java - Cosine Similarity of Vectors, with < O(n^2) complexity -
I found this site looking around for similar issues: and this:
However, this run in O (n ^ 2) seems to be clustering some documents and found that this level of complexity was also not possible when dealing with small document sets for dot product, only we vector contained in both vectors Conditions are required, it is possible to put the vectors in a tree and In this way, calculate the dot product with N log n complexity, where N is the number of unique positions 1 of 2 documents
Am I missing something? Is there a Java library that does this?
Thanks
If you store vector elements in a hashtable Do not just log in anyway, do not you? Loop on all the keys in small documents and see if they exist in large ..?
Comments
Post a Comment