Final Report Summary - K-DRIVE (Knowledge Driven Data Exploitation)
The development of innovative technology for the discovery new insights (such as trends) of digital data and real time guidance of using digital data is one of the priorities in the areas of semantic technologies.
The K-Drive (Knowledge Driven Data Exploitation, URL: http://www.kdrive-project.eu/) project focussed on the construction, understanding and exploitation of Knowledge Graph, which is has become popular in knowledge representation and knowledge management applications widely across search engines, biomedical, media and industrial domains. In 2012, Google popularised the term Knowledge Graph (KG) with a blog post titled ‘Introducing the Knowledge Graph: things, not strings', while simultaneously applying the approach in their core business, fundamentally in the web search area. Inspired by the successful story of Google, knowledge graphs are gaining momentum in the World Wide Web arena. Recent years have witnessed increasing industrial take-ups by other internet giants, including Facebook's Open Graph and Microsoft's Satori.
The project Coordinator and Chief Scientist of the K-Drive project is Dr Jeff Z. Pan (University of Aberdeen). There are altogether 25 Marie Curie Fellows from the 5 partners (University of Aberdeen, IBM, iSOCO, Expert System Iberia and Expert System) working in the K-Drive project.
The first objective of the K-Drive project was the identification of novel techniques to understand knowledge graphs, such as query generation techniques to facilitate experts to identify key dimensions within semantic data and to recommend users with related queries. Key results in this strand, in addition to query generation, include knowledge graph simplification, system categorisation of redundancies in knowledge graphs and knowledge graph compression.
The second objective of the K-Drive project was to develop novel reasoning and querying techniques on knowledge graph exploitation, including knowledge graphs with addition streams and deletion streams. Key results in this strand include stream reasoning, stream querying with consistent knowledge discovery over incomplete knowledge graphs, instance retrieval with negative atomic concepts, question answering, handling vagueness and uncertainties in knowledge graphs.
The third objective of the K-Drive project was to develop novel hypothesis and guidance techniques on knowledge graph exploitation. Key results in this strand include TBox learning over incomplete knowledge graphs, vague knowledge extraction from mini-posts, approximate justifications, approximate deduction, reasoner performance and energy consumption predictions, troubleshooting and optimising named entity resolutions.
To the best of our knowledge, the K-Drive project produced the first book on systematic introduction of Knowledge Graph and its applications in Enterprise: ‘Exploiting Linked Data and Knowledge Graphs in Large Organisations‘ (URL: http://www.springer.com/gp/book/9783319456522). Editors of this book include Jeff Z. Pan, Guido Vetere, Jose M. Gomez-Perez and Honghan Wu. The Foreword of the book is written by Chris Welty (Google). This book had over 500 paid downloads within the first week that it was available from the Springer's web site.
The K-Drive project helped organise, among others, the 12th International Reasoning Web Summer School in Aberdeen. A second book has also been published based on the lecture notes of the summer school. This book complements the above first book by focusing on the logical foundation of knowledge graphs: ‘Reasoning Web: Logical Foundation of Knowledge Graph Construction and Query Answering’ (URL: http://www.springer.com/gp/book/9783319494920).
The K-Drive project has attached quite some media coverage, including the collaborations between the University of Aberdeen and IBM on Watson and Knowledge Graph (e.g. URL: http://www.businesscloudnews.com/2016/03/18/ibm-partners-with-aberdeen-university-to-bring-watson-to-medical-research/). Also, the joint efforts between the University of Aberdeen and Expert System on the Brexit prediction have media coverage in UK, US, France and Italy (such as https://www.yahoo.com/news/expert-system-university-aberdeen-64-75-tweets-britain-142700726.html ).
The K-Drive (Knowledge Driven Data Exploitation, URL: http://www.kdrive-project.eu/) project focussed on the construction, understanding and exploitation of Knowledge Graph, which is has become popular in knowledge representation and knowledge management applications widely across search engines, biomedical, media and industrial domains. In 2012, Google popularised the term Knowledge Graph (KG) with a blog post titled ‘Introducing the Knowledge Graph: things, not strings', while simultaneously applying the approach in their core business, fundamentally in the web search area. Inspired by the successful story of Google, knowledge graphs are gaining momentum in the World Wide Web arena. Recent years have witnessed increasing industrial take-ups by other internet giants, including Facebook's Open Graph and Microsoft's Satori.
The project Coordinator and Chief Scientist of the K-Drive project is Dr Jeff Z. Pan (University of Aberdeen). There are altogether 25 Marie Curie Fellows from the 5 partners (University of Aberdeen, IBM, iSOCO, Expert System Iberia and Expert System) working in the K-Drive project.
The first objective of the K-Drive project was the identification of novel techniques to understand knowledge graphs, such as query generation techniques to facilitate experts to identify key dimensions within semantic data and to recommend users with related queries. Key results in this strand, in addition to query generation, include knowledge graph simplification, system categorisation of redundancies in knowledge graphs and knowledge graph compression.
The second objective of the K-Drive project was to develop novel reasoning and querying techniques on knowledge graph exploitation, including knowledge graphs with addition streams and deletion streams. Key results in this strand include stream reasoning, stream querying with consistent knowledge discovery over incomplete knowledge graphs, instance retrieval with negative atomic concepts, question answering, handling vagueness and uncertainties in knowledge graphs.
The third objective of the K-Drive project was to develop novel hypothesis and guidance techniques on knowledge graph exploitation. Key results in this strand include TBox learning over incomplete knowledge graphs, vague knowledge extraction from mini-posts, approximate justifications, approximate deduction, reasoner performance and energy consumption predictions, troubleshooting and optimising named entity resolutions.
To the best of our knowledge, the K-Drive project produced the first book on systematic introduction of Knowledge Graph and its applications in Enterprise: ‘Exploiting Linked Data and Knowledge Graphs in Large Organisations‘ (URL: http://www.springer.com/gp/book/9783319456522). Editors of this book include Jeff Z. Pan, Guido Vetere, Jose M. Gomez-Perez and Honghan Wu. The Foreword of the book is written by Chris Welty (Google). This book had over 500 paid downloads within the first week that it was available from the Springer's web site.
The K-Drive project helped organise, among others, the 12th International Reasoning Web Summer School in Aberdeen. A second book has also been published based on the lecture notes of the summer school. This book complements the above first book by focusing on the logical foundation of knowledge graphs: ‘Reasoning Web: Logical Foundation of Knowledge Graph Construction and Query Answering’ (URL: http://www.springer.com/gp/book/9783319494920).
The K-Drive project has attached quite some media coverage, including the collaborations between the University of Aberdeen and IBM on Watson and Knowledge Graph (e.g. URL: http://www.businesscloudnews.com/2016/03/18/ibm-partners-with-aberdeen-university-to-bring-watson-to-medical-research/). Also, the joint efforts between the University of Aberdeen and Expert System on the Brexit prediction have media coverage in UK, US, France and Italy (such as https://www.yahoo.com/news/expert-system-university-aberdeen-64-75-tweets-britain-142700726.html ).