Taking Sherlock for a Walk
Sherlock is our revolutionary, AI-powered digital document bloodhound. It is a smart, machine-learning algorithm designed to make it easier to find information in large document populations.
Sherlock eliminates the need to craft complicated keyword searches. Instead, send one or more documents to Sherlock and let him find more. In milliseconds, Sherlock will analyze the documents you’ve sent, build an AI model, apply it to the larger document population, analyze and rank them for relevance and then present the next likely relevant documents for your review.
In short, Sherlock works like Pandora Internet Radio, only it finds great documents rather than great music. Just say: “Go get ’em, Sherlock.”
Sherlock’s strength is its speed, scalability and flexibility. It can analyze and rank a million documents in 100 milliseconds–10 million in a second. It can then deliver new documents in order of likely relevance, allowing you to review and mark them relevant (“Thumbs Up”) or not (“Thumbs Down”).
Send one document to Sherlock or send hundreds. Sherlock will analyze them, extract key terms, build an AI model and apply the model to millions of documents in milliseconds. Once you see Sherlock in action, you won’t want to go back to using keywords alone again.
Just say: “Go get ’em, Sherlock!”
Sherlock can start the hunt based on one document alone. But why not send hundreds of them to get him going?
"What amazing magic is this? It’s almost spooky."
Sherlock's on the Hunt
The screens below were taken from our demo site which holds just under two million documents, primarily emails from Jeb Bush’s two terms as governor. The topic for this exercise is climate change, which was a concern in Florida in the early 2000s.
The request (RFP) was this:
Climate Change — All documents concerning climate change, global warming, or carbon emissions, whether in Florida or otherwise.
It was created by NIST, the National Institute of Standards and Technology for their annual TREC (Text Retrieval) Conference. The screens we present were taken directly from the site. You can duplicate these results yourself if you like with access to our site.
There are a number of approaches you could take to begin this exercise. The most obvious in a keyword world is to use Boolean syntax to create a keyword search. Let’s start with some of the keywords in the request.
In this case, we will start with a simple keyword search. Rather than worry about search syntax, we will just enter some logical keywords using our unique “Free Form” search feature. Simply enter words or phrases enclosed with square brackets. Merlin will find documents with the greatest number of search terms without worrying about search syntax.
You can use this handy format to paste in paragraphs of text or even the text of a specific request for production. While you still have full-featured keyword search syntax in Merlin IS, many users find this a simpler and easier way to begin a keyword search. Remember to sort your results by Relevance (High to Low).
Sending a Document to Sherlock
Our search has found a good document about regulating carbon dioxide. We might continue advancing through the results to see what other documents our search has turned up. Or, we might instead just send the document to Sherlock and see if he can find more good documents on climate change.
We click on Send to Sherlock to start the process. That’s all there is to it.
Naming Your Session
Sherlock gives you the option to name your session and to have Sherlock place positive documents (“Thumbs Up”) in a selected folder.
Naming a session allows you to return to it at a later time or let others participate. Sherlock will retain the earlier training as it looks for more good documents.
Finding Another Great Document
In an instant, Sherlock will analyze the document you sent, build an AI model around its key terms and apply that model to the other documents on your site.
Notice that Sherlock has already identified other key terms that were not included with your search. This is a key part of what makes Sherlock unique. You can start with initial terms you think might be effective. Sherlock will quickly find others and highlight them for you.
Take a look at the document from Sherlock. If it is relevant to your inquiry, give it a Thumbs Up. If not, give it a Thumbs Down to help Sherlock get on the right track. Sherlock will quickly bring back another document that will further your investigation.
As we noted earlier, Sherlock will highlight key terms (often not part of your initial search) that you may find important. We offer two ways to view them. Just click on the Key Terms link at the top right.
You can view key terms in a Word Cloud. The more important terms are larger in size.
Or view the top key terms in list format. They too are ranked in order of importance.
