You can contact me already today if you are looking for a supervisor for your thesis. If you have a topic in mind that you think might fit my general research direction, please suggest it.
Send me an email or contact me for a consultation (see page “For students”)
Supervised research themes
- Digital transformation – digitalization of business processes, management of digital changes
- Conversational agents (CA), virtual assistants (VA), chatbots, large language models (LLMs)
- Predictive analytics – forecasting and other automated calculations and analyses
- BPM (business process management) – management of business processes using AI
- Implementation of AI hardware in business processes – cloud computing, edge computing – at the production line, service unit, department, company or consortium level.
- AI in management – decision support systems, rule-based approaches, decision trees, etc.
- Industry 4.0, smart manufacturing, supply chain optimization, preventive maintenance, automation of production processes, etc.
- AI in marketing – customer behavior analysis, customer satisfaction, automated content creation and content analysis, customer behavior forecasting, personalized marketing, consumer feedback analysis, etc.
- AI in customer service – chatbots, sentiment analysis, customer support automation, etc.
- AI in finance – financial analysis, risk assessment, fraud detection, etc.
- AI in Human Resources – Recruitment, Employee Performance Analysis, Employee Wellbeing/Absenteeism Assessment, HR Analytics, Employee Productivity Assessment and Improvement, etc.
- AI in occupational health and safety (AI OHS) – automated risk analysis, risk monitoring, risk forecasting, risk communication, etc.
- other feasibility studies, effectiveness studies on the use of artificial intelligence in business.
5 questions when devising a thesis topic
When choosing the topic of the thesis, we must first clarify the following questions. During the first conversation, I generally also ask about your professional interests in order to find out how the planned job could support your development and career.
- In which field is the work carried out? (field of activity) Determine the field in which the research will be conducted, for example clothing manufacturing.
- What business process is being looked at? For example, customer service.
- What is the innovative solution? The research generally focuses on some innovative solution or method that enables a certain activity within a company to be carried out more efficiently. An innovative solution can be an existing solution that is already in use, or something the student tests with available resources or asks test subjects to evaluate. The research aims to determine whether the new method is indeed more effective than the conventional method.
- What data is collected? In research work, the central question revolves around data. Consider whether 1) data is taken from an existing business process or 2) data needs to be generated through experimentation, testing a solution, or conducting surveys.
- What should the results show? Generally, research finds out whether a new innovative solution/method is more effective than a traditional method. How does research demonstrate this?
+------------------------------------------+
| 1. Agreeing on the topic and methodology |
| with the supervisor |
+------------------------------------------+
|
v
+------------------------------------------+
| 2. Data collection (plan |
| min 6 weeks) |
+------------------------------------------+
|
v
+------------------------------------------+
| 3. Data analysis and writing |
| results (plan min 2 weeks) |
+------------------------------------------+
|
v
+------------------------------------------+
| 4. Submit the final version of |
| the work |
+------------------------------------------+
|
v
+------------------------------------------+
| 5. Powerpoint Slide Series Oral |
| Presentation |
+------------------------------------------+
Joonis 1. Uurimistöö peamised tegevused (Koppel 2022)Sample topics
Below is a selection of sample topics and methodological options related to them. The described methodology is illustrative and does not express the details that will be agreed upon with the supervisor when planning the study. This is a limited selection of topics, and the student may propose their own topic or focus on any of the topics below. The topics are suitable for research papers at all levels, including term papers, bachelor’s and master’s theses. A doctoral thesis solves research questions from several angles and usually covers several topics. If technical tools are needed, they will be provided by the instructor’s laboratory if possible.
CATEGORY: Virtual agents taking over human activities in front office and back office
TOPIC: Virtual agent instructs employees to acquire new know-how and to carry out specific work processes
METHOD: training a virtual agent (chatbot) to find the information you are looking for faster and present it to the user; intervention – testers, n employees test virtualti, measure whether there is faster intervention or manual solution (time is measured and subjective assessment is asked)
DATA: data tables, free text guides containing key business know-how
RESULTS: whether working with a virtual agent was faster than the usual manual tasks such as reading n instructions
TOPIC: Building a retail chatbot to diversify the company’s contact channels and facilitate customer communication
METHOD: chatbot assembly, testing by a pilot group, questionnaire survey
DATA: chatbot usage indicators (number of calls, length, etc.); customer feedback (questionnaire survey)
RESULTS: whether the chatbot – increases customer satisfaction, – is able to bring more customers to the company, – increases sales
TOPIC: Virtual secretary, mail robot – Using large language model to respond to customer emails
METHOD: Building a virtual secretary that receives customers’ emails, analyzes those, composes and sends out responses. Testing by a pilot group, questionnaire survey.
