[DataScience] LTR(Learning To Rank)

TF-IDF와 BM25가 검색결과를 도출하는 하고 상관없는 문서를 빼는것에 초점을 둔 반면 LTR은 문서의 정렬을 하는 것에 초점을 두고 있다
TF-IDF와 BM25와 같은 알고리즘의 고도화로 불필요한 doc는 제거가 잘되고 있으니, 무엇을 먼저 보여줄것인가로 관점이 이동했다고 보면 된다

1. Create Judgement List
2. Define features for the model
3. Log features during usage
4. Training and testing the model
5. Deploying and using the model
6. Feedback loop

grade,keywords,movie
4,Rambo,First Blood     # Exactly Relevant
4,Rambo,Rambo
3,Rambo,Rambo III       # Fairly Relevant
3,Rambo,Rambo First Blood Part II
2,Rambo,Rocky           # Tangentially Relevant
2,Rambo,Cobra
0,Rambo,Bambi           # Not even close...
0,Rambo,First Daughter

grade,keywords,docId
4,rambo,7555
3,rambo,1370
3,rambo,1369
4,rocky,4241

# grade (0-4)queryiddocIdtitle
#
# Add your keyword strings below, the feature script will
# Use them to populate your query templates
#
# qid:1: rambo
# qid:2: rocky
# qid:3: bullwinkle
#
# https://sourceforge.net/p/lemur/wiki/RankLib%20File%20Format/
#
#
4qid:1 #7555Rambo
3qid:1 #1370Rambo III
3qid:1 #1369Rambo: First Blood Part II
3qid:1 #1368First Blood
0qid:1 #136278Blood
0qid:1 #102947First Daughter
0qid:1 #13969First Daughter
0qid:1 #61645First Love
0qid:1 #14423First Sunday
0qid:1 #54156First Desires
4qid:2 #1366Rocky
3qid:2 #1246Rocky Balboa
3qid:2 #60375Rocky VI
3qid:2 #1371Rocky III
3qid:2 #1375Rocky V
3qid:2 #1374Rocky IV
0qid:2 #110123Incredible Rocky Mountain Race
0qid:2 #17711The Adventures of Rocky & Bullwinkle
0qid:2 #36685The Rocky Horror Picture Show
4qid:3 #17711The Adventures of Rocky & Bullwinkle
0qid:3 #1246Rocky Balboa
0qid:3 #60375Rocky VI
0qid:3 #1371Rocky III
0qid:3 #1375Rocky V
0qid:3 #1374Rocky IV

POST tmdb/_search
{
  "query": {
      "sltr": {
              "params": {
                  "keywords": "rambo"
              },
              "model": "my_model", <- 모델 지정
          }
  }
}

POST tmdb/_search
{
  "query": {
      "match": {
          "_all": "rambo"
      }
  },        
  "rescore": {   # 결과값 도출 이후 상위 1000개의 대해서만 재정렬 진행
      "window_size": 1000,
      "query": {
          "rescore_query": {
              "sltr": {
                  "params": {
                      "keywords": "rambo"
                  },
                  "model": "my_model",
              }
          }
      }
  }
}