Converse is a Platform for building Conversational Bots, which allows businesses to automate conversations with their customers. It provides an AI-powered conversation experience to users. Omnitive Converse allows end users to query information through a conversational interface, making questions in natural language and engaging in dialogues with Converse. The UI is a familiar chat-like window where users type or speak, and Converse replies.
Make a question that initiates several dialogs and allows the user to choose from the options. For example, “I would like to order pizza?”, then select from the given options, followed by “Any toppings that you would like to add?” and so on.
The users can formulate their questions in natural language. This is aimed at making it easy and natural to express their Queries. Omnitive Converse has a number of techniques to facilitate that. NLU Engine understands a wide range of natural input. For example, Singlish/malformed sentences.
Syntax analysis. Converse understands the structure of the sentence to derive its meaning.
Entity recognition. Converse will recognise domain specific terms within the user input, e.g. “how to apply for BTO”. These concepts can be managed from the Administration module.
Spelling Errors & Grammar Handling. Converse can detect typos and spelling errors in the user messages, and it will try to interpret the input as the correct word. Often the users won’t type proper sentences but simply lists of words, since they may be used to other UIs such as common search engines (e.g. “loan apply”). Omnitive Converse will still try to infer the user intent from the words, and provide the most suitable answer.
Synonyms. The user may express some concepts with different but equivalent words, and Converse will understand them as the same thing, e.g. “what is the price/cost/charge/fee of season parking”. For domain-specific synonyms, users can configure through the Ontology Manager.
Converse can have multiple unique bots. Each bot can answer questions based on their knowledge base as well as certain questions or intents that are shared across all bots. Multi-Bot is managed by a Super Administrator, who manages intent sets (intent library) between all bots and one bot respectively.