In the Information Retrieval & Web Search course I'm taking, we look at an algorithm for document-at-a-time scoring called MaxScore (also written as max_score, maxscore or max score in the literature). Now MaxScore is a pretty cool algorithm, but unfortunately you wouldn't know that, because to learn MaxScore you have to piece together the ideas behind the technique from half a dozen hand-waved explanations in various academic papers.
This is a great loss to the recreational programming community, so I've written (to my knowledge) the first thorough and in-depth explanation of MaxScore. It includes a general outline, examples, and implementation details, as well as a detailed explanation and proof of why it works. Hopefully it will be useful to others.
No comments:
Post a Comment