News
Demonstrator v.2
We have extended the demonstrator of the SubSift matching application. SubSift was originally developed by the Intelligent Systems Laboratory at the University of Bristol as an innovative “submission sifting” application to support academic peer review. The first demonstrator included a pre-defined set of people from the TAILOR project that could be match to arbitrary text. The match was based on the set of previous publications available in their DBLP article repositry.
In this extension, we have included an explanation for each match using Shapley values, and the option to upload your own data to match to free text and obtain an explanation.
The demonstrator can be accessed at the following link. This link may take up to one minute to launch initially.
European Commision's Innovation Radar
The European Commission’s Innovation Radar has highlighted the SubSift tool of WP9 as an excellent innovation (innovation-radar.ec.europa.eu/innovation/51621). WP9 is using AI-powered collaboration tools to facilitate the search for experts in given contexts. SubSift has been used in various TAILOR Conferences and a Summer School to match like-minded attendees. It is currently in an exploratory phase and WP9 is working on extending a demonstrator with additional transparency and explainability functionalities. Future plans involve the integration of SubSift with the AI-on-demand platform (www.ai4europe.eu) to provide AI-powered search functionalities across its catalog.
The demonstrator can be accessed at subsift-webapp.onrender.com/subsift/, which may take up to one minute to launch initially.
D9.1 AI-Driven Collaboration Tools Report
The report for the deliverable 9.1 is now available in the following link.
This report describes the SubSift matching algorithm as an AI-tool that can be used to facilitate the collaboration and networking of researchers. This tool allows to rank a set of people by expertise with respect to a provided text. This can be used to find potential collaborations that were previously unknown, or even to match people with similar interests. In preparation for this deliverable we have applied the algorithm successfully in two TAILOR events, which is described in their respective deliverables and summarised in this report. In this deliverable we provide an online public demonstrator v.1.0 of the tool, which can be used by researchers to find possible collaborators among the TAILOR partner representatives. We expect TAILOR and non-TAILOR members to use the tool to explore the expertise of the TAILOR partners to potentially start new collaborations. Furthermore, in the appendix we present information about a workshop organised by TAILOR which provides additional AI-tools to facilitate the collaboration of AI researchers.
Demonstrator
We have released a demonstrator of the SubSift matching application. SubSift was originally developed by the Intelligent Systems Laboratory at the University of Bristol as an innovative “submission sifting” application to support academic peer review. Within the TAILOR project the application has been extended to match arbitrary text to a pre-defined set of people, based on the similarity of the text to titles of published works in the DBLP article repository.
The demonstrator can be accessed at the following link. This link may take up to one minute to launch initially.