Contents
PundIt general info
Can be found at the tool's web-page: PundIt.
PundIt workshop, den Haag 23/05/2013
This is a short summary of the PundIt workshop on 23/05. The slides of the presentations can be found on the workshop's web-page, http://dm2e.eu/pundit-workshop-at-the-national-library-of-the-netherlands/.
Users of the tool. "Europeana cloud" can be one of them. Currently they consider if the tool suits their aims.
License. Open source since 1 month.
OA-compatibility. Not 100%, the plan is to make it 100% though.
An RDF-triple is the only schema of any PundIt annotation, which must have the form "subject-predicate-object". The workshop audience was rather critical about it. Indeed, you really need to get used to the way document fragments are annotated. While simple annotations like "text A"-"created by"-"author B" can be created and saved quickly, a more involved relation needs a specific workaround.
In the future, the developers plan "to hide the triples" behind user friendly composite predicates. In particular they consider adding variables to be able to express something similar to search queries, e.g. "text X created by Marx where the creation date of X is between 1830 and 1840" (not sure about exact formulation).
Vocabularies for predicates: are provided or can be configured by the administrator. The client requests vocabularies e.g. via KORBO api.
Administrator. Both, the client and the server, parts of PundIt need administrating. The administrator needs to configure and update vocabularies for the client and to configure the servers where annotations are saved.
Server. At the moment the tool is well-thought as a one-server system. Connecting multiple servers is in the future plans. One of the foreseen problems is export/import of notebooks, with annotators registered on one server and not registered on another.
Notebooks. At the moment you cannot copy an annotation from one notebook to another.
Authentication/sharing. An authentication can be done via an openID, a google or yahoo account, or the credentials are created by an annotation server. A user can create either a fully personal or a fully public notebook. No group notebooks are possible at the moment. The workshop audience was not happy about that. A workaround: to fake a group you make a single account for a group of people. In this case, however, it would not be clear who is annotating what, unless the participants make an agreement how each individual marks his/her annotations.
Named-content methodology. You may annotate not only a web-document but its content, if its web-publishers agrees to provide a content vocabulary. For instance, let there be 2 images of the same person in a document. If the publisher allows to mark them by the same content (the name of the person) then any annotation on one image is an annotation on the other one and vice versa. Entities with the same content does not need to belong to the same document (???). In the same way you create linking between a high- and a low-resolution versions of the same image.
Another situation where content annotating can be used is resolving an annotation for a dynamically updated document (in some cases). For instance, if an image or a text fragment are annotated as content then when they are moved from one place to another their annotations will be still resolvable.
Annotating image. You can annotate an image part as well, by surrounding it by the corresponding polygon.
Visualisation. In principle it is possible to connect PundIt to some fancy visualising tool. There is a demo example where a participant of the workshop creates two annotations of the form "text A is created by a philosopher B", and "text A cites a philosopher C". The visualisation tool collects the annotations from all the participants and makes a time-line in which "B is influenced by C" connections can be seen for all B and C used by the participants of the workshop to create their triples. Technically, connecting to a visualisation tool needs an intermediate layer which sends the necessary requests to the annotation server and transforms the returned values to the data for the visualisation tool.
UI. Can be better. The PundIt developers want to make it user-friendlier, e.g. by adding help, etc.
