The use of conversational robots to support intranet: a case study in Elektro Ljubljana

Authors

  • Živa Hudobivnik Elektro Ljubljana d. d.
  • Maja Savinek Elektro Ljubljana d. d.
  • Marina Trkman Univerza v Ljubljani, Fakulteta za upravo

DOI:

https://doi.org/10.31449/upinf.219

Keywords:

conversational robot, large language model, intranet, implementation, focus group

Abstract

One of the important technologies that are key to the popularity of conversational robots today are the large language models. These allow conversational robots to use contextual memory to support dialogue in a way that is similar to that between humans. Conversational robots benefit customers in entertainment, industry and education. Few papers have addressed the use of a conversational robot in a corporate setting as support to intranet. The aim of this paper is to present the process of designing a conversational robot in Elektro Ljubljana. The Belko prototype was created based on the GPT-3.5-turbo model. We used focus groups to prepare its primary knowledge base. A group of employees tested Belko to see if it could support the company's intranet. Based on the findings, a list of user requirements for the professional development of the Belka conversational robot was drawn up. The case study can be useful to companies deciding to use a conversational robot as a company intranet, as it provides a unique insight into its creation.

Author Biographies

Živa Hudobivnik, Elektro Ljubljana d. d.

Živa Hudobivnik je magistrica družboslovne informatike. Magistrirala je z delom Načrtovanje pogovornega robota za pomoč zaposlenim v podjetju Elektro Ljubljana, d. d. Zaposlena je v podjetju Elektro Ljubljana, d. d., v Službi za napredno analitiko, kjer deluje na področju podpore uporabnikom, analitike in umetne inteligence.

Maja Savinek, Elektro Ljubljana d. d.

Maja Savinek je doktorska študentka na Fakulteti za računalništvo in informatiko Univerze v Ljubljani. Zaposlena je v podjetju Elektro Ljubljana, d. d., kjer se ukvarja z razvojem napredne analitike in algoritmov na področju elektroenergetike, s poudarkom na tehnologijah velepodatkov. Njeni strokovni prispevki so s področja umetne inteligence, podatkovnega rudarjenja in analize podatkov v elektroenergetiki. 

Marina Trkman, Univerza v Ljubljani, Fakulteta za upravo

Marina Trkman je docentka na Univerzi v Ljubljani; zaposlena na Fakulteti za upravo in Fakulteti za družbene vede. Doktorat je opravila na Fakulteti za računalništvo. Svoj podoktorski projekt je vodila na Institutu Jožefa Štefana. Raziskovalno se ukvarja: 1) s preiskovanjem dejavnikov, ki vplivajo na privzemanje informacijsko komunikacijskih tehnologij in 2) z analiziranjem poslovnih procesov za optimizacijo poslovanja. Kot prvi avtor je objavila članke v prestižnih revijah kot sta International journal of information management in pa Government information quarterly

Published

2024-03-07

How to Cite

[1]
Hudobivnik, Živa, Savinek, M. and Trkman, M. 2024. The use of conversational robots to support intranet: a case study in Elektro Ljubljana . Applied Informatics. 31, 4 (Mar. 2024). DOI:https://doi.org/10.31449/upinf.219.

Issue

Section

Scientific articles