Date of Award

2013

Degree Type

Thesis

Degree Name

Doctor of Philosophy

School

School of Computer and Security Science

Faculty

Health, Engineering and Science

First Advisor

Dr Mark Brogan

Second Advisor

Dr Mike Johnstone

Abstract

Subject access to images is a major issue for image collections. Research is needed to understand how indexing and tagging contribute to make the subjects of historic photographs accessible.

This thesis firstly investigates the evidence of cognitive dissonance between indexers and users in the way they attribute subjects to historic photographs, and, secondly, how indexers and users might work together to enhance subject description. It analyses how current indexing and social tagging represent the subject content of historic photographs. It also suggests a practical way indexers can work with taggers to deal with the classic problem of resource constraints and to enhance metadata to make photo collections more accessible. In an original application of the Shatford/Panofsky classification matrix within the applications domain of historic images, patterns of subject attribution are explored between taggers and professional indexers.

The study was conducted in two stages. The first stage (Studies A to D) investigated how professional indexers and taggers represent the subject content of historic photographs and revealed differences based on Shatford/Panofsky. The indexers (Study A) demonstrated a propensity for specific and generic subjects and almost complete avoidance of abstracts. In contrast, a pilot study with users (Study B) and with baseline taggers (Studies C and D) showed their propensity for generics and equal inclination to specifics and abstracts. The evidence supports the conclusion that indexers and users approach the subject content of historic photographs differently, demonstrating cognitive dissonance, a conflict between how they appear to think about and interpret images.

The second stage (Study E) demonstrated that an online training intervention affected tagging behaviour. The intervention resulted in increased tagging and fuller representation of all subject facets according to the Shatford/Panofsky classification matrix. The evidence showed that trained taggers tagged more generic and abstract facets than untrained taggers. Importantly, this suggests that training supports the annotation of the higher levels of subject content and so potentially provides enhanced intellectual access.

The research demonstrated a practical way institutions can work with taggers to extend the representation of subject content in historic photographs. Improved subject description is critical for intellectual access and retrieval in the cultural heritage space. Through systematic application of the training method a richer corpus of descriptors might be created that enhances machine based information retrieval via automatic extraction.

Share

 
COinS