Virtual assistant, aka chatbot, is a cognitive computing system that acts like a human operator in engaging, guiding, and providing users with contextually relevant information. It is based upon the natural language question answering technique that allows a user to pose everyday questions, processes the questions to determine the informational requirements, and delivers information against such requirements. It can ask clarifying questions to decipher the user’s intent, helping to pinpoint the correct answer. Powered by Artificial Intelligence (AI), Virtual assistant continuously learns from new information to refine its effectiveness in delivering the most relevant and accurate responses.
Why Virtual Assistant?
Virtual assistant provides a highly scalable way to automate cognitive search and knowledge discovery processes and services for both customers and staff, without compromising the quality of interaction. The human operator-like capability reduces friction inherent in traditional text entry-based search while enabling organizations to address massive volumes of diverse inquiries in an efficient and consistent manner, which results in significant cost savings, shorter response time, and improved customer satisfaction.
How does Virtual Assistant work?
Virtual assistant has multiple elements working in close concert to deliver a seamless experience. First, there is the conversation component to handle the exchange of messages between user and system. Next, the intent classifier analyses the conversation to determine what the intent or topic of the conversation is most likely to be. The intent classifier uses machine learning (a branch of artificial intelligence) to develop and refine its ability to accurately classify intent.
To address a wide variety of questions, there are multiple engines in the answer server. Answer Bank provides FAQ-like responses to standard questions such as, “How can I turn off roaming on iPhone 7?” Fact Bank interprets a question to determine what the user is asking for and translates this into the correct syntax for querying whichever system this fact is stored in, such as a local repository or a remote database queried through a custom API. It provides factual answers to questions such as, “Who is Apple’s CEO?” Passage Extraction provides verbatim text from relevant sources in response to questions such as, "What did Apple’s CEO say about augmented reality?”