You can boost key terms to let Sherlock know you deem them particularly important. Or choose to suppress terms that Sherlock should downplay. Either way, Sherlock will take your signals into account the next time you send a document judgment. Think of it like waiving a hankie under your digital bloodhound’s nose.
Continue Your Review
As you keep judging documents, Sherlock gets smarter, much like Pandora gets better at finding new music. Continue with Sherlock until you find what you need, or he stops finding relevant documents.
We track your progress at the top of the screen.
Sherlock Just Gets Smarter
As you keep judging documents, Sherlock gets smarter, much like Pandora gets better at finding new music. You can watch the Key Terms evolve as Sherlock hones in on his quarry.
Notice the new terms Sherlock has moved up in the ranking.
Continue Finding Great Documents
As you continue your investigation, Sherlock just keeps finding great documents. Click Thumbs Up or Thumbs down to guide Sherlock as he works the spoor.
Click Thumbs Up or Thumbs down to guide Sherlock as he works the spoor.
And Even More
We continue the hunt, learning that some call Governor Bush a “fossil fool.”
That is an interesting keyword we might never have thought of on our own.
Moving to Cluster Mode
We continue the hunt, learning that carbon dioxide in the state has soared by more than 347 percent in the last 40 years. But rather than continue document by document, let’s switch to Cluster mode.
Sherlock will send us the next fifty likely relevant documents, but grouped in clusters of documents with highly-similar content.
Viewing Documents by Cluster
You can enhance review efficiency by using Sherlock’s unique Cluster Batching algorithm. Rather than review documents one by one, ask Sherlock to send the next 50 documents in ranked order for your review. Sherlock will group your results in individual clusters of documents with similar content. You can then review documents individually or tag them by cluster.
The number of clusters will vary depending on the documents. But typically Sherlock creates about a dozen individual clusters.
The Advantage of Cluster Views
In this case, Sherlock has grouped 36 documents together in Cluster 10. You can review fields and document excerpts quickly. in this view, which is similar to our Snippets View for search results.
Cluster view is a powerful tool that can improve review efficiency by an order of magnitude. Imagine being able to review and tag 50 documents in ten minutes. This is in sharp contrast to doc by doc review, which might take an hour to complete.
Viewing Document Text
At any point during Cluster Review, you can view individual document text or switch to Native View. You can also tag or folder documents individually.
The Text Preview mode is a quick way to make sure Sherlock keeps on track with the documents he brings.
Finishing the Session
When you are done, choose an exit option to return to where you started. From there, you can continue with your keyword search or start fresh with a new search.
At any time, you can return to the session you created by sending Sherlock another document.
Sending Multiple Documents
You can send multiple documents to Sherlock as easily as one. From a folder or from search results, select multiple documents and select Send to Sherlock.
More Good Documents
As you can see, Sherlock has found another relevant document. You can continue the hunt to see if there are others.
Sherlock in Action
Want to see more examples of Sherlock in action? Here are two more of his most important cases.
As we mentioned earlier, Sherlock can start with one or hundreds of training documents. Watch fast he creates a new AI model and analyzes just under two million documents in this short video.
During Governor Bush’s term as governor, the Florida Legislature considered a bill to allow venue’s to resell tickets. Watch as Sherlock quickly finds relevant documents in about ticket scalping.
What makes Sherlock revolutionary is his blinding speed. Many existing TAR (technology assisted review) engines require hours to analyze and rank a million documents. Sherlock can do the job in 100 milliseconds, the blink of an eye. And there are no practical scaling limits. Indeed, Sherlock can rank ten million documents in a second or two.
We believe Sherlock’s speed and scalability will change document search, much like how Pandora changed the way we find new music. It allows users to find relevant documents quickly and easily, without having to master complex keyword search syntax. Its utility goes beyond legal, reaching to healthcare, corporate investigations, FOIA requests and other searches involving large document populations.
Welcome to Search 2.0
Contact us to find out how you can move up to Search 2.0 as well.