clean-up-log.text

From LOGICS.EMAP.FGV.BR to WNPT.BRLCLOUD.COM
============================================

Notes:

a) All methods mentioned here are from WORDNET-EDITOR; 

b) "Context" and "graph" are the same terms.  Allegro Graph uses
   "graph", we'll use "context".

High level steps:

1. Fix incorrect IRIs in logics.emap.fgv.br
2. Rename namespaces
3. Fix default graph
4. Process suggestions
5. Deal with blank nodes

Detailed steps:

1. Some IRIs in the logics.emap.fgv.br had spaces in them, which is
   not acceptable to Allegro Graph when loading the files.  To fix
   this, run the method PROCESS-ALL-WORDS-WITH-INVALID-CHARS in
   (clean-up.lisp)

2. After this is done, export all triples with context information
   using the method EXPORT-TRIPLES (export.lisp)

3. Execute the script rename-namespace.sh

4. Re-import the files preserving the context information as follows:

   agload -g "<file://own-pt.nt>" wn30 nomlex.nt
   agload -g "<file://wnlite.nt>" wn30 wnlite.nt
   agload -g "<file://own-pt.nt>" wn30 own-pt.nt
   agload -g "<file://wordnet-en.nt>" wn30 wordnet-en.nt
   agload wn30 default-graph.nt

5. At this point we have the triple-store with the same contents of
   the original LOGICS.EMAP.FGV.BR triple-store, but with proper
   namespaces and context information.

6. Process suggestions from the wnpt.brlcloud.com/wn site: execute
   PROCESS-SUGGESTIONS (suggestions.lisp)

7. Make sure none of the newly added suggestions were added to the
   default graph:

   (length (get-triples-list :g (default-graph-upi *db*))) should be 0.

8. Fix the remaining default-graph issues by executing
   FIX-DEFAULT-GRAPH (clean-up.lisp) and
   fix-remaining-blank-nodes-in-default-graph.

9. We need to clean the literals of the synsetId and tagCount
   properties (they need to be xsd:nonnegativeInteger).

   # Run: (FIX-INCORRECTLY-TYPED-LITERALS) on file CLEAN-UP.LISP

10. Fix all blank word senses, by executing
    PROCESS-ALL-BLANK-WORDSENSES

11. Fix all blank word nodes, by executing PROCESS-ALL-BLANK-PT-WORDS
    and PROCESS-ALL-BLANK-EN-WORDS.

========================================================
= After verification is complete, these need were done =
========================================================

12. New wn30.turtle was provided: need to replace the one in the
    triple-store (context <file://wnlite.nt>

    # First, execute the following SPARQL:
    # drop graph <file://wnlite.nt>

13. Next, imported wn30.ttl from the openWordnet-PT with the
    proper context