Model Design and Development for a User Support System Using Artificial Intelligence Techniques in Enterprise Resource Planning Software

For organizations, the error-free, consistent, and transparent management of business processes is crucial. Process management relates to the proper handling of data, achievable through enterprise resource planning (ERP) software. This software enables all departments, structured around a central database, to work together. From an organizational perspective, one of the most critical factors for the success of ERP software is its users. Users often require support when using ERP software for various reasons, such as lack of training, experience, or technical issues. Providing this support, whether internally or from external sources, incurs a cost.This study seeks to determine whether a structure can be developed where users can resolve their support requests independently. To achieve this goal, we propose a model for an intelligent user support system with an AI-based process engine for ERP users. Using this model, unstructured user support requests, created in natural language, are transformed into structured forms using natural language processing techniques and then classified using multi-class machine learning algorithms. The process relevant to the identified class is executed to generate a solution for the user's problem. In this study, 1,000 samples of text written in natural language across 10 different categories, based on real data, were used in the learning process. The data were analyzed using machine learning algorithms (K-Nearest Neighbors, Naive Bayes, C4.5, Support Vector Machine, Random Forest, Sequential Minimal Optimization, LibSVM) and deep learning algorithms (Long Short-Term Memory). The best classification was achieved with Sequential Minimal Optimization (SMO) using TF-IDF weighting. The developed chat and process robot helped convert requests or problems into structured forms and provided solutions. In addition to solving the problem through chat or process, technologies such as augmented reality and virtual reality were used.