University of Twente Student Theses

Login

A Closed Domain Question Answering System for a Genetic Counselor within the ECG Family Clinic

Keurhorst, Maaike (2023) A Closed Domain Question Answering System for a Genetic Counselor within the ECG Family Clinic.

[img] PDF
14MB
Abstract:Background: The eCG Family Clinic is a platform developed for online genetic counseling to help family members make informed decisions regarding the testing of hereditary cardiovascular diseases. Objective: This research explores how an open question answering feature can be developed for a virtual assistant for genetic counseling. We aim to answer this by: 1) exploring which techniques for question answering systems could be used and 2) examining the users’ attitude towards this open question answering feature. Method: Using literature, a retrieval-based approach was selected. This approach aimed to detect similarities between user input and potential answers in the knowledge base. Named entities, shared bigrams, semantic similarity, and shared part-of-speech tags are used as features to detect the similarity. The knowledge base consisted of excerpts extracted from websites approved by specialists. Additionally, the prototype was evaluated with users by an qualitative study guided by the Technology Acceptance Model (TAM). User evaluation consisted of a set of tasks using the eCG Family Clinic and a semi-structured interview. The user study was analysed using a thematic analysis. In total 8 participants, all master’s students, were recruited. Results: We found that the algorithm had a top-1, top-2, and top-3 accuracy of 19.2%, 25.3%, and 26.3%. This accuracy is lower than related open question features found in the literature. Thematic analysis using the themes of ’Perceived Usefulness’, ’Dialogue Management’, and ’Ease of Use’ showed us that participants see the value of the addition of the open question feature. However, more accuracy in this feature is needed to increase its usefulness. Conclusion: While the accuracy of the methods needs to be improved, this study forms a foundation for future research. User testing demonstrated that users recognized the value and potential of the open question feature but indicated that an increase in accuracy and personalization is needed. For future research, it is suggested to focus on increasing the accuracy and addressing user suggestions.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Programme:Interaction Technology MSc (60030)
Link to this item:https://purl.utwente.nl/essays/96195
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page