SORT
VOTES
CREATED
metamitya
METAMITYA
1
36:37 7
we're in the dark ages of search relevancy, now entering the age of enlightenment
metamitya
METAMITYA
1
26:39 7
lists features we can use for learning to rank, including comparative signals which i believe is a feature that describes eg. similarity between a query and a document.
metamitya
METAMITYA
1
25:03 7
building a relevance ml model. reminds me of the learning to rank video. mentions semantic cohesion, reminded me of word2vec. use model via a lucene custom operator.
metamitya
METAMITYA
1
22:26 7
predictive analytics: search engine rankings not based on math or statistics. tf/idf is a heuristic based ranking.
metamitya
METAMITYA
1
19:57 7
graph of iterative engine score improvement, with blog mentioned: http://www.searchtechnologies.com/blog/search-engine-relevance-scoring
metamitya
METAMITYA
0
39:13 7
pulling scores out of solr like tfidf
metamitya
METAMITYA
0
29:27 7
demo
metamitya
METAMITYA
0
27:38 7
thresholding. meaning if the probability of relevancy is low you can retry the query before returning results to users!
metamitya
METAMITYA
0
18:35 7
offline engine analysis. the search algorithm becomes the independent variable passed to a function which runs a static set of queries against a static index and outputs a score.
metamitya
METAMITYA
0
16:03 7
score each query from the perspective of each user