Getting Answers to Fuzzy and Flexible Searches by Easy Modelling of Real-World Knowledge Victor Pablos-Ceruelo and Susana Mu˜ noz-Hernandez susana@fi.upm.es Universidad Polit´ ecnica de Madrid (T.U. of Madrid), Spain FCTA’2013, Vilamoura, Algarve, Portugal (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 1 / 29
Overview • Motivation − Crisp Databases − Fuzzy Databases − Real Databases − Expressive Queries − Limited Queries • Goal • Query Framework − Query Components − Query Structure − Basis − Semantics & Syntax − Syntactical Constructions • Present work • Conclusions (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 2 / 29
Overview • Motivation − Crisp Databases − Fuzzy Databases − Real Databases − Expressive Queries − Limited Queries • Goal • Query Framework − Query Components − Query Structure − Basis − Semantics & Syntax − Syntactical Constructions • Present work • Conclusions (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 3 / 29
Crisp Databases Restaurants table name distance price avg. food type Il tempietto 100 30 italian Tapasbar 300 20 spanish Ni Hao 900 10 chinese Kenzo 1200 40 japanese (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 4 / 29
Fuzzy Databases Restaurants table name close cheap similar to mediterranean Il tempietto 1 0.5 0.8 Tapasbar 0.9 0.8 0.9 Ni Hao 0.4 1 0.1 Kenzo 0.1 0.2 0.2 (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 5 / 29
Real Databases • Crisp values (traditional) • Easy update values • Fuzzy concept personalization • Less number of fields • Data objectivity (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 6 / 29
Expressive Queries I’m looking for a ... • restaurant of mediterranean food or similar and not far from the city center • cheap flat of two rooms very close to the beach (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 7 / 29
Limited Queries I’m looking for a restaurant or a flat • mediterranean food • 2 rooms • cheaper than 30 euros menu (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 8 / 29
Limited Queries I’m looking for a restaurant • food type = mediterranean • number of rooms = 1 • price � 30 euros (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 9 / 29
Overview • Motivation − Crisp Databases − Fuzzy Databases − Real Databases − Expressive Queries − Limited Queries • Goal • Query Framework − Query Components − Query Structure − Basis − Semantics & Syntax − Syntactical Constructions • Present work • Conclusions (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 10 / 29
Goal • Defining a flexible and expressive query structure • Providing a sound framework with a search engine of these queries over traditional databases (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 11 / 29
Overview • Motivation − Crisp Databases − Fuzzy Databases − Real Databases − Expressive Queries − Limited Queries • Goal • Query Framework − Query Components − Query Structure − Basis − Semantics & Syntax − Syntactical Constructions • Present work • Conclusions (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 12 / 29
Query Components I’m looking for a ... • restaurant of mediterranean food or similar and individual value nfp co not far from the city center negation fp • cheap flat of two rooms very close to the beach fp individual value nfp q fp (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 13 / 29
Query Components • individuals (restaurant, film, house, ...) • comparison operands co (equal, distinct, greater, less, similar, ...) • fuzzy concepts fp (big, cheap, close to the beach, ...) • quantifiers q (quite, rather, much, very, little, ...) • crisp concepts nfp (prize, size, distance, food type, ...) • values (30000, 3, mediterranean, comedy, ...) (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 14 / 29
Query Structure I ′ m looking for a / an individual � � not q fp (1) AND whose nfp co value (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 15 / 29
Basis RFuzzy (very expressive fusion information tool) • Multi adjoint Logic • Over Prolog (distributed computation, constraints, finite domains, ...) • Sugar Syntax • Similarity • Types • Combine crisp and fuzzy information • Incomplete information (default values, conditioned) (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 16 / 29
Semantics and Syntax • Framework Semantics ← RFuzzy Semantics • Framework Syntax ← defined using RFuzzy predicates • Fuzzy concepts ← fuzzyfication of crisp values of nfp • Similarity ← explicit or advance definition • Configuration file ← fp definition (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 17 / 29
Syntactical constructions Database definition define database ( pT / pA , [( pN , pT ′ )]) (2) define database ( restaurant / 4 , [ ( id , string type ) , ( food type , enum type ) , ( years since opening , integer type ) , ( distance to the city center , integer type )]) . (3) (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 18 / 29
Syntactical constructions Similarity Definition similarity between ( pT , pN ( V 1) , pN ( V 2) , TV ) (4) with credibility ( credOp , credVal ) (5) only for user ′ UserName ′ (6) similarity between ( restaurant , food type ( mediterranean ) , food type ( spanish ) , 0 . 8) (7) (p , v) , & i similarity ( pT ( pN ( V 1 , V 2))) if COND (8) TV ← − − − − − − (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 19 / 29
Syntactical constructions Truth value Definition fPredName ( pT ) : ∼ value ( TV ) (9) if ( pN ( pT ) cond value ) . (10) cheap ( restaurant ) : value (0 . 1) if ( id ( restaurant ) is equal to zalacain ) . (11) (p , v) , & i fPredName TV if COND (12) ← − − − − − − (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 20 / 29
Syntactical constructions Fuzzification Definition fPredName ( pT ) : ∼ function ( pN ( pT ) , [( valIn , valOut )]) (13) traditional ( restaurant ) : function ( years since opening ( restaurant ) , [(0 , 0) , (5 , 0 . 1) , (10 , 0 . 4) , (15 , 1) , (100 , 1)]) . (14) (p , v) , & i fPredName ( valIn ) OP if COND (15) ← − − − − − − ( valOut 2 − valIn 1) OP = valIn (16) ∗ ( valIn 2 − valIn 1) COND ′ = ( valIn 1 < valIn < valIn 2) (17) (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 21 / 29
Syntactical constructions RFuzzy Rule Definition fPredName ( pT ) : ∼ rule ( aggr , complexBody ) (18) fPredName ( pT ) : ∼ rule ( simpleBody ) (19) (p , v) , & i aggr ( complexBody ) if COND (20) fPredName ← − − − − − − (p , v) , & i fPredName simplexBody if COND (21) ← − − − − − − (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 22 / 29
Syntactical constructions Default Values Definition fPredName ( pT ) : ∼ defaults to ( TV ) (22) near the city center ( restaurant ) : ∼ defaults to (0) . (23) (p , v) , & i fPredName TV if COND (24) ← − − − − − − (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 23 / 29
Overview • Motivation − Crisp Databases − Fuzzy Databases − Real Databases − Expressive Queries − Limited Queries • Goal • Query Framework − Query Components − Query Structure − Basis − Semantics & Syntax − Syntactical Constructions • Present work • Conclusions (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 24 / 29
Present work Prototype • Web Interface Application • Parametric (databases) • Personalized (fuzzy concepts) (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 25 / 29
Overview • Motivation − Crisp Databases − Fuzzy Databases − Real Databases − Expressive Queries − Limited Queries • Goal • Query Framework − Query Components − Query Structure − Basis − Semantics & Syntax − Syntactical Constructions • Present work • Conclusions (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 26 / 29
Conclusions • Useful, expressive query structure (friendly user interface) • Flexible and fuzzy search included combination with crisp search and similarity • Easily implemented over CLP • Adequate for any kind of traditional database • Multiple possibilities of personalization (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 27 / 29
Questions Thank you (UPM, Spain) Getting Answers to Fuzzy and Flexible Searches FCTA’2013, Portugal 28 / 29
Recommend
More recommend
