University NLP Research: Qualitative Methodology for Data Science

This report examines the application of qualitative methodologies in data science research, specifically within the context of Natural Language Processing (NLP) for chatbot development. It begins by introducing NLP and its significance, particularly in creating chatbots for customer support. The core of the report focuses on comparing three qualitative research methods: action research, narrative inquiry, and grounded theory, to determine the most suitable approach for the research problem. The author argues for the superiority of action research due to its collaborative and iterative nature, which allows for continuous input from clients and dynamic adjustments to the chatbot's design. The report then outlines the stages of action research, including planning, acting, developing, and reflecting, and details how these stages would be applied to the chatbot development process. The report also includes a discussion of why narrative research and grounded theory are less appropriate for this specific application. Finally, the report concludes by emphasizing the importance of client feedback and iterative testing in the development of a successful NLP-based chatbot.

Running head: QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
Application of Qualitative Methodology on Data Science Research (Natural Language
Processing)
Name of the Student:
Name of the University:
Author note:

1QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
Introduction
Natural Language Processing (NLP) is one of major trends in data science. NLP refers to
the technology in which people converse with machines and those machines are able to
recognize the speech, analyze and understand the language and sometimes they are to give a
reply also (Young et al., 2018). For example, Alexa, powered by Amazon, and Siri, powered by
Google are two major examples of NLP, in which the machines are not only able to perceive and
analyze the speech but can also reply accordingly. This report will give an overview of the
suitability of action research methodology on conducting research on NLP based research study
by comparing with narrative inquiry and grounded theory, followed by an outline of applying the
methodology on the study.
Research problem and qualitative research questions
A chatbot is required to be developed that would interact with customers for providing
customer support by addressing their queries and complaints. Hence, NLP is required in the
chatbot to design and implement the human touch in the customer support communications
without actually employing a human to do that. Thus, NLP system of data science should be
integrated in the chatbot so that the machine can perceive the speech of the customers, analyze
those and reply them accurately. The following research questions should be addressed using
action research methodology.
The research questions are:
1. What qualitative aspects should be programmed into the NLP of the chatbot to enhance
the analytical power of the chatbot?
2. How the reciprocation vocabulary can be improved for the chatbot?

2QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
Explanation of action research methodology and reasons for it to be best fit in the intended
problem and topic
Action research methodology is one of the most suitable methodology that is used in
qualitative research. As defined by McNiff (2016), action research is a form of participatory
research, in which collaborative and cyclical ways of producing knowledge and solution to a
problem. In many research issues it is found that the clients and the solution providing
companies work together by following the action research methodology to create, share and
apply the knowledge to get the solution to the research issues. Under this particular qualitative
research method, the goals are primarily change-oriented, than only knowledge gathering and
this methodology is highly suitable for diagnosing a research problem that requires much inputs
from the clients or organization to get the most optimized and customized solution to the
organizational problems. As active participation of both the researcher and client occurs in this
method, the solution is continuously changing to get the optimized outcome (Chevalier &
Buckles, 2019).
Narrative research is a qualitative research method in which the research is based on the
written or verbal words or the visual representation of the individuals who are subject of the
research. In this method, the emphasis is put on the words and images provided by the
individuals and those are analyzed as put forward originally. Hence, as stated by Bruce et al.
(2016), the aim of narrative research is to explore and hypothesize the human experiences as
represented in the textual form. The interpretation often depends on the knowledge, and
rationality of the researcher.
Grounded theory is another methodology for qualitative research. In this method, the
techniques of data collection and analysis has reciprocal relationship. This method is concerned

3QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
with new theory generation from the information. As defined by Thornberg (2017), the
methodology of grounded theory involves the process of generating new theory from
systematically gathered and analyzed data. The most striking feature of this method is the
simultaneous process of data collection and analysis (Flick, 2018). This theory is mostly suitable
for social research, that is, to explore and uncover the social behaviors and relationships among
the people and the researcher must not have any pre-conceived notion while applying the
grounded theory into the social research.
By comparing the definitions and features of the three qualitative research methods, it is
understood that in the given research study, action research would be most suitable because as
the chatbot is required to reply to the customers by identifying and analyzing their queries,
hence, it is important to provide the inputs to the machine. This is possible only with action
research method. In this method, the clients and the researcher would come together to determine
the replies that the chatbot would give to the customers.
An explanation for not selecting the other two methodologies
While narrative research is focused on textual analysis of the words spoken by people,
grounded theory involves the process of new theory generation through simultaneous data
collection and analysis, and none of this methods are suitable for natural language processing
(NLP). Moreover, both these methods generate static outcome. Through action research, the
immediate outcome is obtained and for developing the chatbot, it is important to provide most
relevant input to the machine by applying artificial intelligence and NLP (Verlinde et al., 2019).
Hence, narrative research and grounded theory methodology were not chosen for this study.

4QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
Reason for Narrative research and grounded theory not fitting the intended research
Narrative research and grounded theory do not involve the clients to provide knowledge
and inputs for the chatbot and the processes do not generate change oriented outcomes. The NLP
process for developing the chatbot requires dynamic outcome as the machine will address variety
of customer queries and complaints. And hence, the NLP and the implementation of artificial
intelligence must be extremely advanced and continuous input from the client and testing of the
programming is required. These are possible only under action research method and both the
narrative research and grounded theory methodologies are not suitable for applying and
analyzing NLP to a research.

5QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
A preliminary outline for the process of applying methodology in designing the study
There are four stages of action research methodology:
 Planning stage
 Acting stage
 Developing stage
 Reflecting stage (Mertler, 2009)
These stages of research include nine steps (Mertler, 2009) and their application in the study is as
follows:
Stages of action
research
Action research steps Application in research study
Planning stage Identification and limitation
of topic
Application of NLP in developing a
chatbot for efficient customer support.
Information collection Secondary research on the existing chatbots
in the market, such as, Alexa and Siri
Review of literature Exploration and review of existing
literature on the relevant research subjects,
such as, NLP, artificial intelligence and
their efficient implementation
Research plan development Qualitative research method to be
developed as the chatbot would perceive
and analyze qualitative texts from the
customers and would reply qualitatively.
Research questions are made according to
the client requirement
Acting stage Plan implementation and data
collection
Qualitative data about customer support
will be collected from the company

6QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
database.
Potential responses of the chatbot will be
collected from the company’s customer
support executives through qualitative
interview.
Data analysis Qualitative analysis method will be applied
on the data to sort out the potential
responses to be given by the chatbot
NLP programming will be integrated to
develop the chatbot with the analysis
outcome
Developing stage Development of action plan Testing of the chatbot and the data will be
done in collaboration with the client.
The clients will examine the working of the
chatbot and its programming to check
whether its working perfectly to resolve
customer issues
Reflecting stage Sharing and communication
outcomes
Cyclical and iterative process will be
followed after testing of the chatbot and its
programming.
Outcomes will be shared and
communicated with the clients.
Based on the feedback, possible responses
and its NLP programming will be modified
for betterment.
Reflection on the process Clients’ feedback, testing outcome and the
entire process of NLP programming and
integrating in the chatbot will be reviewed
for improving the features and efficiency of
the chatbot in the next version

7QUALITATIVE METHODOLOGY ON DATA SCIENCE RESEARCH
References
Bruce, A., Beuthin, R., Sheilds, L., Molzahn, A., & Schick-Makaroff, K. (2016). Narrative
research evolving: Evolving through narrative research. International Journal of
Qualitative Methods, 15(1), 1609406916659292.
Chevalier, J. M., & Buckles, D. J. (2019). Participatory action research: Theory and methods for
engaged inquiry. Routledge.
Flick, U. (2018). Doing grounded theory (Vol. 9). Sage.
McNiff, J. (2016). You and your action research project. Routledge.
Mertler, C. A. (2009). Action research: Teachers as researchers in the classroom. Sage.
Thornberg, R. (2017). Grounded theory. The BERA/SAGE handbook of educational research, 1,
355-375.
Verlinde, S., De Wachter, L., Laffut, A., Blanpain, K., Peeters, G., Sevenants, K., & D’Hertefelt,
M. (2019, August). Writing assistants: from word lists to NLP and artificial intelligence.
In EUROCALL Conference 2019 (p. 161).
Young, T., Hazarika, D., Poria, S., & Cambria, E. (2018). Recent trends in deep learning based
natural language processing. ieee Computational intelligenCe magazine, 13(3), 55-75.