Pengembangan Chatbot Berbasis WhatsApp Menggunakan Dialogflow dan NLP untuk Layanan Informasi Toko Fotokopi

The use of WhatsApp-based information services has become a primary option for providing information to customers. However, manual information services at
photocopy shops still face constraints such as limited response times and inaccuracies in handling inquiries. This study aims to develop a WhatsApp-based chatbot by integrating Dialogflow and Landbot with a Natural Language
Processing (NLP) approach to deliver automated information services at the Sahabat Photocopy Shop. The research methods include collecting a conversational dataset, data preprocessing (cleaning and tokenization), intent creation, data balancing via oversampling, chatbot training, and integration with WhatsApp through Landbot. Performance testing using the F1 score demonstrated high results for the majority of intents, with a value of 1.0 for the Barang (Goods),
Informasi (Information), Jadwal (Schedule), Selamat Datang (Greeting), and Tempat (Location) intents, although the Jasa (Services) and Default Fallback intents exhibited a recall of 0.8 due to limited phrase variation. User evaluation using the System Usability Scale (SUS) yielded a score of 80.0, which falls into the “acceptable” category, indicating that the chatbot is easy to use and well accepted by users. Overall, the developed chatbot successfully mitigates time-management
issues and provides efficient information services, although improvements in conversational consistency and expansion of the dataset for specific intents are required to ensure accurate interactions.