Home
By Know-Center: Know-Center

Matchbook is a new open science application that provides tailored recommendations for new scientific collaborations based on information about research projects and institutions indexed by  OpenAIRE . It enables researchers to systematically seek out collaborators within specific research areas based on their previous success in securing project funding within those domains and on the previous collaboration history of institutions. Users can search across information linked to all funders indexed by OpenAIRE, or filter by specific funding organisations.

Matchbook works by ingesting data regarding funders, projects and project partners from the OpenAIRE API and applying customised algorithms to create tailored recommendations for individual users based on their preferences and history. The recommender system is built upon Know-Center’s highly flexible and scalable recommender framework ScaR . Recommendations can be returned based on three criteria:

1. Keyword-Filtering: This algorithm recommends organisations for given keyword(s) (e.g., data science) entered in the text field below. Therefore, we search the titles and keywords of all OpenAIRE projects and recommend the organisations that were involved in most matching projects based on the term frequency–inverse document frequency. (See Wikipedia entry on tf-idf score ).

2. Collaborative-Filtering: This algorithm recommends organisations based on project collaborations for a given organisation id entered in the text field below. Thus, we search for similar organisations by means of common projects and ranks/recommend according to the organisations that were involved in most projects as the given organization. (See Wikipedia entry on Collaborative-Filtering ).

3. Content-based Filtering: This algorithm recommends organisations based of project metadata for a given organization id entered in the text field below. The means that we search for similar organisations by matching the titles and keywords of involved projects based on the TF-IDF score (See Wikipedia entry on Content-based Filtering ).

Matchbook functionality can also be embedded on other websites using a few simple lines of JavaScript code. For example, individual funders indexed by OpenAIRE may wish to enable searching just on projects and institutions connected to their own funding information to easily create a local collaboration recommender system. Any parties interested in using Matchbook to embed such functionalities in their own sites are welcome to contact us for advice and assistance.

For any questions or feedback, including advice on re-using Matchbook functionality, please contact the Matchbook project lead Tony Ross-Hellauer:tross@know-center.at