Computational Social Science and Web Science

Supervisor: Henry Thompson

Systems such as ChatGPT pose an existential threat to the value of the Web.  Web archives such as Common Crawl provide a substantial fraction of the training material for LLMs.  As more and more web pages are generated by LLMs, we risk a feedback spiral of increasing garbage.  This project will explore potential mitigation strategies based on detecting how easily extracted information from Common Crawl index entries for pages of varying age change at two key points in Web history: the advent of Web 2.0 early this century and the release of ChatGPT in 2023.

Supervisor: Henry Thompson

Finding or making and recording connections is constitutive of scholarship of all kinds.  The recent enormous growth of material available to scholars, beginning with the advent of the Web and accelerating as a result of widespread movement towards open content, has made the business of making connections at the same time more important and more difficult.

References on the Web are URIs, Uniform Resource Identifiers, but common usage patterns for these mean they often fail to be reliable, leading to growth in the design and use of so-called _persistent_ identifiers (PIDs), with tightly controlled usage regulations intended to improve reliability.

But, many discipline-specific PID systems have failed 'in the marketplace'.  The purpose of this project is to produce a set of case studies of successes and failures, and a set of guidelines for success.

Supervisor: Henry Thompson

Patterns of Web use change over time, and analysing these changes can help predict where the Web is going.  Data from web proxy logs provides evidence of what kinds of HTTP requests are being made, for example showing a recent upsurge in use of encrypted requests.  Data from large-scale collections of Web pages such as Common Crawl and the Internet Archive give a different kind of evidence. The purpose of this project is to identify and quantify major shifts in usage, using multiple sources to ensure validity.

Supervisor: Henry Thompson

This project combines aspects of the previous two, looking specifically at how usage of different contemporary PID systems are developing, both in numeric terms and with respect to developments in the contexts in which PIDs are appearing.