Since implementing the whole plugin during the hackfest is a not realistically achievable I tried to break it down into smaller tasks:
create a barebone plugin, packaged according to the OJS plugin format
for the time being the enriched HTML will not be stored (but will need to be done in the long run) meaning that every time a journal paper is viewed the automatic extraction will be performed
webservice hook: the communication between the plugin and the 3rd party NER web service has been almost implemented. I managed to have a PHP class which send a call to the web service and prints the JSON output.
the automatic markup of the html may be tricky: it is needed tokenisation plus keep the position (start/end) for each token in the string. best to do this at a later stage. I will start with
containing pure text
templates to translate JSON into HTML ** what is being used in OJS for templating? ** with 1 to 1 correspondence between IOB entity and template: if entity=["LOC"] in the config file, then there shold be a LOC.tpl in the template directory and so on.
display inline citations and a sort of summary on the right hand side
A first implementation is expected to:
- install properly with config file
- for each paragraph in the article: ** call webservice; ** run output through template and push it into HTML; ** display summary on right hand side
== Questions ==
- what's best recommended workflow with GIT when working on OJS
- any plugin file structure to get started?
- which template to change to display citation on the right
- what's the best way to develop-test OJS