diff --git a/Report.md b/Report.md index 58da43f856da3e447cf671ac69d531799ce86103..7396223959751ef81bb92ea601364877abdc831d 100644 --- a/Report.md +++ b/Report.md @@ -83,6 +83,16 @@ You can see the output file [here](https://gitlab.my.ecp.fr/2014guom/BigDataProc All the details are written in my code [Preprocess.java](https://gitlab.my.ecp.fr/2014guom/BigDataProcessAssignment2/blob/master/output/Preprocess.java). ## Set-similarity joins +> You are asked to efficiently identify all pairs of documents (d1, d2) that are similar (sim(d1, d2) >= t), given a similarity function sim and a similarity threshold t. Specifically, assume that: + each output line of the pre-processing job is a unique document (line number is the document id), + documents are represented as sets of words, + sim(d1, d2) = Jaccard(d1, d2) = |d1 Ո d2| / |d1 U d2|, + t = 0.8. + +For this part, I can't use directly 'pg100.txt' with 124787 lines because it will take too much time while finding similary documents. So I made a sample where I chose the first ten sonnets with total 174 lines. +You can find the sample text [here](https://gitlab.my.ecp.fr/2014guom/BigDataProcessAssignment2/blob/master/input/pg100_Sample.txt) and the output file after preprocessing [here](https://gitlab.my.ecp.fr/2014guom/BigDataProcessAssignment2/blob/master/output/output_preprocess_sample). + +### Naive Approach + -For this part, I can't use directly 'pg100.txt' with 12