i used the binary version on windows the first version is galagosearch version 1.0.4 the command line galago eval TF_IDF_baseline TF_ID-2 qrels.txt >TF-type2 The second one for galago3.10 and also 3.13 the command C:\galago-3.13-bin\bin>galago eval --judgments=qrels.txt --baseline=Hiemstra_baseline --treatment=Hiemstra_comb > stat-comb
Hi all I used galago to do thesignificance test i use the version 1.0.4 i have this result for the MAP wich seems correct (not significant with Ttest) averagePrecision baseline 0,3681 averagePrecision treatment 0,3659 averagePrecision basebetter 31 averagePrecision treatbetter 41 averagePrecision equal 0 averagePrecision ttest 0,7445 averagePrecision signtest 0,1444 But when i used galago 3.10 here the result is significant map baseline 0,3681 map treatment 0,3659 map baseBetter 31,0000 map treatBetter...
Thanks
Hi all, I indexed my documents collecttion using Porterstmmer with Indribuildindex, for the retireval, do I must specify that the queries must be stemmed? if yes, it would be in command line with <setmmer><name>Porter</name></stemmer> Regards
Hi, In the file src/Indrilang.g we can find this command "charVocabulary = '\u0001'..'\u00ff'; // UTF-8 format ", this line mean that it can handle french because for example the caracter "è" is" U+00E8" So my question : Is it Indri able to index fench caracter like é è à...? Regrads.
Thank you
Than you very much
Hi Mr Fisher, i have some questions: 1) if we don't specify feedback paramters, the results obtained by indri still of the model RM3? Regards
Hi all, I know that the TRECformat is the the format accpeted by indri ,and it allow indexing the text between <TEXT>. 1- Now if my collection contain other tags, for exemple < TX>, is it INDRI able to index this collection without specifying the TX as field? 2.If my collection contain many fields <TX>, <Title>, it is possible to do retrieval without using field retrieval? for example by simply using #combine( text ....query) Regards
Thank you very much Mr Fisher
Hi all, I want to know how to specify the paramter for the RM3 to do query expansion with Indri or Lemur, because the only indication for Indir is the paramter <fbOrigWeight>number</fbOrigWeight>. what is the number to specify that allow to use RM3 Thank you
thank you
Hi, is there an implemntation of the relevance model created by Dr. Victor lavrenoko?...
Thank you David In my case i indexed unigram and bigram, and i want a mixture of...
Thank you Craig In my case i indexed unigram and bigram, and i want a mixture of...
Hello I want to use this formula with indri or lemur P(Q|Q)=(alpha)P(t|D)+(1-alpha)...