the cellular network or other networks (e.g., GPS), to find your location and compare it to a map, as well as a set of your personal preferences located on some server. The method then notifies you of sales and promotions within a 5 minute walk or drive from your current location. The invention has to know if you are walking or driving based on the average speed you are moving. A five minute drive could translate into an hour walk, and who wants to walk an hour to save 25 cents on a Starbucks coffee (OK, you're never more than a 5 minute walk from a Starbucks. Bad example, but you get the point.) So that is the basic invention.
First, I am going to pick a few terms that I think could possibly identify patents similar to what I am searching for. I don't have to be perfect because you can quickly iterate if you miss. I'll try the full text for this:
FT:("mobile device" AND (location NEAR3 monitoring) AND drive AND walk AND promotion)
Notice I used a proximity search for the terms (location NEAR3 monitoring). This just says that "location" and "monitoring" need to be within three terms of each other in any order (ADJ in place of NEAR will specify the order the terms have to be in). I just intuited that there would be many variants and modifiers of this idea. All terms are required in this example (since I used the AND operator each time).
AcclaimIP returned 349 results without any other filters. I was hoping for a few more, but this might work. Scanning the class titles and hierarchy I found H04W4/02 which looks pretty close, and G06Q30/0629 which seems like it's also in the same wheelhouse as my invention. I do note that in the figure above, the query window is obfuscated, but I ran the query above.
The likelihood that my technology will perfectly fit my guessed keywords is low (remember the vagaries of human language). So I am going to do two things...
- Remove the keyword restrictions (1) --> I deleted the keywords from the query window.
- View all patents in the CPC class from the facets (2) --> I selected the facets for H04W4/02 and G06Q30/0261
Once I've run the search, you can see I actually expand my results to 24,020 matching documents. This is because by removing the keyword restrictions, I am now looking for ALL the patents in the two CPC classes I selected.
I next scan a few patents and develop a new set of keywords to limit the results within the class. I tried...
sale OR promotion AND (walk AND drive)
...and added these terms back to the query window.
I think these results are ok. Scanning the titles I can see I am right on the sweet spot.
I chose to just click the + on the first patent to just see it, but I can also expand them all, by clicking on the + on the top bar. Also, right now I have chosen the Bibliographic view of the expander row, by using the View-->Expander Row Content-->Bibliographic, which is helpful at this point. If you prefer, you can choose the Images view with or without abstracts.
For a more detailed view, you can open the Document Details window by double clicking any patent in the Search Result grid. This will give you the full details.
Click Forward (1) in the Document Details window and AcclaimIP will advance to the next patent on its parent Search Result window. The highlighting on the patent in focus in the Search Results grid (2) will also increment so you know where you are.
This is an easy and almost foolproof process to find what you're looking for, and it will become second nature as you get more familiar with AcclaimIP. Of course, there are subtleties to the process. There is no guarantee that there are NOT other classes that might also contain relevant patents. I find these other classes by repeating the process from the top with new keywords that I learned from the first round.
Some of the other relevant classes will almost certainly be cross reference classes (the ones that are not listed first and not bold type, as shown in the circle in the figure above). Your first line of investigation will tell you where to go next.