iris

In indri, query likelihood model is compeleted by "#combine(term1 term2)", so what's the difference between "term1 term2" and "#combine(term1 term2)"? In galago, is query likelihood model also compeleted by "#combine(term1 term2)"? And, how to use galago to achieve the MLE method?

Thank you very much, Michael. I find the function. according to the result, the first column is word, the second column is term frequency in the collection, the third column is document count, is that right? And can you explain the meaning of extent , posting, and field ? It confuses me a lot after I created the apendix.

How to get tf or idf of a term with galago? Every time I run the command line — batch-search, I can see the term frequence in the screen, But how can I get the frequence of a term in my code rather than in output screen,so that it can be computed in my way?

After browsing the pages about galago, I know there are two ways to use galago, one is using command line, the other is writing Java codes. Here is the problem. Can I complete all galago’s operators using command lines without writing even one line code of Java?

Can anyone tell me the right way to use galago? From the document, I know there are two ways to use galago. One of them is command line, the other is to write Java codes. So, here is the problem. Is it possible to complete all opertators with command line? For example, if I want to use sdm model, the thing I need to do is put #sdm before query in the json file for configurating parameters, right? #sdm(white house) Then it will return the result by sdm model. Or, I can only write Java codes to call...

After browsering relevant pages of galago. I find there are two ways to use galago. One of them is using command lines, another is writing Java codes. Here is the problem. Can I complete all galago's operators using command lines without writing even one line of Java codes?

θq: the update query model θo: the original query model, which is estimated with MLE without smoothing, θR: the relevance model

query: milosevic mirjana markovic term：plan "queries": [{ "number": "423", "text": "#combine:0=0.2:1=0.8 (#combine(#stopword(milosevic mirjana markovic )) plan)” " }] I found a formula in a paper-- the relevance model is used to enhance the original query model by the following intrerpolation: P(w|θq )=(1-λ) P(w|θo )+ λ P(w|θR ) Is it equal to set the weight in #combine ? In other ways, I want to impletate the formula above, so I set λ=0.8. Then I write the text as "#combine:0=0.2:1=0.8 (#combine(#stopword(milosevic...