Conversational Web Interaction Guided By Automated Web Page Segmentation

Date

2024-9-04

Author

Suleiman, Suleiman Abubakar Tanko

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

42
views

0
downloads

Cite This

The web has grown to become an essential part of our daily lives. And yet, even with its widespread use, we mostly explore it through its graphical interface, using clicks and typing operations for interaction. However, this is limited in the level of accessibility it can provide to visually impaired people and those who are temporarily unable to use normal interaction methods. This can be due many reasons, such as when trying to use the computer in a very bright or dim environment, thus making it difficult to see the screen. These situations make users experience situationally induced disabilities (SIID). So far, users have screen readers and voice assistants to help them navigate the web in such situations. Recently, however, research has been done into the creation of conversational web interfaces to allow users to interact with the web using their voice. The most prominent of these studies employ the use of heuristics and manual annotations in identifying the different sections within the web page structure. These solutions, however, are not without their challenges. Manual annotation for websites is costly, and heuristics are difficult to implement in a generalized way, given the diverse nature of websites. In this thesis, we propose a conversational web architecture that incorporates a special automated segmentation algorithm, and present our prototype implementation of said architecture called ConvoBot, as a proof of concept tool, using existing technologies to address the limitations of manual annotations, while also testing the usability of such a prototype. Our implementation was tested by 12 participants and evaluated using the System Usability Scale (SUS). Our results show that ConvoBot achieved an average SUS score of 80, which is considered acceptable and `Good' by the SUS adjective rating. The main contribution of this thesis is the automated segmentation-driven architecture and a usable prototype implementation of that architecture for conversational web interaction.

Subject Keywords

Conversational Web Interaction, Conversational Web Architecture, Voice- based Interface, Automated Webpage Segmentation

URI

https://hdl.handle.net/11511/111590

Collections

Northern Cyprus Campus, Thesis