University of Twente Student Theses

Login

Natural language search for personal information retrieval : Benefits and limitations to the user’s experience in bookkeeping software.

Smienk, Jeroen (2023) Natural language search for personal information retrieval : Benefits and limitations to the user’s experience in bookkeeping software.

[img] PDF
2MB
Abstract:Natural language processing (NLP) is rapidly progressing and more specific areas such as natural language understanding (NLU) and natural language generation (NLG) are becoming incredibly advanced. However, the use of NLU in personal information retrieval is limited and disadvantages or lack of advantages to the user's experience may have hindered its adoption. Studies on how natural language search affects the user's experience are scarce and dated. NLP in search may transform the norm of forcing users to cognitively transform their information need into a structured path or keyword query towards the answer. The goal of this master's thesis was to find the benefits and limitations of natural language search to the user's experience of personal information retrieval in bookkeeping software. A limited prototype was developed of a state-of-the-art natural language search interaction for search in bookkeeping software, based on the interaction and technical designs for natural language search in related work. A scenario-based user study was designed to evaluate how the prototype system helps or limits users with their information needs. The qualitative and quantitative data collected from nineteen participants was analysed to identify four major themes, assess the prototype's performance, and produce concrete learnings. The main benefits of natural language search are described in terms of providing a more intuitive way for users to express their information needs, including the ability to describe complex information, unfamiliar or unknown items, and domain-specific language. Limitations discussed include conditioning to keywords and lack of trust in the system's capabilities that come with increased cognitive effort, and additional physical effort for inexperienced typists. Adoption by expert users is also limited due to perceived greater efficiency of their common methods. Clear introduction, but especially building trust through successful performance are crucial for overcoming limitations. The major themes, benefits, limitations, performance assessment, and learnings are discussed in detail and the thesis is concluded with advice for future work and continued development.
Item Type:Essay (Master)
Clients:
Moneybird, Enschede, The Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:50 technical science in general, 54 computer science, 77 psychology
Programme:Interaction Technology MSc (60030)
Link to this item:https://purl.utwente.nl/essays/94310
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page