Information Retrieval Systematic 2 Ranking and Evaluation Efficient List Intersection

It’s important to be able to rank the inverted index like the red font relevance value here.

A simple method is TFIDF.

the parameter b and k here is in IR jargon called BM25 method

To evaluate the ranking, there are several ways. First , precision using P@K

Others are AP(average precision). Discounted Cumulative Gain(DCG)

Inverted Index can be intersected(AND) in industry for retrieving searching result, even merger(OR) is better, but more “expensive” way. This concept is relatively easy so the professor went length to discuss the algorithm efficiency.

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.