Analysis and Policy Observatory (APO) is a evidence platform for policy and practice that is both open access and open-sourced. APO has collected policy resources for over 15 years that are accessed by researchers and policymakers. The open source approach has included user contributions, including the collection of subject tags. Over this period, APO has collected over 5,000 subject terms in its subject repository. The repository, therefore, exhibits ‘folksonomy’ characteristics – it is both rich in user-defined natural language, and lacking in key quality controls that are typical of controlled vocabularies.
The Linked Semantic Platforms LIEF project has resourced further development of APO subject terms into a standards-based, share-able and integrate-able thesaurus. By implementing well-known thesaurus conventions such as hierarchy, synonymy and association relationships, the APO subjects are already improving the discovery experience within the APO platform. We present techniques for improving the semantic and structural integrity of APO subject repository using Artificial Intelligence and conclude by discussing a road-map for a ‘policy thesaurus’, including potential applications and integrations.
Dr Yong-Bin Kang
Dr Yong-Bin Kang is a data science researcher in the Data Science Research Institute at Swinburne with expertise and experiences in the fields of natural language processing (NLP), machine learning and data mining. He completed his PhD in the areas in the Faculty of Information Technology at Monash University, 2011. In his PhD research, he developed a new retrieval algorithm in Case-Based Reasoning (CBR) by enhancing the traditional similarity-based retrieval with data mining techniques. Following the completion of his PhD, he has been actively investigating a wide range of real-world NLP and machine learning problems, and leading various data science research via collaborations within Swinburne and with domestic / international research groups. More recently, his research has been focused on identifying technological trajectories of patent documents by dynamic topic modelling, and semantic interferences on heterogeneous graphs.
Les Kneebone has worked in information management roles in government, school, community and research sectors since 2002. He has implemented metadata, taxonomy and cataloguing products and services in these sectors. Les is currently supporting APO by developing and refining metadata standards and services that will help to link policy research with policy practice.
11.00AM – 12.00PM
Wednesday 18 September 2019
AR104 VR Theatre, Swinburne University of Technology, Hawthorn VIC 3122