Recommender System with Apache Spark Apache Kafka and Elasticsearch
ApacheBigDataEU16-NPentreath
ApacheBigDataEU16-NPentreath
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>Elasticsearch</strong><br />
Custom scoring function<br />
Scoring<br />
• Native script (Java), compiled for speed<br />
• Scoring function computes dot product by:<br />
§ For each document vector index (“term”), retrieve<br />
payload<br />
§ score += payload * query(i)<br />
• Normalizes <strong>with</strong> query vector norm <strong>and</strong><br />
document vector norm for cosine similarity