When you’re deciding on how to integrate a chatbot into your customer service or marketing strategy, if you’re not a developer, more than likely, it’s an added step to understand the terminology around chatbots.
You’re not alone if you’re not clear on how each term works together. Buzz-terms and acronyms around chatbots can have (somewhat) similar meanings. Alas, this isn’t Coachella; you don’t need to pretend to know every band name in the line-up (or in chatbot vocabulary) if you don’t. Tech moves fast. It’s okay. We’re here to help.
Here are 10 acronyms commonly used in conjunction with chatbots and helpful definitions for each:
First, Chatbots — A Quick History
Haven’t bots been around for a long while now? What’s the difference between chatbots and chat? Why the surge in interest now?
At the core, bots are software programs created to perform specific, automated, repetitive tasks. They work faster and more efficiently in these repetitive tasks than do humans and have been a helpful liaison for almost 30 years in executing mundane but essential tasks. Googlebot, referred to as a spider, of course, has been crawling the internet, indexing pages for a few decades. Like the Night’s Watch in Game of Thrones (because, Game of Thrones), bots can patrol and sweep a site for White Walk — err, any nefarious bots that can enter.
Bots have evolved from performing simple tasks to simulating conversations either by text or voice. Work on AI and bot-to-human conversation has been advancing since the Turing test of 1950. But the tipping point occurred in 2016 when Facebook’s Messenger app opened to third party developers. The intersection of a ripe marketing channel and advancement in AI expanded opportunities for businesses to use chatbots for better customer experience.
So what’s the difference between chatbots and chat? When you interact with a chatbot, you’re interacting with a machine. In some customer service structuring, a hand-off can occur to a human. In live chat, you’re interacting with a human, which you’ve probably become accustomed to as a pop-window or “chat” option on a business site.
1. Natural Language Understanding (NLU) and Natural Language Processing (NLP)
NLU and NLP are the input and output process for understanding language. Natural language understanding is the reading comprehension level of a machine, or input, while natural language processing is the steps it takes to reach a goal, which in this case is responding to the information entered. These aren’t new terms, of course, but the more advancement here the better a chatbot functions. To date, NLU is the more challenging area to surmount because of how contextual human conversation is.
2. Natural Language Processing (NLP)
For a more detailed look at this output process, natural language processing is how computers take the language received and create a response. The method combines approaches in artificial intelligence, computer science and computing linguistics. Words, then phrases then sentence structures are analyzed and a response is made based on analyzed data. NLP algorithms are built on machine learning algorithms, which analyze data and use inference to produce the best answers. The more data given, the better the answers.
Real World Application-Common tools like spell check and language translation, and automated questions, use NLP.
3. Machine Learning (ML)
Machine learning is how computers are programmed to automatically learn from experience through data gathering. ML algorithms digest data iteratively and, effectively “learn” from the process so that hand-coding or programming is not necessary.
Real World Application- An example are those uncanny Facebook or Amazon ads that pop-up in your feed related to your recent searches.
4. Artificial Intelligence (AI)
Artificial intelligence is the end-goal of these processes. It includes computer science, linguistics, psychology and neuroscience, in essence, every field needed to replicate humanly intelligent output by a machine. The better these underlying processes function to make contextual, concise meaning within any and every conversation, the better simulation of human intelligence.
5. Deep Learning (DL)
Deep learning is one way to approach machine learning. Instead of task algorithms, a machine learns experientially and more comprehensively based on learning data sets or representations. It can be directed, partially-directed or not explicitly directed, and deep learning can be performed completely by algorithm.
6. Messaging as a Platform (MaaP)
Over 4 billion users are expected within the messaging space by year’s end. Seemingly, everyone uses messaging. We bet your next slice of pizza your grandparent’s have even texted you recently. Messaging has become a remarkably sizable platform for brands and organizations to raise customer engagement by using application-to-person (A2P) messaging with chatbots and artificial intelligence (and yes, we’ve covered A2P in #8 below). Instead of leaving one site to go to another for a purchase or information, these interactions can all occur in a personalized, secure fashion in a messaging app. Messaging as a platform is the next generation of how we can use messaging and interact online.
7. Conversation as a Platform (CaaP)
The premise within conversation as a platform is that applications can perform tasks based on interacting as part of a conversation. At the Build 2016 conference, Microsoft CEO Satya Nadella contended that conversation as a platform will change the way people interact with their computers.
8. Application to Person (A2P)
Any mobile message sent from an application to a person is an A2P interaction.
Real World Application-If you’ve received a verification code via text or received an alert, these are all initiated from an application.
9. Representational State Transfer (REST) API-
This application programming interface is the most common way of designing networked applications, with HTTP, the application protocol for the web. This is the way applications work and function well together. A well-programmed API allows a seamless and wide breadth of functionality and is designed so that it doesn’t inhibit a user’s performance. You may also see SOAP, or simple object access protocol, which has historically been the approach to application interface, but in terms of chatbots, more and more organizations are opting for REST.
10. Conversational API
In most chatbots you’ll encounter now, conversational application programming interface is backed by NLP-NLU tools. It works with natural language and combined with messaging (these can be REST-based) APIs to search and respond to your queries. Conversational APIs are evolving, particularly as it relates to voice queries. Conversational APIs using natural language input will be used by bots to network across platforms and domains to deliver the best reply.
The goal of chatbot technology is to combine ease, speed and quality to arrive at the best response. The better the logical output and contextual response, the better an experience you’ll have with a chatbot. Currently, chatbots are still categorically specific use bots or generalist. To date, specificity wins in customer experience. Though, if we’re being honest, a generalist bot that has no command limitations is the expectation for customer experience with chatbots (#workgoals).
Use this list as a reference, and if you’re still unclear about how chatbots work, check out our post on chatbots and trends in 2018.