Bachelor Thesis Defence: "Analyzing Behavioural Patterns in Online Knowledge Collaborations: A Case Study of Wikidata" by Hong Zhu (Nov 07, 2019)
News from Nov 06, 2019
November 7th, 3.15pm, the Thesis Defense: "Analyzing Behavioural Patterns in Online Knowledge Collaborations: A Case Study of Wikidata" by Hong Zhu will take place at HCC Lab.
Research into online knowledge ontologies like Wikidata has often overlooked the important interaction patterns in exploring the collaboration temporal dynamics. This thesis outlines a step-by-step procedure for investigating the frequent patterns of sequential behaviors in terms of data quality. Using a dataset of 146,450 revisions to 500 Wikidata items as a case study, the thesis employs a machine learning based tool as a proxy to evaluate the temporal item quality, and utilizes the Jensen-Shannon distance as a metric to quantify the quality variations. Referenced from the methods already developed in fields like
bio-informatics, the case study mined the revisions collected using the PrefixSpan algorithm in a constraint-based fashion. Having identified the frequent sequential editing patterns that satisfy specific quality constraints, the study demonstrated the impacts of different sequence identifications regarding data quality.
The defence will be held in English.
Location: HCC Lab, Königin-Luise-Straße 24/26, Room 120