Persian-English Tagged Test Corpus

Glass-Box Evaluation Sample

===== Component Summary =====

Name: CRL.component.per.DictionaryLookup.SingleWordLookup
Number of documents: 200
Average words per document: 21
Average total speed per document: 28101ms
Average total speed per word: 1311ms
Number of corpus annotations: 3766
Number of annotations produced by system: 13220
Number of matching annotations (matching spans and matching structure): 5517
Number of annotations with matching spans and non-matching structure: 3322
Number of annotations with non-matching spans: 4381
Recall: 89 %
Precision: 25 %

 ===== Document Diagnostic =====

Document 1
Content
dvrh byk|rs|zy gstrdh dr alm|n p|y|n my~y|bd

[0,4] {dvrh} (3,0,0)
Matches
1. per.Noun "period, course"  Key values: {dvrh}
2. per.Noun "gathering"  Key values: {dvrh}
3. per.Noun "review"  Key values: {dvrh}

[5,14] {byk|rs|zy} (1,0,0)
Matches
1. per.Noun "unemployment"  Key values: {byk|rs|zy}

[15,21] {gstrdh} (2,3,0)
Matches
1. per.Adjective "spread"  Key values: {gstrdh}
2. per.Adjective "widespread"  Key values: {gstrdh}

Non-matching structures
1.
[diff1 : per.Entry[gram.pos : per.Adjective],
 diff2 : per.Entry[gram : per.Grammar[pos : per.Verb]]]
Sense: "lay, spread"  Key values: {gstr} {gstrdn}
2.
[diff1 : per.Entry[gram.pos : per.Adjective],
 diff2 : per.Entry[gram : per.Grammar[pos : per.Noun]]]

Back to Bilingual Tagged Corpus