This projects aims at grouping utilities to deal with the numerous and heterogenous datasets, as for example in Information Retrieval tasks. For each type of task, it tries to automate/standardize common operations like downloading topics, assessments or evaluating.
- Each task is uniquely identified by an ID, e.g.
- Each task is associated to a definition containing all the necessary information; tasks can be output in JSON
- Resources (e.g. assessments or topics, when available online) can be automatically processed
- Resources can be transformed before being fed to a particular software (e.g. Indri)
- Integrates with the experimaestro experiment manager
Note that this is beta software, in particular the JSON format is still subject to change. Please contact me if you use the software so I can keep you in the loop when doing so.