DATA: Human expert analysis of automatically composed/sent out emails (questionnaire); customer feedback (questionnaire)
TOPIC: Text – speech/ speech to text – bringing apps into the business processes, e.g. Amazon Alexa style virtual consultant
METHOD: replacement of traditional manual processes, e.g. compilation of meeting transcription into speech-to-text by automatic method; measurement of accuracy, measurement of user satisfaction (questionnaire survey)
DATA: meeting or other auto transcriptions vs human-written transcriptions, accuracy/effectiveness of auto transcriptions for business process
RESULTS: whether an AI-based automated method can replace a manual method; what are the strengths, weaknesses of both methods; user satisfaction
TOPIC: Artificial intelligence as an interviewer/interviewer
METHOD: text – for speech / kõne – using text applications to conduct interviews and transcribe statements. Use cases include job application interviews; subordinate-boss periodic conversations.
DATA: automatic transcriptions, their comparison with manual work; user satisfaction with the automated solution (questionnaire survey v interview)
RESULTS: Can AI as an interviewer replace a human worker;
CATEGORY: Automated content creation when replacing human workers
TOPIC: Applying content creation tools to business texts
METHOD: Two types of content generation: a) human writer, b) AI-generated. Comparing the results of s isulome – can a person understand which ones created AI, – which text would be chosen by the test client. Human testers should know the theme of the text generated by the AI or belong to the target audience of the text.
DATA: AI generated texts, n sales texts; user feedback (questionnaire survey and comparison evaluation)
RESULTS: Descriptive statistics (number of characters in texts, number of sentences, etc.); semantic analysis;
Human tester’s assessment of texts (questionnaire). The assessment of the generated texts by the professional, n the person doing the work.
CATEGORY: Content analytics in supporting the company’s work processes
TOPIC: Content analytics, semantic analysis application of tools to automate processes in the enterprise
METHOD:
1) free text analysis (centition analysis)
2) Automation – tools become part of some business processes
3) evaluation of efficiency through a) error measurement and b) user questioning. Example
DATA: documents, e-mails (correspondence with customers, colleagues), customer feedback, etc. business information exchange
RESULTS: assessment of significance, positivity and other aspects.
TOPIC: Automated meeting transcription and memomaking
METHOD: speech-to-text tools, transcription, comparison of auto transcription with what a person has built
DATA: speech recordings of meetings, memos of human-written co-opism
RESULTS: comparison of auto transcription and auto-meeting memo with human performance, comparison of total errors, highlighting types of errors
CATEGORY: Predictive business analytics
TOPIC: Status/situation prediction
METHOD: different methods of solving the research question are possible
DATA: data collected by the organization or data collection with monitoring devices (e.g. health data of workers, building data or other time series)
RESULTS: different research question resolution metrics
TOPIC: Forecasting of activity indicators; Forecasting the conditions, influences of the external environment of the company’s activities
METHOD: historical analysis of company data
tforecasting leases (time series)
tulemuss display. Example
DATA: the company’s financial statements, financial ratios or other performance indicators; data on factors that affect performance directly (internally) and indirectly (n weather data)
RESULTS: e.g. comparison of the prediction of turnover and other activity indicators with actual results, calculation of accuracy (%)
TOPIC: Forecasting customers ‘ solvency/ creditworthiness
METHOD: Machine learning methods, predictive analysis. Validation of the solution in real-life conditions.
