Can any one suggest what is the BM25 score cutoff in biological text data mining --- Rather what is the perspective / lookout to fix the cut of?? ---- Please help giving a basic cue .. needed little bit urgent. I am stalled in analyzing my data while writing a paper .... plzzzzzz
BM25 is that probabilistic approach to scoring.
Suggested reading
https://berlinbuzzwords.de/sites/berlinbuzzwords.de/files/media/documents/bm25_buzzwords_britta_weber.pdf
Atish,
The default values for Okapi BM 25 k_1=1.2 and b=0.75
The parameter b
controls the level of the normalization of document length in BM 25. The parameter b plays an important role for example when a query of a single term in a document may vary because of the length of the document will have a diluting effect on the term.
The parameter k_1
puts an upper bound of the TF contribution of a single term. A single term in a retrieval should not exceed the match of 2 terms or more
The cutoff
The cutoff in ranking will be tied to the recall and precision established in your domain which you can establish empirically. Both the the parameter b and k_1 will influence the recall and the precision.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining