Integrating CRM and ERP Insights for Optimized Product Development Using CNN-LSTM Hybrid Models

Enhancing product development processes through insightful data integration in business finance is crucial for aligning offerings with market demand and optimizing resource allocation. Financial sentiment analysis, which involves extracting and analyzing sentiments from financial news, social media, and other textual sources, plays a pivotal role in understanding customer feedback and market trends. However, existing models often struggle to integrate local feature extraction with long-range dependency modeling effectively. To address this gap, a hybrid model is proposed, combining the strengths of Convolutional Neural Networks (CNN) and long short-term memory (LSTM) networks. The hybrid model integrates local feature extraction with long-range dependency modeling for financial sentiment analysis, utilizing customer feedback from CRM systems and cost data from ERP systems to prioritize product development. Dropout regularization is employed to prevent overfitting, while L2 regularization is used to penalize large weights, promoting simpler models. For hyperparameter tuning, an extensive grid search and cross-validation were conducted. The evaluation results demonstrate the superior performance of the hybrid model, achieving an overall accuracy of 95.9%, with individual label accuracies of 93.7% for negative sentiment, 95.6% for neutral sentiment, and 98.4% for positive sentiment. These findings indicate significant improvements over other hybrid models in the literature, with the proposed model outperforming recent works by margins ranging from 0.9% to 5.7%. The hybrid model provides a robust solution for integrating CRM and ERP insights, offering significant advancements over existing models and paving the way for more sophisticated tools in the domain of business finance.