We had a great review on Twitter this week from Andrew Gray / @generalising – Librarian for the Antarctic! He highlighted some really good points here on Twitter that we wanted to provide some answers to:

Thank you – we are building on initiatives like I4OC to make the most of openly available data by enhancing and linking it with other sources.

We are trying to strike a positive balance to make data accessible, usable and at the same time making our efforts sustainable.

The main two reasons for removing filters (for now) from the public versions is computational load and performance of the free app and associated operations costs of the free offering and – of course as well the differentiation between the free and paid for offering to get to sustainability.

… you can as a matter of fact also search with an entire piece of text to find similar documents – a little hidden, look out for ‘abstract search’ in the right lower corner when the search box is active.

It is actually not a bug but a feature – since selecting one facet might lead to the other facets to change – it is a ‘funnel approach’ which always presents the facets derived from the data. You need to go facet by facet to ensure that you are not running into zero results due to not fitting facets selected.

We are going to integrate ORCID as an additional data source for disambiguation and will allow researchers to use ORCID to improve their profile. Also the algorithmic disambiguation will be continuously improved, especially to map across the data types.

The suggestion to add affiliations to the author name search is a great suggestion – we will look into that.

Yes, as mentioned – that has to do with the computational load and performance (and costs) for the free application.

Pre-print archives are currently being added – just a matter of time constraints to get this done!

The data is a starting point, but a huge one. We are calling on a larger journal pool since we integrating Crossref data as a backbone and work with publishers to increase the discoverability of their content. So as usual – a different pool of records leads to different citation counts.

Thank you – we will pass it on the team!

Thank you – we will share the feedback with the Unpaywall team.

We are looking into this – stay tuned.

The idea is to provide whitelists and blacklists going forward – we are looking forward to receive suggestions for lists of both kinds!

The machine learning based classification is meant to allow to classify across content type which is obviously not possible with journal classification.

We are going to add features to provide feedback and ability to improve the classification. We already have identified areas where we will work with subject matter experts to improve the training set.

We are aware of the challenge and are discussing with development partners different options. One option could be to add different iterations and the user can select which one is active in the interface to stay consistent – but that goes beyond what we can provide in the free version for now.

Feedback noted… we also got the opposite when in a previous version a paging was still implemented. Will watch feedback in this area.

The download functionality is available – but related to personal user accounts and that the user can create these. We deactivated these for the launch to lower the barrier to use Dimensions and since we are making changes in this space as we implement the institutional single sign on integration.

The explanation: the classification is done on  an abstract level to be consistent across the corpus- and currently only a low number of monographs has an abstract. We are investigating how to overcome this challenge but it will take some time due to too many possible improvements and too little resources.

Thanks very much to Andy for posting such an engaging and insightful review – we really enjoyed reading it and hope these answers are useful for everyone – please keep your comments and feedback coming – via Twitter @DSDimensions, or direct to our team at support@dimensions.ai.