Intelligent Patent Searching? It’s All About Pulling the Right Strings
Published on 06 May, 2016
In a couple of previous posts, we’ve discussed the necessity of intelligence in IP searches as well as the importance of drawing on additional data sources in order to derive more meaningful insights.
In this post, we’ll talk about an important aspect of IP analytics — "search query" or "search string".
In layman’s terms, a search query is an intelligent combination of words/terms that are used to dig a database for references/results of interest. It’s not unlike what you’d do while looking something up on Google.
"Intelligent" is the key word here.
A proper search query ought to help IP researchers shave off precious time while fetching the appropriate results they’re looking for.
“Appropriate results” here means finding those two or three crucial patent references from among the millions of worldwide patents they’re sifting through.
As you may imagine, it’s as hairy as finding a needle in a haystack.
Your best bet is to weave a brilliant search query. It can be something as handy in an age of multi-million dollar infringement lawsuits as a silver bullet on a full moon night.
Making proper search query however, is easier said than done.
Some say it’s a difficult art to master; others think it cannot be mastered at all.
The usual drill for an IP researcher is to use combinations of various keywords.
You’d have to prepare and run your first query, scour the results, figure out whether you’re in the ballpark, and modify your query to get the results you’re looking for.
Then you run your second query, throw in some more keywords, and then start over again till you’ve got something better.
Lather, rinse, repeat …
The process is as cumbersome as it is fallible.
Hey, wait a minute.
Aren’t we living in an age of AI and Big Data Analytics?
Think they’d help?
Imagine an intelligent query forming tool.
Something embedded in the patent databases you’re searching.
You’d have to enter the first round of keywords, and the tool “intelligently” combines and rearranges your terms to provide one or more queries.
The tool could have more features too.
Instead of entering keywords, what if you could just scan a patent document that you’re looking into — perhaps looking for prior art — and the tool figures out the relevant keywords all by itself.
What if it also figured out and assigned a weightage to your keywords, a priority order to determine what the most important keywords are, and which are to be located first.
How would such a tool even function?
For starters, it could use a simple word database to look for synonyms and semantic variations of your keywords. It can then analyze the citations of your subject patent and figure out how the keywords have been used to create logic, perhaps even with the help of some crafty AI. The tool can then use the observed logic to combine keywords and create a relevant query. Finally, the tool can conduct multiple iterations to perfect the query.
The process could involve manual intervention, or it could be completely automated.
The tool could also be heuristic, capable of creating and validating intelligent queries without any human oversight at all.
As Big Data and AI technologies mature, intelligent and self-learning tools that assist you in your day-to-day research could be as commonplace as cellphones.
As far as I’m concerned, the future can’t get here fast enough.