Simple chatbots

Simple chatbots do not include components that are capable of understanding the human language of their users – either in written or in spoken form. Although the interface is reminiscent of messenger services such as WhatsApp, users cannot enter free text, but can only choose from response alternatives provided by the bot via a button.

Dialogues usually follow a predefined path. In certain application situations, this can be an advantage over conventional user interfaces from the user’s perspective.

However, real user dialogues with natural language are not possible in this way.

From the provider’s point of view, the fact that simple chatbots of this type can be used as a basic surface technology is an advantage. They can be gradually expanded to include speech understanding functionality.

Assistants with speech comprehension

If the chatbot is to conduct real user dialogues, appropriate components for understanding and natural language processing (NLP) must be integrated.

Ensuring that the chatbot’s NLP component is sufficiently trained is crucial for the quality of language comprehension. For this purpose, possible intentions of the users are anticipated and classified as intents. Typical linguistic expressions must be stored as training data for each intent.

For this reason, assistants with speech comprehension are usually developed for a specific, precisely defined content domain.

They can be supplemented by components that convert spoken language into text (speech to text) and transmit it to the assistant, which can then process the text in the same way as typed text. The assistant’s answer is in text form and can be output – read out – to the user by a component for speech synthesis (text to speech).  

Example: Alexa-Watson-Connect (PoC)

Alexa-Watson

Play video

Typical application scenarios and objectives

From an economic point of view, assistants with speech comprehension are particularly suitable in the following application scenarios:

  • For frequently recurring service requests and processes
  • For the manual processing of simple enquiries, where this is very time-consuming
  • For answering enquiries when there are long waiting times
  • For customers who appreciate having their enquiries handled efficiently and who only need to speak to a human in difficult cases

Users mostly have the following objectives

  • Cost reduction
  • Availability 24 hours a day, 7 days a week
  • Relief for employees
  • Focus on difficult cases and special cases
  • Availability of novel customer interfaces

What can the laboratory offer?

Together with our partners in practice, we carry out research into innovative technologies that make artificial intelligence usable in concrete fields of application.

An important aim of the laboratory is to transfer knowledge from the university to business practice – mainly, but not only, to make it easier for small and medium-sized enterprises (SMEs) to access these technologies.

Get involved and try out your ideas in our laboratory!

Further information and contact details

Name Contact
Thomas Bahlinger Thomas Bahlinger
Prof. Dr.

Further activities at the Laboratory for Applied Artificial Intelligence

AI conference in summer semester 2019

OHM-Journal of the Nuremberg Institute of Technology: Interview with Prof. Zimmermann and Prof. Bahlinger

Future Engineering (with Fraunhofer Institute)