Social identity information retrieval system (SIIRS) : final report / by Brian Riordan. : D68-3/135-2014E-PDF

Development of the SIIRS system focused on the specific problem of detecting and classifying sociostructural beliefs from Social Identity Theory -- beliefs about the legitimacy, stability, and permeability of social groups and their status. A dataset of social media comments centering on the issue of income inequality was collected and analyzed, and a system was built to classify implicit author beliefs on the legitimacy of class-based income disparity. Because of the heavily implicit nature of sociostructural belief expression, the detection problem was formulated as a form of text classification. The SIIRS system achieved classification accuracies competitive with results from similar tasks in social media analysis, demonstrating the feasibility of the approach.

Lien permanent pour cette publication :
publications.gc.ca/pub?id=9.821005&sl=1

• Format MARC XML
• Format MARC HTML

Demander des formats alternatifs