A Chatbot for Searching and Exploring Open Data: Implementation and Evaluation in E-Government

Abstract

In this paper, we present a chatbot to access open government data. Differently to similar systems reported in the research literature, the developed chatbot not only allows searching for data collections, but also exploring information within the collections. The exploration is done via complex queries that are easily built by non-expert users through a natural language conversation. Moreover, as another novel, differentiating contribution, we report a conducted user study aimed to evaluate the chatbot according to the achievement of a number of public service values, as well as measuring distinct objective and subjective metrics. Experimental results show that the proposed system outperforms traditional methods followed in open data portals. CCS Concepts: • Applied computing → E-government; • Information systems → Search interfaces; • Human-centered computing → Natural language interfaces; User studies.

The Adoption and Implementation of Artificial Intelligence Chatbots in Public Organizations: Evidence from U.S. State Governments

“…Our data also help to answer the research questions presented in this study. In general, our results indicate that chatbot adoption and implementation are multidimensional processes, involving both technical and non-technical factors; these results identify factors that determine chatbot adoption and implementation, contrasting with studies that emphasize the role of technical factors (Androutsopoulou et al, 2019; Cantador et al, 2021; Carvalho & Barbosa, 2019; Keyner et al, 2019; Lommatzsch, 2018). More importantly, the factors that determine the two stages differ significantly, a point that will be further elaborated below.…”

“…From the users' perspective, chatbots alleviate user information overload by displaying the information that is most relevant to the requested service or question (Lommatzsch, 2018;van Noordt & Misuraca, 2019). In addition, artificial chatbots ensure a constant level of answer quality (Cantador et al, 2021;Keyner et al, 2019;Lommatzsch, 2018). They also reduce waiting times, thanks to their 24/7 availability (Carvalho & Barbosa, 2019;Makasi et al, 2020Makasi et al, , 2022.…”

“…Third, another topic illuminated by the existing literature is the process of implementing chatbots and the potential challenges that public organizations may confront and need to deal with. In this regard, scholars have focused on the technical aspects of chatbots, with some studies offering demonstrations of system design and development (Androutsopoulou et al, 2019; Cantador et al, 2021; Carvalho & Barbosa, 2019; Keyner et al, 2019; Lommatzsch, 2018). While researchers have proposed various system architectures, these approaches have several components in common.…”

E-Government in Nigeria: Can Generative AI Serve as a Tool for Civic Engagement?

“…Moving beyond theoretical arguments, researchers have actively developed and tested generative AI prototypes for civic use cases. Notably, Cantador et al (2021) implemented an AI conversational tool powered by language models to facilitate conversational search and exploration of open government data repositories. This system demonstrated how citizens could intuitively query vast open data resources through natural language interactions, potentially democratising access to public information reserves.…”

“…One of its many merits consists in the ability of the generative AI to create automated chatbots that can interact with citizens in a natural, conversational manner. These chatbots can answer queries, provide information and guide users through various processes which improves the accessibility and efficiency of e-government services (Cantador et al, 2021;Mahapatra et al, 2012). In addition to chatbots, generative AI may also be used to develop interactive policy-guides which can present complex policy information in a simplified, easy-to-understand format.…”