Add Thesis

Chatbot

The future of customer feedback

Written by Kevin Hoang Dinh

Paper category

Master Thesis

Subject

Computer Science

Year

2020

Abstract

Thesis Chatbot: Chatbots, also known as conversational agents, use natural language processing (NLP) to interact with users through interactive conversations based on text and voice [8]. Implemented on many different platforms (for example, applications, web services, chat channels), it responds to customer inquiries and questions. The uses of chatbots are endless, because it is becoming a part of people's daily lives. Apple's Siri, Amazon's Alexa, and Google Assistant are well-known multi-purpose agents, such as training, education, travel booking, transactions, customer service, or route planning [9]. Developers are constantly evolving complex agents, adding more capabilities and functions. In addition to these complex multi-purpose agents, there are agents dedicated to a single task, and these agents can be combined through a cloud communication platform as a service (PaaS). Popular platforms that specialize in conversational chatbots are Dialogflow, Twilio, and Azure Bot Service. The development of chatbots is on the rise. From simple code patterns to hardware products and embedded deep learning systems, chatbots have been created for many purposes and have proven their usefulness in many different situations [10]. Although each chatbot has different goals, they all focus on the same domain, conversational knowledge. The oldest chatbot ELIZA generates answers by matching keywords in user sentences [11]. This simple implementation has become the most cited conversation technology [12]. By using these technologies (for example, NPL, keyword extraction, deep learning, or machine learning), chatbots can understand user input and respond with appropriate answers, making the conversation more human-like, thus encouraging participants to continue using the service [13]. 2.2 Natural language processing Natural language processing (NLP) is a field study on the understanding of human languages ​​(including text and speech) by computers. The basic part of all chatbots is to interpret the user's input and return the desired output. NLP is the accumulation of many application fields such as machine translation, natural language text processing and summarization, user interface, multi-language and cross-language information retrieval (CLIR), speech recognition, artificial intelligence and expert systems [14]. NLP involves three main problems of natural language understanding (NLU), and computers need to solve these problems to complete human-like language processing. The first problem is the human thinking process, the second problem is the understanding of semantic input, and the last problem is world knowledge, that is, knowledge outside of the default program. In order not to confuse "processing" and "understanding", NLP focuses on the process of natural language interpretation, while NLU focuses on interpretation. 2.3 The focus of the extraction, conversion and loading project is to compare the quality, effectiveness and response rate of ordinary surveys with chatbots. 2.4 REST API Representational State Transfer (REST) ​​is a style of building distributed hypermedia systems. HTTP (Hypertext Transfer Protocol) is the main protocol for the most common REST implementation. RESTful APIs work around resources. Resources can be any type of data, objects, or services that can be accessed through identifiers, which are URLs (website addresses). The REST API uses HTTP as the protocol, so it can execute common HTTP methods. The use of resource identifiers and HTTP methods allows the client to perform common operations on the resource: • GET: Retrieve a representation of the resource. •POST: Create a new resource. The client must provide data to the new resource to create it. •Delete: delete the resource. The client must have an identity to delete a specific resource. The project will use REST API to build a web service to receive POST requests through webhooks. 2.5 Related work The goal of the project is to allow consumers to participate in voting through dialogue chatbots. Similar work is divided into two aspects and detailed as follows. Information agents Many developers have established agents to interact with their customers to retrieve information. Knowing this information, the agent can find relevant data and perform the correct tasks, such as educational support [15], booking travel, customer service [16] or advice [17]. If the agent does not know what kind of education the student has received, it will not be able to provide you with educational support. The traveler’s name, credit card and personal information are required to book a trip. Suggestions made without knowing the customer's reference material are irrelevant and will waste time. There are researches on agents that collect information, but there are not many studies in the questionnaire. Unlike surveys, the quality of information is not considered. The goals of this project may be similar to the other studies mentioned above. But it goes one step further and discovers more discoveries to improve the quality of the investigation. Improving the quality of surveys As mentioned above, the main purpose of this project is to find a way to improve the quality of surveys. Not only this project, many other studies have put forward a lot of efforts to improve the response rate. A 2006 study showed that personalization can increase the response rate by 4.4 percentage points, from 50.3 to 54.7 [18]. Other research on interactive feedback also shows improvements in online surveys [19]. Another study used inquiry to obtain better quality answers from respondents [20]. Learning from previous research, the project can apply these techniques to conversational agents to further improve the quality of investigations. When the chatbot communicates with the interviewee, the chatbot will let the server know what the interviewee answered in each interaction. Even if the interviewee did not complete the entire conversation, the project wanted to collect as many answers as possible. Thechatbot will send the data to an application suitable for processing it. Read Less