Chatbot Research – An Industry Overview Pt 2
IBM Chatbot Tools
View IBM Website Here
IBM has created a set of tools known as ‘Conversation’ in their Watson range of machine learning tools. Watson Conversation allows the developer to quickly build, test and deploy a bot or virtual agent. IBM Watson has a variety of complexity levels available to the developer to make use of, from its a graphical user interface tool or to the developer’s ability to make use of Watson’s REST API.
Both sets of tools make use of certain types of functions that dictate the flow and structure of the chatbots conversation, they are intents, entities and dialog.
An Intent represents the purpose or rationale behind a user’s input, normally consisting of a question within the chatbots capabilities. It is advised when making a chatbot to supply lots of examples of user inputs and which intents they map to in order to help better train the chatbot.
An Entity represents a term or object that is relevant to the intents and provides a specific context for the intent. To be able to efficiently train the chatbot to recognise the entities, the developer should list the possible values for each entity and synonyms the user might enter.
A Dialog is the branching flow of the desired conversation that defines how the chatbot responds when it recognises the defined intents and entities.
IBM Watson attempts to make the creation of a chatbot as easy as possible for the developer by making many of the management systems autonomous or dealt with on IBM’s side. One example is that, as the developer adds the various functions and their information, IBM is training the chatbot so the developer doesn’t have to initiate training whenever a new set of information is available. IBM Watson’s Conversation service also allows developers to connect additional Watson services to the chatbot functionality allowing it to perform actions such as analyse user inputs and have speech-to-text functionality. This also means that IBM offers the ability to add machine learning capabilities into chatbots through the use of their own Watson Machine Learning service.
How IBM Watson Conversation works is that the developer’s application is configured to send signals to the IBM Watson Conversation service, from here the service is trained to recognise the various intents and entities provided by the application. IBM Watson Conversation makes use of a dialog flow which allows the chatbot to respond in a natural way to the input provided by the user. A Dialog flow is a flow type which consists of interactions between a user and a system and is used to document the nature and sequence of these interactions.