Chatbots or virtual assistants in Public Administrations to democratize the use of open data

Fecha de la noticia: 16-08-2022

Image to illustrate the content of the post on chatbots

According to the latest analysis conducted by Gartner in September 2021, on Artificial Intelligence trends, Chatbots are one of the technologies that are closest to deliver effective productivity in less than 2 years. Figure 1, extracted from this report, shows that there are 4 technologies that are well past the peak of inflated expectations and are already starting to move out of the valley of disillusionment, towards states of greater maturity and stability, including chatbots, semantic search, machine vision and autonomous vehicles.

Figure 1-Trends in AI for the coming years.

In the specific case of chatbots, there are great expectations for productivity in the coming years thanks to the maturity of the different platforms available, both in Cloud Computing options and in open source projects, especially RASA or Xatkit. Currently it is relatively easy to develop a chatbot or virtual assistant without AI knowledge, using these platforms.

How does a chatbot work?

As an example, Figure 2 shows a diagram of the different components that a chatbot usually includes, in this case focused on the architecture of the RASA project.

Figure 2- RASA project architecture

One of the main components is the agent module, which acts as a controller of the data flow and is normally the system interface with the different input/output channels offered to users, such as chat applications, social networks, web or mobile applications, etc.

The NLU (Natural Languge Understanding) module is responsible for identifying the user's intention (what he/she wants to consult or do), entity extraction (what he/she is talking about) and response generation. It is considered a pipeline because several processes of different complexity are involved, in many cases even through the use of pre-trained Artificial Intelligence models.

Finally, the dialogue policies module defines the next step in a conversation, based on context and message history. This module is integrated with other subsystems such as the conversation store (tracker store) or the server that processes the actions necessary to respond to the user (action server).

Chatbots in open data portals as a mechanism to locate data and access information

There are more and more initiatives to empower citizens to consult open data through the use of chatbots, using natural language interfaces, thus increasing the net value offered by such data. The use of chatbots makes it possible to automate data collection based on interaction with the user and to respond in a simple, natural and fluid way, allowing the democratization of the value of open data.

At SOM Research Lab (Universitat Oberta de Catalunya) they were pioneers in the application of chatbots to improve citizens' access to open data through the Open Data for All and BODI (Bots to interact with open data - Conversational interfaces to facilitate access to public data) projects. You can find more information about the latter project in this article.

It is also worth mentioning the Aragón Open Data chatbot, from the open data portal of the Government of Aragón, which aims to bring the large amount of data available to citizens, so that they can take advantage of its information and value, avoiding any technical or knowledge barrier between the query made and the existing open data. The domains on which it offers information are: 

  • General information about Aragon and its territory
  • Tourism and travel in Aragon
  • Transportation and agriculture
  • Technical assistance or frequently asked questions about the information society.


These are just a few examples of the practical use of chatbots in the valorization of open data and their potential in the short term. In the coming years we will see more and more examples of virtual assistants in different scenarios, both in the field of public administrations and in private services, especially focused on improving user service in e-commerce applications and services arising from digital transformation initiatives.

Content prepared by José Barranquero, expert in Data Science and Quantum Computing.

The contents and points of view reflected in this publication are the sole responsibility of the author.