Artificial Intelligence in Natural Disaster: Data Crawling- and Text Mining-Based Literature Review

Abstract

A literature review is an essential part of research. Beginning researchers who would like to conduct research in any field commonly review previous papers to identify trends and gaps in research. However, conducting a literature review is challenging, particularly for novice researchers. First, synthesising previous related works can be time-consuming. Second, accessing expensive databases such as Scopus and Web of Science might be inaccessible for novice researchers in developing countries. Thus, the purpose of this paper is to present a review performed by data crawling ArXiv and using latent Dirichlet allocation (LDA), a topic modelling algorithm. In this paper, we provide an overview of the topic models of natural disasters and artificial intelligence. Papers from 2017–2022 were crawled using Python with specific keywords, and only papers written in English were eligible for further analysis. Then, the papers were pre-processed by lowering the text, removing punctuation and stop words, tokenising, and lemmatising with POS tagging techniques. Pre-processed papers were then trained using the LDA algorithm. For evaluation, coherence metrics were used to determine the optimal number of topics. There were five topics resulting from the topic modelling algorithm. The clusters were qualitatively analysed and labelled as image classification, algorithms, social media analysis, natural language processing, and machine learning. Further findings are also discussed.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 2 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.