Design of Indicators to monitor Strategic priorities: Mapping Knowledge Clusters 5 th International Workshop Sharing the Best Practices in R&D and Education Statistics - Informing National and Institutional Policy Isabel Reis Lisbon, 9 June 2015
Agenda 1. Framework 2. Project 3. Goals 4. How to do it? How can we delineate priorities? 5. Co-word analysis 6. Methodology 7. First results of the case study: Ocean Economy 8. Problems 9. Next Steps
The problem 1. Definition of indicators to monitor the evolution of the strategic priorities of Portugal 2020 and the ENEI 2. How to delineate the priority? (Priority = theme /multidisciplinary / Problem solving) 3. Indicators are commonly defined by research fields (FOS - international standardization of the fields) We need to match #2 and #3 in order to achieve #1
Project Portugal 2020: monitoring indicators of POs and ENEI - DG REGIO Developing internal competencies Work Group started in Summer 2014 Collaboration with IFRIS-LISIS-ESIEE (Paris-Est Université) in January 2015 Results: in December 2015
Goals 1. Delineation of knowledge clusters associated to each priority - Matrix Priorities/WoS/FOS 2. Design of indicators to monitor the evolution of the strategic priorities of ENEI 3. Quantification of input / output indicators
How to do it? How can delineate priorities? Inspired by nanotechnologies delineation of each priority in Portugal 2020 and ENEI based on bibliometric tools using co-word analysis on WoS.
Co-word analysis Identification of relations between terms and grouping terms into a number of disjoint clusters of related terms The research profile is built by ranking fields according to their size (in terms of number of publications)
Methodology Keywords identification of terms relevant to the priority by experts Building the Query and database - extraction of relevant papers to the priority from the WoS database (tittle, abstract and Keywords) Normalisation/cleaning database (institutions/countries and cities) the main problem related to the retrieval of information from the WoS is the enormous number of unstructured data Analysis of the dataset in CorTEXT digital plataform a) Co-word analysis - identification of interrelationships among these terms using co-word analysis b) Clustering (mapping the structure and the dynamics of the dataset) - The contents of these articles were analysed and organized into thematic clusters through Heterogeneous Networks Mapping. Filtering and refining the query removing terms based on level of generality and speciality Geo-location - enabling us to map how a given priority is distributed over the regions, to see how they evolve in each region and make comparisons with relevant countries. Validation by experts
Case Study: Ocean Economy Preliminary results
Knowledge Cluster Co-occurrence of Terms (300) 2005-2014 Query Ocean Economy WoS
Knowledge Cluster Co-occurrence of Terms (300) 2005-2014 Query Ocean Economy WoS
Knowledge Cluster Co-occurrence of Terms (300) 2005-2014 Query Ocean Economy WoS
Country Cluster Co-occurrence of Countries Co-authorship (300) 2005-2010 WoS
Country Cluster Co-occurrence of Countries Co-authorship (300) 2010-2014 WoS
Country Cluster Co-occurrence of Countries Co-authorship (300) 2010-2014 WoS
Cities cluster Co-occurrence of geo-referenced (300) 2005-2014 WoS
Ocean Economy FoS Minor Code
Problems to be solved Affiliations: rules are necessary Identification of too generic and too specific terms: vessel: cardiology (97) Cleaning the DB to achieve robust results
Next Steps Until now, we have achieved promising results, but improvements can and should be made. Mainly, the keyword-based extraction and clustering could benefit from further refinement. Refining the query - removing terms based on level of generality and speciality. (e.g. Vessel, wave) 2% articles, 21% of fields
Thank you http://www.fct.pt/gabestudosestrategia/ Isabel.reis@fct.pt