Unveiling Latent Topics in Robotic Process Automation - an Approach based on Latent Dirichlet Allocation Smart Review

2024-04-08

arXiv.org

AI Summary

This study systematically investigates the thematic distribution, evolutionary trajectories, and scholarly impact of Robotic Process Automation (RPA) research to construct a scientific knowledge map. Methodologically, it pioneers the application of an unsupervised LDA topic model to over 2,000 RPA-related article abstracts, integrating standardized text preprocessing and interactive visualization to identify 100 fine-grained topics, subsequently synthesized into 15 high-impact core themes and their temporal evolution patterns. The primary contributions are threefold: (1) a reproducible, AI-augmented systematic literature review framework; (2) the first dynamic science map of the RPA domain; and (3) empirically grounded insights into the field’s knowledge structure, shifting research foci, and cross-topic interconnections—thereby providing actionable evidence for strategic research planning and frontier identification.

Technology Category

• Intelligent Robots: Other Foundations and Applications
• Knowledge Representation and Reasoning: Automated Reasoning and Theorem Proving
• Natural Language Processing: Safety and Robustness

Application Category

• Web Mining and Content Analysis: Topic discovery and tracking
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Robotic process automation (RPA) is a software technology that in recent years has gained a lot of attention and popularity. By now, research on RPA has spread into multiple research streams. This study aims to create a science map of RPA and its aspects by revealing latent topics related to RPA, their research interest, impact, and time development. We provide a systematic framework that is helpful to develop further research into this technology. By using an unsupervised machine learning method based on Latent Dirichlet Allocation, we were able to analyse over 2000 paper abstracts. Among these, we found 100 distinct study topics, 15 of which have been included in the science map we provide.

Problem

• Identify latent topics in Robotic Process Automation research
• Analyze research interest and impact of RPA topics
• Develop a science map for RPA using AI methods

Innovation

• Unsupervised machine learning for topic analysis
• Latent Dirichlet Allocation for abstract processing
• Science map creation from 2000+ papers