University of Twente Student Theses


Navigating Semantic Shifts : A Visual Tool for Exploring Word Meaning Change

Kazi, Raef (2023) Navigating Semantic Shifts : A Visual Tool for Exploring Word Meaning Change.

[img] PDF
Abstract:Detecting and tracking semantic shifts in language is a complex task with signifi- cant implications for various fields. Computational methods like topic modeling have emerged as powerful tools for this purpose. However, challenges persist in evalu- ating, selecting, and interpreting suitable models. Existing metrics like coherence scores often fall short in capturing the nuances of word sense changes, leading to laborious manual inspection and parameter tuning. This research aims to bridge this gap by developing a user-friendly visualization tool that enhances the understanding of semantic shifts and facilitates model selection. This study combines insights from semantic shift literature and data visualization techniques. It begins with a comprehensive review of semantic shift visualization methods to identify the requisite features to visualize for the creation of an innovative tool. This tool integrates various visualization elements, including intertopic maps, stacked bar charts, steamgraphs, and coherence scores. Using a topic modeling approach, the study detects word senses across seven decades, enabling users to visualize semantic shifts and assess the impact of different parameters. A user study, involving a diverse group of participants, evaluates the tool’s effectiveness in exploring word senses and making model selections. The research yields a comprehensive visualization tool that empowers users to explore diverse word senses and their evolution over time. Users leverage temporal features to navigate subtle changes in word senses. The tool aligns users’ intuitive model choices with quantitative measures, bolstering their confidence in model se- lection. However, challenges related to user familiarity, visual complexity, and the subjectivity of semantic shift identification emerge. This research contributes to the semantic shift analysis field by evaluating a vi- sualization system on the basis of its usability and task support. It represents an interdisciplinary intersection of linguistics, data visualization, and user experience design. To further enhance the tool’s utility, future work should integrate user feed- back, broaden participant demographics, enable real-time computation, and offer multilingual support.
Item Type:Essay (Master)
OCLC, Leiden, Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:02 science and culture in general, 17 linguistics and theory of literature, 54 computer science
Programme:Business Information Technology MSc (60025)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page