*This is the second in a series of posts from Dr Bernhard Haslhofer, one the recipients of the [DM2E Open Humanities Award](http://openhumanitiesawards.org).*

In my [previous blog post](https://dm2e.eu/open-humanities-awards-maphup-update-1/) I briefly described the two main construction areas we are working on in order to provide semantic tagging functionality as part of the [Annotorious image annotation library](http://annotorious.github.io/). First we need to add GUI elements that display proposed semantic tags to users which allows them to accept or reject tags. Second, we need a backend-service that proposes tags based on textual annotation and context features.

Work on the front-end is making good progress and the [code base](https://github.com/annotorious/annotorious-semantic-tagging-plugin) is growing. Recently we were working on UI design issues with the goal to come with a appealing and intuitive interface. We also published first demos on the [Annotorious Github site](http://annotorious.github.io/demos.html).

At the moment, it is possible to use the Annotorious semantic tagging plugin with a running instance of [Wikipedia Miner](http://sourceforge.net/projects/wikipedia-miner/), which returns relevant Wikipedia articles for given textual input. We use this information to generate semantic tags, which are then proposed on the user interface. Connecting Annotorious with instances of [DBPedia Spotlight](https://github.com/dbpedia-spotlight/dbpedia-spotlight), the [Freebase API](https://developers.google.com/freebase/) or any other commercial named entity recognition engine are possible extension points of the current front-end code base.

Besides connecting the Annotorious semantic tagging plugin with previously mentioned services, we will also continue working on Contextualism (https://github.com/behas/contextualism), which should be a light-weight alternative operating on simple gazeteers or taxonomies. We already defined the interface (see dev branch) and will work on the implementation in the upcoming weeks.