Tools
Sorted by:
Try your query at:
 |
 |
 |
 |
 |
 |
Results 21 - 30
of
21,173
Table 13. Bilingual Spanish: experiment groups according to the Tukey T Test.
in WITH
"... In PAGE 15: ...nalysis with respect to the Jarque-Bera test. The value of alpha for this test was set to 5%. The difficulty to transform the data into normally distributed samples derives from the original distribution of run performances which tend towards zero within the interval [0,1]. The following tables, from Table 8 to Table13 , summarize the results of this test. All experiments, regardless the topic language or topic fields, are included.... ..."
Table 13. Bilingual Spanish: experiment groups according to the Tukey T Test.
in WITH
"... In PAGE 15: ...nalysis with respect to the Jarque-Bera test. The value of alpha for this test was set to 5%. The difficulty to transform the data into normally distributed samples derives from the original distribution of run performances which tend towards zero within the interval [0,1]. The following tables, from Table 8 to Table13 , summarize the results of this test. All experiments, regardless the topic language or topic fields, are included.... ..."
Table 2 shows the scores achieved for bilingual tasks. For each couple of languages the same runs developed for monolingual tasks were carried out.
"... In PAGE 7: ... Table2 : GeoClef 2005 officials results for Bilingual tasks The results achieved for the bilingual runs achieved have a similar problem that the result for monolingual tasks; when the retrieved documents are from the English collections our system obtains better results than when the retrieved documents are from the German collections. The Spanish-English and Portuguese-English scores are very similar, whereas the result achieved for German-English task are better.... ..."
Table 2: NLSR language independent products for corpus
Table 3: NLSR language independent product for morphology
Table 5: NLSR language independent product for speech
Table 6: NLSR language independent product for syntax
Table 2. Non-interpolated average precision of bilingual runs (source language English) for all relevant documents averaged over queries.
"... In PAGE 7: ... Retrieval in the lemmatized indexes where compounds were split performed best in all the bilingual runs. In English-Finnish and English-German runs, the next best was the run in the lemmatized compound index, and the stemmed run achieved the worst result (see Table2 and Figures 1, 2 and 3). In the English-German run, the difference between the result of the run in the lemmatized compound index and the result of the stemmed run was only minor: the stemmed run performed only 2.... ..."
Table 2. Test-set error rates for monolingual and bilingual naive classifiers based on smooth n-gram language models in Traveller and BAF.
2005
"... In PAGE 8: ... Results are given in Table 2. From the results in Table2 , we can see that 1-gram language models are similar to our 1-component mixture models. In fact, both models are equivalent except for the parameter smoothing.... ..."
Cited by 2
Table 1. Acoustic data for each language We have trained 8 monolingual recognizers for all 8 corpora (7 languages). Furthermore we have trained an Italian-German-G1 bilingual recognizer (which showed best performance for bilingual recognition), as well as one for Slovak and Slovenian. Further- more, we have trained a trilingual recognizer for the Slavic languages (Sa-Se-Cz). Finally, we have trained a bilingual recognizer for English and German-G2.
1999
Cited by 4
Results 21 - 30
of
21,173
