Romanian and Catalan/Workplan

From Apertium
Jump to navigation Jump to search

You can find the detailed goals for each week here.

Week Dates Work done Bidix WER / PER Coverage
Post-application period 28 March - 13 May
  • Build frequency lists for Romanian and Catalan
  • Fix broken bidix entries
  • Improve testvoc scripts
12,819 ron > cat (~36%)
cat > ron (~61%)
ron (79%)
cat (78%)
1 14 May - 20 May
  • Add new entries to dictionaries
  • Fix broken bidix entries
  • Fix transfer rules that didn't work as expected
14,890 (~15,000) ron > cat ~33% (~34%)
cat > ron ~56% (~60%)
ron 82.2% (80.1%)
cat 82.7% (79.1%)
2 21 May - 27 May
  • Add new entries to dictionaries
  • Fix broken bidix entries (adj clean)
  • Rewrite ron-cat transfer rules to use chunking
  • Improve freqlist generation script (explains the leap in coverage)
17,026 (~17,000) ron > cat ~31% (~32%)
cat > ron ~53% (~59%)
ron 83.6% (81.1%)
cat 86.2% (80%)
3 28 May - 3 June
  • Add new entries to dictionaries
  • Fix broken bidix entries
  • Plan transfer rule changes
18,487 (~19,000) ron > cat ~31% (~30%)
cat > ron ~53% (~58%)
ron 84.9% (81.9%)
cat 86.4% (80.9%)
4 4 June - 10 June
  • Add new entries to dictionaries
  • Fix broken bidix entries
  • Upgrade transfer rules and write rules for new patterns
  • Add CG to Romanian to improve disambiguation
20,324 (~21,000) ron > cat ~31% (~28%)
cat > ron ~53% (~57%)
ron 85.4% (82.7%)
cat 86.8% (81.7%)
5 11 June - 17 June
  • Add new entries to dictionaries
  • Fix broken bidix entries
  • Upgrade transfer rules and write rules for new patterns
  • Add more Romanian CG rules
  • Improve evaluation of test texts with diff files

First evaluation

22,381 (~23,000) ron > cat ~30% (~26%)
cat > ron ~53% (~56%)
ron 86.1% (83.4%)
cat 87.4% (82.4%)
6 18 June - 24 June
  • Add new entries to dictionaries
  • Fix broken bidix entries
  • Write new transfer rules
22,640 (~25,000) ron > cat ~30% (~25%)
cat > ron ~51% (~53%)
ron 86.4% (84.1%)
cat 88.2% (83%)
7 25 June - 1 July
  • Upgrade Indonesian-Malaysian pair to monolingual package system (testvoc clean)
  • Upgrade Welsh-English pair to monolingual package system
  • Testvoc fixes (Romanian-Catalan and Welsh-English)
8 2 July - 8 July
  • Testvoc fixes (Romanian-Catalan and Welsh-English)
9 9 July - 15 July
  • Upgrade Catalan-Italian pair to monolingual package system (testvoc clean)
  • Testvoc fixes (Romanian-Catalan)

Second evaluation

10 16 July - 22 July
  • Upgrade Afrikaans-Dutch pair to monolingual package system (testvoc clean)
  • Testvoc fixes (Romanian-Catalan and Welsh-English)
11 23 July - 29 July (~27,000) ron > cat (~24%)
cat > ron (~50%)
ron (84.7%)
cat (83.6%)
12 30 July - 5 August (~29,000) ron > cat (~23%)
cat > ron (~47%)
ron (85.3%)
cat (84.2%)
13 6 August - 14 August

Final evaluation

(~31,000) ron > cat (~22%)
cat > ron (~45%)
ron (85.8%)
cat (84.7%)