Chatbot Interface

Written by Hannah Boast on 19 Aug, 19
City-wide innovation ecosystem
C/O Elzware: Royalty-free stock photo ID: 1426044410

 

How do we support people in accessing and engaging with data like that found at Open Data Bristol?

Winner of the Data Brains competition in April, Elzware is doing just this through developing a skeleton ChatBot system, otherwise known as Dotty, to assist with interactions with Open Data Bristol. This system will enable engagement through a conversational AI system with the Open Data themes on the basis of natural language with text input through a webpage.  

This is much more than a simple ChatBot. Using a publicly available open source ChatScript A.I/NLP engine, a conversational AI system has been created. Compared to the traditional ChatBot, this type of system is much more sophisticated, nuanced, and can be updated accurately and quickly.  This is mainly due to its more sophisticated layering of pattern matching, dealing with various levels and structures of language and proactive/naturalistic interactive tone and style.

It has a number of conversational functions including:

  • an exact set of pattern matches on keywords related to and the entire theme text strings, and a lower ranked fuzzy matching on terms;
  • an FAQ topic for non data specific questions, but those that are relevant to Open Data more broadly;
  • a “no-match” topic where user inputs which are not matched are handled conversationally, and proactive suggestions are made; and,
  • an intent matching methodology to create basic user-typing for future development and reporting.

Future work includes looking at how to best improve the user experience, and loading it onto https://www.elzware.com/open_data/ The back end of the system runs on Ubuntu, an open source operating system which is installed and maintained in line with GDPR principles.

Font size