Well, one reason is that we just keep creating more and more of it all the time. We continually interact with information and those interactions, which are the key to understanding our intent, are captured and used to make our next interactions smarter. It’s the feedback loop!
And businesses depend on that feedback to better engage with their customers and employees.
As we get smarter and leverage the cloud to rapidly analyze massive amounts of data, the machine learning models that are being developed can understand natural language better than ever.
The deep learning technologies that are available today can virtually read and understand the words we feed into them and then help drive the artificial intelligence behind question-answering systems and conversational interfaces, such as chatbots and virtual assistants.
How Do Chatbots and Virtual Assistants Make Search Smarter?
Hold on. Wait a minute... Isn’t a chatbot and virtual assistant really the same thing? Pretty much.
A virtual assistant is just an extension of the chatbot, but there are some differences to note especially as we start to think about connecting them to “smart search” technology. Let’s take a minute to make sure we’re all on the same page here.
A chatbot is an artificial intelligence that simulates human conversation. They’re generally used in lieu of providing direct human contact for basic tasks such as retrieving product details or offering “canned” responses to frequently asked questions.
Natural Language Understanding (NLU) platforms like Google’s Dialogflow or Amazon’s Lex allow you to quickly and easily build conversational interfaces and embed them into major messaging platforms, websites, search and mobile applications.
Ditto for the virtual assistant. VAs can do all that and more! Think of Siri, Amazon Alexa or Google Assistant. Not only can they hold chatbot conversations, but they can also “assist” by answering questions and performing basic tasks. “Hey, Siri... How high is the space needle?” Or, “Alexa... Play my morning playlist.”
Now you can even say things to them like, “Hey Google... Make me a restaurant reservation, and start a conversation with a robot that will ultimately result in a belly full of your favorite meal from a local food establishment.
So, how does search fit into all of this? It’s quite simple actually. Search is made smarter by taking those deep learning models we talked about earlier and applying them to information as it flows in and out of the search index.
Some of the latest search offerings now on the market have embraced deep learning technology and are delivering some eureka moments to enterprises with an eye toward the future of search. Amazon Kendra is the new enterprise search service from Amazon Web Services and Smart Answers is now available in the latest enterprise version of Lucidworks Fusion. Both of these search platforms are great for delivering powerful “smart” applications wherever you need them.
Picture this… First, your information is fed into the system and analyzed using deep learning analysis derived from models included with the search engine plus models that you can train yourself. All of that analysis is captured and stored with the data inside the search engine.
Now we can ask questions using natural language, send those questions through the same deep learning model and voilà! Search can now easily match the analysis of your question, regardless of how you asked it, with the analysis stored in the index and give you the answer in a few milliseconds. Not just a blue link and a paragraph with your keywords highlighted… but a real, honest-to-goodness answer in large font, just as you would expect from a modern-day search engine.
This is powerful. Now we can wire up our favorite apps to our smart search index and start that conversation with SharePoint that we’ve always wanted.
“Hey, SharePoint. Show me all the latest marketing documents for product XYZ.”
Expand that line of thought out further and the potential to develop and streamline company processes is endless. Helpdesk and customer support applications are some of the most common starting places.
Building state-of-the-art smart applications is not rocket science, but it is computer science and therefore not trivial either. Choosing a strong and stable search platform plus teaming up with a good search partner is essential for making the most out of today’s smart search technology.