Optimization of Chatbot Model in Industry
Abstrak
A lot of people are now working in AI and ML. These two methods make data processing a lot better, which means that the results are faster, more accurate, and more adaptable. A report from McKinsey says that the use of AI and ML went up by 57% from 2020 to 2021. There was also a lot of growth in the Natural Language Processing (NLP) market in the Asia-Pacific region. The market grew from $2.7 billion to $4.2 billion in 2021. It is also expected to grow quickly until 2030. The goal of this study is to make industrial chatbot models like Sinar Mas Land's Digital Hub better so they can help with operational problems. This study used the CRISP-ML framework, which includes Business Insight, Data Insight, Data Preparation, Modeling, Evaluation, and Deployment. The company gave us a dataset with questions and answers in text form. We processed it by tokenizing it, removing stop words, and stemming it. Then we used TF-IDF to extract features from it. The SVM algorithm with a linear kernel was applied to develop a classification model capable of understanding and responding to user queries. The model was deployed using Streamlit for easy user access. Experimental results show that the developed model achieves an accurate rate of 95.77% with a linear kernel. Closed beta testing results indicate the chatbot received an average score of 87.50%. Future development recommendations include expanding the dataset, applying feature selection techniques to improve accuracy, and integrating generative-based features to enhance chatbot flexibility and performance.
Identifier
- DOI: https://doi.org/10.62527/joiv.9.6.3681
- OpenAlex: https://openalex.org/W7117275184