Abstract
As businesses increasingly adopt digital tools to streamline operations, artificial intelligence (AI)-based chatbots have emerged as vital components for enhancing customer communication and supporting financial management within accounting services. This research focuses on reliable AI-powered accounting chatbots capable of handling complex financial tasks while enhancing customer communication and user satisfaction. The goal of this study is to establish AI-based chatbots in accounting services to improve financial management assistance and customer communication. This paper presents a novel Raven Roosting-tuned Adaptive Bidirectional Long Short-Term Memory (RR-ABiLSTM) model designed to classify financial queries and enhance contextual understanding in conversations to improve customer communications. The dataset encompasses both structured and unstructured data from accounting conversations, constituting a domain-specific corpus focusing on common accounting tasks. Data preprocessing included text cleaning and tokenization applied to the acquired data. Subsequently, feature extraction was performed using Word2Vec. The RR algorithm was utilized to optimize hyperparameters and feature selection, while BiLSTM ensures a deep understanding of contextual relationships in conversations, thereby enhancing accuracy and efficiency in processing financial queries. Furthermore, a dynamic training mechanism was integrated, allowing the chatbot to continually adapt to increasing consumer demands without downtime. The proposed method was implemented using Python software, and its performance was compared with traditional algorithms. The overall metrics—F1-score (87.75%), precision (89.25%), recall (86.24%), and accuracy (90%)—illustrate that the suggested model significantly improves customer engagement, reduces the workload of accountants, and enhances the overall efficiency of accounting services by providing reliable financial support.
Keywords
Get full access to this article
View all access options for this article.
