CCASH: Cost-Conscious Annotation Supervised by Humans

See CCASH Setup for instructions on how to setup CCASH for development.

• 0.1 - Demoed in the ALFA meeting on 10 Dec 08

See CCASH Schedule for current goal deadlines.

• English user-study with AL in the loop
• Syriac user-study with AL in the loop

• Format for storing additional lexicon information

• Dictionary v. no dictionary
• Sentence v. word v. other granularity
• Waiting: human or machine
• Modes of keyboard interaction
• Dictionary: building from scratch or pre-formed
• Scope of annotation: project vs. document

This is the wishlist of features we will use to create version 0.2's target feature set. Right now this is a fluid pool. Please add things. We can decide which features stay in future meetings.

• High Priority
  • projects created
  • complete work flow for a document: it may be assigned to a user, annotated, reviewed, and marked complete.
  • user administration panel: create, assign roles, delete
  • project admininstration panel: create, assign to users, delete
  • link to dictionary entries in the annotation lens
  • create new dictionary entries from words during the annotation process
  • deal gracefully with English's tagset (single category, 45 possible values)

• Low Priority
  • security (password hashing, email verification)

• No Assigned Priority…yet
  • active reviewing - just seeing conflicts not everything?
  • token correction - by reviewer or annotator, token was input incorrectly into database, etc.
  • annotator comments to reviewer - about documents, tokens, any granularity
  • annotation color codes - yellow=computer, blue=human, green=reviewed/approved, red=problem
  • subversion style annotating - each annotation is a revision/multiple annotations per word
  • change database engine to innodb