Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Browse files

Merge branch 'ferhanture-master'

  • Loading branch information...
commit 25e23fc73409f2a99aec23ac377b78d34cea53e7 2 parents 390d0c6 + 23de4d6
@lintool authored
View
18 src/java/integration/ivory/integration/wikipedia/VerifyWikipediaProcessingCrosslingual.java
@@ -37,19 +37,19 @@
// en side: part 00000, key = 92101
private ImmutableMap<String, Float> enTermDocVector1 = ImmutableMap.of(
- "total", 0.048103902f, "external", 0.004541542f, "time", 0.033348884f, "refer", -0.011754768f);
+ "total", 0.0521711f, "extern", 0.0045928364f, "side", 0.052283954f, "refer", -0.012635737f);
// en side: part 00010, key = 138960
private ImmutableMap<String, Float> enTermDocVector2 = ImmutableMap.of(
- "external", 0.004776824f, "cofound", 0.09919491f, "he", 0.023234092f, "devianc", 0.18071339f);
+ "extern", 0.004580953f, "perspect", 0.097292185f, "deal", 0.07025129f, "devianc", 0.18621536f);
// en side: part 00002, key = 150251
private ImmutableMap<Integer, Float> enIntDocVector1 =
- ImmutableMap.of(35202, 0.034152746f, 34129, 0.054591186f, 27261, 0.039973103f, 34140, 0.08634214f);
+ ImmutableMap.of(35202, 0.033555865f, 34129, 0.053415705f, 27261, 0.039032873f, 34140, 0.08449726f);
// en side: part 00011, key = 184192
private ImmutableMap<Integer, Float> enIntDocVector2 =
- ImmutableMap.of(8777, 0.10271761f, 73827, 0.077015184f, 75933, -0.016551014f, 44992, 0.11264816f);
+ ImmutableMap.of(8777, 0.103838794f, 73827, 0.07746056f, 9147, -0.016266173f, 44992, 0.11387909f);
private static final String dewikiPath =
"/shared/collections/wikipedia/raw/dewiki-20110131-pages-articles.xml";
@@ -58,19 +58,19 @@
// de side: part 00000, key = 1001242228
private ImmutableMap<String, Float> deTermDocVector1 = ImmutableMap.of(
- "auction", 0.00926886f, "total", 0.011755229f, "hors", 0.06490202f, "store", 0.003023784f);
+ "foundat", 0.0034755506f, "external", 0.024032094f, "programm", 0.08090772f, "htv", 0.26859477f);
// de side: part 00010, key = 1000034130
private ImmutableMap<String, Float> deTermDocVector2 = ImmutableMap.of(
- "portray", 0.02833135f, "profession", 0.007643698f, "asund", 0.025962f, "suitabl", 0.02116417f);
+ "ombudswomen", 0.18138693f, "profession", 0.039470334f, "ascrib", 0.016959665f, "great", 0.0019023749f);
// de side: part 00002, key = 1000943946
private ImmutableMap<Integer, Float> deIntDocVector1 =
- ImmutableMap.of(27255, 0.034241054f, 59321, 0.19270006f, 39099, 0.08531962f, 37992, 0.006224899f);
+ ImmutableMap.of(34132, 0.004382227f, 26285, 0.009954007f, 33034, 0.028362243f, 66084, 0.07469488f);
// de side: part 00011, key = 1000347854
private ImmutableMap<Integer, Float> deIntDocVector2 =
- ImmutableMap.of(2110, 0.02419825f, 14287, 0.27120075f, 75805, 0.15010615f, 49109, 0.20328416f);
+ ImmutableMap.of(51505, 0.33379892f, 16336, 0.21256998f, 80281, 0.7150921f, 78262, 0.37479895f);
@Test
public void runBuildIndexEnSide() throws Exception {
@@ -219,7 +219,7 @@ public void verifyIntDocVectorsDe() throws Exception {
new Path(dewikiEn + "/wt-int-doc-vectors/part-00002"), fs.getConf());
reader.next(key, value);
verifyIntDocVector(deIntDocVector1, value);
-
+
reader = new SequenceFile.Reader(fs,
new Path(dewikiEn + "/wt-int-doc-vectors/part-00011"), fs.getConf());
reader.next(key, value);
View
10 src/java/integration/ivory/integration/wikipedia/VerifyWikipediaProcessingMonolingual.java
@@ -34,7 +34,7 @@
// Galago: part 00000, key = 92101
private ImmutableMap<String, Float> galagoTermDocVector1 = ImmutableMap.of(
- "total", 0.036282938f, "posit", 0.047018476f, "valid", 0.07093949f, "formula", 0.06923077f);
+ "total", 0.036282938f, "posit", 0.047018472f, "valid", 0.07093948f, "formula", 0.069230765f);
// Galago: part 00010, key = 34222
private ImmutableMap<String, Float> galagoTermDocVector2 = ImmutableMap.of(
@@ -53,19 +53,19 @@
// Opennlp: part 00000, key = 92101
private ImmutableMap<String, Float> opennlpTermDocVector1 = ImmutableMap.of(
- "extern", 0.002431489f, "zero", 0.054258674f, "theorem", 0.06642f, "prime", 0.04304153f);
+ "extern", 0.0021218103f, "zero", 0.047699325f, "theorem", 0.057848703f, "prime", 0.037537705f);
// Opennlp: part 00010, key = 34222
private ImmutableMap<String, Float> opennlpTermDocVector2 = ImmutableMap.of(
- "direct", 0.07711119f, "titl", 0.08201428f, "congress", 0.1344831f, "soundtrack", 0.15833028f);
+ "direct", 0.07568382f, "titl", 0.080587946f, "congress", 0.131854f, "soundtrack", 0.15528105f);
// Opennlp: part 00002, key = 100984
private ImmutableMap<Integer, Float> opennlpIntDocVector1 =
- ImmutableMap.of(2101, 0.07527498f, 12, 0.023745911f, 3156, 0.0844875f, 15, 0.030846044f);
+ ImmutableMap.of(1, -0.012728759f, 12, 0.024670692f, 3266, 0.084026895f, 5, 0.01582835f);
// Opennlp: part 00011, key = 34222, (terms: conjunto, histori, film, cultur)
private ImmutableMap<Integer, Float> opennlpIntDocVector2 =
- ImmutableMap.of(50365, 0.31359836f, 284, 0.0925163f, 201, 0.0978275f, 419, 0.10097963f);
+ ImmutableMap.of(37730, 0.31801265f, 92316, 0.3315111f, 294, 0.09105158f, 2336, 0.15528107f);
@Test
public void runBuildIndexGalago() throws Exception {
Please sign in to comment.
Something went wrong with that request. Please try again.