DATA: historical data of the company, payments from customers, outstanding payments
RESULTS: forecast – probability of whether the client will fulfill payment obligations / comparison with actual data
TOPIC: Client service time forecasting
METHOD: sentiment analysis
DATA: customer communication of the company (free text inquiries, emails, speech transcriptions and answers to them – the content of the text, length and time required to respond)
RESULTS: forecast for each in-service – how long does it take to have customer service time, comparison of the forecast with reality;
TOPIC: Identifying potential new buyers/orders to know who it makes sense to make direct offers to
METHOD: Predictive analysis. Validation of the solution in real-life conditions. Näide
DATA: data collected about the potential buyer of the company, n joining the mailing list, visits to our e-shop v website, etc.
RESULTS: Accuracy of the prediction, key factors. Real-life/estimable benefits of implementing the solution.
TOPIC: Predicting customers who are likely to opt out of our service in order to make direct offers to them to maintain a customer relationship
METHOD: Machine learning methods, predictive analysis. Validation of the solution in real-life conditions. Example
DATA: Data collected in the course of the customer relationship (customer database with on-demand services and purchase indicators)
RESULTS: Accuracy of the prediction, key factors. Real-life/estimable benefits of implementing the solution.
TOPIC: Assessment of employee satisfaction/absenteism/departure
METHOD: Evaluation of employee satisfaction with predictive analytics tools, the goal is to identify the employee’s dissatisfaction / decrease in productivity early in order to increase their motivation and prevent the employee from leaving; validation of the solution in a real-life environment / data;
DATA: Data characterizing the activities of employees – data collected by the company, as well as available from public sources.
RESULTS: Accuracy of the prediction, identification of key factors. Real-life/estimable benefits of implementing the solution.
TOPIC: Implementing automated business analytics to assess risks
METHOD: real-time or predictive calculation of the risk level; real-life testing/validation of the tool and evaluation of its accuracy and usefulness (computationally and estimatingly – questionnaire)
DATA: data on variables related to the selected risk; assessment of the risk analysis providers (questionnaire survey)
RESULTS: Important metrics from which it is concluded,
CATEGORY: Audio/video model training to improve business processes, improve products and services
TOPIC: Determination of employee well-being, stress, productivity
METHOD: the study of the employee’s activities on the basis of historical data or novelly collected data. Vit is possible to carry out sensor measurements. Possible solution testing test by individuals (employees). Example 1
DATA: time series: employee performance indicators; direct and indirect factors affecting the employee; possible sensor data
RESULTS: Can the developed solution provide a prediction of the employee’s increase in stress, well-being and decrease in productivity
CATEGORY: Industry 4.0, Smart Manufacturing
TOPIC: Planning production needs
METHOD: Machine learning methods, predictive analysis. Validation of the solution in real-life conditions. User feedback via questionnaire survey is also possible.
DATA: Production indicators, direct and indirect agents and other important metrics
RESULTS: Validity of the prediction with reality, i.e. accuracy (%); user feedback (questionnaire survey)
TOPIC: Planning maintenance needs to prevent machine failures and work stoppages
METHOD: Machine learning methods, predictive analysis. Validation of the solution in real-life conditions. User feedback via questionnaire survey is also possible. Example
DATA: Production volume of a production line or machine, data from condition sensors, cost factors and other important metrics
RESULTS: Validity of the prediction with reality, i.e. accuracy (%); user feedback (questionnaire survey)
TOPIC: Edge computing
METHOD: AI applications in small form factor, e.g. Arduino, Raspberry Pi etc. The small computer collects and processes data using machine learning methods. If possible, the output data will be followed by an automatically triggered action.
DATA: 1) Raw data collected by small computers, 2) output data obtained from data processing, n decision support, 3) test users’ assessment of output data (n questionnaire)
RESULTS: Efficiency of the process supported by edge calculation compared to the previous way of working. User satisfaction.
CATEGORY: Managing digital transformation
TOPIC: SWOT for the application of business AI in relation to organizational structure, skills, business strategies, change management, communication – on the example of a company
METHOD: document analysis (measurement of effectiveness by numerical indicators), interviews, questionnaires. Observation of the process, intervention if necessary.
