Fix for NUTCH-2245 NGram Model for Cosine Similarity by bhavyasanghavi

[bhavyasanghavi]

No description provided.

[bhavyasanghavi]

Description:

1. The user should enable the plugin by enabling scoring-similarity in the plugin.includes property in nutch-site.xml.
2. Copy the gold-standard file into the conf directory and enter the name of this file in nutch-site.xml.
3. The user should include the following property in nutch-site.xml to specify the window size for ngram model:
   <property>
<name>scoring.similarity.ngrams</name>
<description>Specifies the 'n' in ngrams</description>
</property>
4. If the user does not include the above mentioned property, the default size will be 1, i.e unigrams will be considered for similarity scoring.

Use Case:
It is extremely useful in cases where the user specifies names entities in the gold standard. For example "Leonardo DiCaprio" is one of the terms in gold standard. The ngram model with a window size 2, will provide distinction between pages pertaining to "Leonardo da Vinci", which otherwise would not be captured. Another example is where user is looking for phrases/dialogues, the window size can be cleverly adjusted depending on the content of the gold standard.

[lewismc]

Please use correct effficient slf4j code notation here
e.g. LOG.info("Value of ngram: {} ", ngram);

[bhavyasanghavi]

Yes, I will update that. Thanks.

[lewismc]

Overall, the patch it looking excellent.

[sujen1412]

This seems like its used for debugging, please change it to LOG.debug(). It helps keeping the log clean.
Thanks!

[bhavyasanghavi]

Updated it. Thanks.

[sujen1412]

I have tried out the patch, works as expected, going to commit now.
Thanks @bhavyasanghavi!