Proceeding of the Faculty of technical Sciences, Novi Sad

Electrotechnical and Computer Engineering

Vol. 40 No. 10 (2025): Proceedings of the Faculty of Technical Sciences

INSURANCE DOMAIN QUESTION-ANSWERING DIGITAL ASSISTANT

  • Стефан Драгичевић
  • Јелена Сливка
DOI:
https://doi.org/10.24867/32BE19Dragicevic
Submitted
October 19, 2025
Published
2026-01-02

Abstract

This paper presents the development of an advanced chatbot system designed to answer insurancerelated questions automatically. The system is implemented as a web application that provides users with quick and clear answers to their questions. The system consists of several key modules: a query preprocessing module, a model for finding relevant questions (high recall), a model for filtering and selecting the most relevant answer (high precision), a sentiment classification module, and a keyword extraction module.

References

  1. [1] Adamopoulou & Moussiades (2020). An Overview of Chatbot Technology. AIAI 2020, IFIP Advances, 584. Springer.
  2. [2] Jatnikaa et al. (2019). Word2VecModel Analysis for Semantic Similarities. ICCSCI 2019.
  3. [3] Lv et al. Siamese Multiplicative LSTM for Semantic Text Similarity.
  4. [4] Devika et al. Sentiment Analysis: A Comparative Study On Different Approaches. ICRTCSE.
  5. [5] Bonta et al. (2019). A Comprehensive Study on Lexicon-Based Approaches. Asian J Comp Sci Tech, 8(S2).
  6. [6] Xu & Zhang (2020). Extracting Keywords from Texts. ICIIK.
  7. [7] https://paperswithcode.com/sota/question-answeringon-quora-question-pairs
  8. [8] https://huggingface.co/datasets/quoracompetitions/quora
  9. [9] https://www.kaggle.com/datasets/lakshmi25npathi/imdbdataset-of-50k-movie-reviews
  10. [10] https://github.com/shuzi/insuranceQA