DATA: analysis of the company, quantitative and qualitative metrics important for solving the research question
RESULTS: How well AI can take on business processes that were previously run by people. How well AI can replace a person in business. Emetrics of affectivity
TOPIC: AI takes over jobs – what are the structural changes in the labour market
METHOD: Analysis of different areas of the economy or one industry, as well as trend prediction
DATA: Identification of important variables, time series of variables, questionnaire for system users
RESULTS: Quantitative (numerical) and qualitative (estimation) description of labour market variables
CATEGORY: Providing a leverage effect on an organization’s data assets
TOPIC: Planning the organization’s graph database. The organization can be an enterprise or other institution.
METHOD: A focus is selected that comprehensively maps a single action function, n customer order acceptance. Transforming an existing relational database(s) into a graph database OR creating a graph database from scratch. Platform n neo4j or other. Data is added to the graph database, tested by users. The differences in the use of two types of databases are compared: speed, functionality.
DATA: Users of the database are a sample of the survey, who is one of the target groups of the company, n employees – feedback is collected from them (questionnaire survey or interview). If possible, the speed of use is measured in time and the efficiency of work is measured by the volume of data being processed.
RESULTS: Conclusions are made regarding the usefulness for the organization.
Currently supervising
Using AI to enhance workers’ productivity, Olga Tšernikova, PhD-student
Artificial intelligence-based decision support system in organization management, Kristjan Vomm, Master’s thesis
Leveraging AI for Corporate Crisis Management, Angelica Tikk, Master’s thesis
Knowledge transfer through a virtual assistant, Kätlin Veesaar, Master’s thesis
Supervised works
ESTONIAN CONSUMER ATTITUDES AND E-MAIL INVOLVEMENT TO ADVERTISING CREATED BY ARTIFICIAL INTELLIGENCE, Henri Valk, Master’s Thesis, 2024
CONSUMER ATTITUDES TOWARDS ARTIFICIAL INTELLIGENCE SYNTHETIC MEDIA, Jana Karimova, Master’s thesis, 2024
FORECASTING THE RESOURCE NEEDS OF THE TICKET CONTROL PROCESS USING ARTIFICIAL INTELLIGENCE, Rakell Pärnamäe, Master’s Thesis, 2024
APPLICATION OF ARTIFICIAL INTELLIGENCE IN THE RECRUITMENT PROCESS AND COMPARISON WITH PERSONNEL EMPLOYEES, Sandra Vallik, Master’s thesis, 2024
DIGITAL TRANSFORMATION IN ALIBABA’S E-COMMERCE ECOSYSTEM: LEVERAGING TECHNOLOGY FOR TRANSFORMATION, Subarna Islam Sharna, Master’s Thesis, 2024
Use of chatbot in customer service to improve efficiency in Indian fintech industry, Swapnil Ashok Kasar, Master’s thesis, 2024
FINNISH CONSUMER ATTITUDES TOWARDS AI-GENERATED MARKETING MESSAGES, Micaela Myllymäki, Bachelor’s thesis, 2024
Possibilities of implementing an artificial intelligence chatbot in the Estonian building chemicals sector using the example of customer satisfaction, Sten Tiidt, Master’s thesis, 2023
The effectiveness of artificial intelligence-based content creation tools in managing the marketing processes of an Estonian e-commerce company, Brenda Lepp, Master’s thesis, 2023
Performance of AI-generated content in content marketing, Sohaib Arshad, Master’s thesis, 2023
Employing AI to evaluate companies’ performance – An approach to utilize ChatGPT for financial ratio analysis to assist managers in decision making, Sonia Ranjbar Sarvandani, Master’s thesis, 2023
Improving employee engagement with the help of Engy chatbot, Aleksei Rozenberg, Master’s thesis, 2022
The combined effect of Virtual Assistants and Augmented Reality tools to enhance customer engagement in furniture industry, Matias Kaijomaa, Bachelor’s thesis, 2022
THE SUITABILITY OF AN ARTIFICIAL INTELLIGENCE-BASED VIRTUAL AGENT IN DEALING WITH WORK STRESS, Signe Bergert, Bachelor’s Thesis, 2022
Smartphone Electromagnetic Fields’ Effect on Human Heart Rate Variability and Cognitive Performance, Valentine Augustine Okoye, Master’s thesis. 2020