1. Speech Recognition Engine and NLP for mixed languages for Banking Industry (Cantonese, English) 2. Natural Language Processor for mixed languages for Banking Industry (Cantonese, English) 3. Dialogue Management System with artificial intelligence 3.1 Logic Graph builder 3.2 Error/Ambiguity handling 3.3 Learn from users’ selections to build dialogue 3.4 Call statistics 4. Text based AI Chatbot for banking 5. Voice-based AI Chatbot for banking CS deliverables 1. CS Deliverable for Customer 1: 1.1 Documentation on the software architectures, feature requirements (CS Deliverable 1 for Customer 1) 1.2 Software on text-based chatbot for banking (CS Deliverable 2 for Customer 1) 1.3 Voice-based AI Chatbot for banking (CS Deliverable 3 for Customer 1) 2. CS Deliverable for Customer 2: 2.1 Executable software on the development of recognition technologies (recognition software) to be applied to Application based on Natural Language Processing technology (CS Deliverable for Customer 2) 2.2 A standalone voice recognition engine to convert speech to text with categorization (CS Deliverable for Customer 2)
Artificial intelligence (AI) and deep learning are here to transform the world of business by offering better customer experience and conducting targeted marketing. Chatbots leverage the power of AI to communicate with users. A Cantonese and English mixed chatbot specifically for the banking industry is the focus of this project. The voice recognition engine can differentiate and recognize English from Cantonese, including slangs, and colloquialisms, and understand the meaning of the sentences, even when spoken with a slight accent. Moreover, chatbots are taught to learn new phrases and handle ambiguity through deep learning to resolve queries by customers. By having an open API, the chatbots developed can be integrated with other systems, such as emails or fax, and can be embedded with promotional material. Big companies like IBM or Google usually may ignore this niche market because of the small market size; yet the demand is here, not just in Hong Kong locally, but also in the clustered cities of the Guangdong-Hong Kong-Macao Big Bay Area. This project could satisfy the needs of this market.