Usability as a success factor in Machine Learning projects

Also in ML projects the usability of the user interface is key to success.

On the IBM Watson Platform, PDF documents can be used for the training of Machine learning (ML) algorithms. For this purpose, a large number of documents need to be annotated manually.

The annotation is an time-consuming work. Making a suitable tool available for this process step is key to success for IBM Watson.



CCS Page Annotator
The page annotator allows the user to mark the various kinds of texts graphically. Thereby, a document model can be established.

On behalf of IBM research, SCS has created a web based tool for this task.

The SCS usability engineer has designed the user interface according to the requirements of IBM. During this process, a special focus was placed on self-explanatory and efficient procedures. Studies at IBM have shown that thanks to this user interface, the time spent on annotation work could be cut by a factor of 10 and more.



CCS Table Annotator
For tables, a dedicated table annotation pagehas been set up.

We congratulate IBM Watson Group, represented by Peter Staar and his team for the presentation of the poster at the SysML conference in Standford, CA.

Peter W J Staar, Michele Dolfi, Christoph Auer, Costas Bekas. Corpus Conversion Service: A machine learning platform to ingest documents at scale. SysML 2018.

Topics: User Centered Design, HTML5, Sass/CSS, TypeScript, Angular, D3



Leave a Reply

Your email address will not be published. Required fields are marked *