TMDB Score shaping

Get Started. It's Free
or sign up with your email address
Rocket clouds
TMDB Score shaping by Mind Map: TMDB Score shaping

1. Untitled

2. signal modeling

2.1. analysis process

2.1.1. Untitled

2.2. ultimate goal of analysis is striking a balance between precision and recall

2.2.1. Untitled

2.3. how to improve recall

2.3.1. stemming

2.3.1.1. Untitled

2.3.2. synonyms

2.3.2.1. specificity

2.3.2.2. capturing meaning

2.3.2.3. acronyms

2.3.2.4. Untitled

2.3.3. edge grams

2.3.3.1. Untitled

2.4. how to improve precision

2.4.1. token filter to remove stop words such as OR, AND

2.4.1.1. Untitled

2.4.2. shingles

2.4.2.1. Untitled

3. ranking func design

3.1. multifield search

3.1.1. multi-match

3.1.1.1. best_fields

3.1.1.1.1. tie_breaker

3.1.1.2. most_fields

3.1.1.2.1. coord

3.1.1.3. cross_fields

3.1.2. match_phrase

3.2. field-centric / term-centric

3.2.1. default scoring DF * ITF

3.2.1.1. Untitled

3.2.2. problem

3.2.2.1. search multiple terms in multiple fields and give out an score

3.2.3. two basic solutions

3.2.3.1. Untitled

3.2.4. field-centric search

3.2.4.1. two forms

3.2.4.1.1. best_fields

3.2.4.1.2. most_fields

3.2.4.2. baked into multi_match

3.2.4.2.1. best_fields

3.2.4.2.2. most_fields

3.2.5. field-centric problems

3.2.5.1. albino elephant

3.2.5.1.1. def

3.2.5.1.2. sample albino elephant

3.2.5.1.3. albino elephant in start trek

3.2.5.2. signal discordance

3.2.5.2.1. def

3.2.5.2.2. star trek example

3.2.6. term-centric search

3.2.6.1. dismax-style

3.2.6.1.1. Untitled

3.2.6.1.2. benefits

3.2.6.1.3. downsides

3.2.6.1.4. fine-tune

3.2.6.2. solve signal discordance

3.2.6.2.1. custom all_fields

3.2.6.2.2. cross_fields

3.2.6.2.3. cross_fields vs custom all_fields

3.2.7. combine field/term-centric search

3.2.7.1. like fields together

3.2.7.1.1. group "like fields" together

3.2.7.1.2. limits of like fields

3.2.7.2. combine greedy naive search / conservative amplifiers

3.2.7.2.1. two factors

3.2.7.2.2. Untitled

3.3. shaping the relevance function

3.3.1. functional query

3.3.2. boolean query

3.3.2.1. via Boolean clause

3.3.3. example

3.3.3.1. problem

3.3.3.1.1. Untitled

3.3.3.2. reason

3.3.3.2.1. TF*IDF has a strong bias towards shorter fields through field normalization

3.3.3.3. solution

3.3.3.3.1. additive boosting

3.3.3.3.2. multiplicative boosting

3.3.3.3.3. filtering

3.4. movie search ranking rules

3.4.1. scoring layers

3.4.1.1. Untitled

3.4.2. Untitled

3.4.2.1. exact name matching

3.4.2.1.1. def

3.4.2.1.2. boolean boost on exact-title matching

3.4.2.1.3. add a clause for big rammed matches

3.4.2.2. another two layers

3.4.2.2.1. Untitled

3.4.2.3. final complete query

3.4.2.3.1. Untitled