Parameter-free and domain-independent similarity search with diversity

Santos, Lucio F. D.; Oliveira, Willian D.; Ferreira, Mônica Ribeiro Porto; Traina, Agma J. M.; Traina, Caetano

doi:10.1145/2484838.2484854

Cited by 27 publications

(33 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Diversity has been exploited in several areas, such as information retrieval [5], recommendation systems [2] and similarity queries [12,14]. Most approaches use metadata associated to the elements, such as taxonomies in document sets [5], cluster attributes in annotated data [2,16], and distances among the elements in dimensional and metric spaces.…”

Section: Search Results Diversificationmentioning

confidence: 99%

“…The optimization approach considers that similarity and diversity must compete to each other, taking a user-defined diversity preference as input, so that the results of pure-similarity algorithms (configured to retrieve more than the k elements requested by the user) can be re-ranked, inducing diversity among elements based on a bi-criteria objective function. However, this diversity definition results in an NP-hard problem [12,17] and restricts their usage to k-nearest neighbor queries.…”

Section: Search Results Diversificationmentioning

confidence: 99%

“…Other recent approach defines diversity without requiring more information from the user about the separation distance, called the Result Diversification based on Influence (RDI) [12]. This technique is based on a minimum distance that can be automatically estimated using the concept of "influence" intensity (I), which is defined as the inverse of the similarity distance between two elements.…”

Section: Search Results Diversificationmentioning

confidence: 99%

“…In order to obtain more relevant photos with such a holistic vision about the search space, several studies introduced a diversity factor in similarity queries [6,12,17]. Query result diversification aims at computing not only a result set with elements similar enough to satisfy the similarity conditions, but also to get a set of elements diverse among themselves to produce a more heterogeneous result.…”

Section: Introductionmentioning

confidence: 99%

“…Query result diversification aims at computing not only a result set with elements similar enough to satisfy the similarity conditions, but also to get a set of elements diverse among themselves to produce a more heterogeneous result. However, to the best of our knowledge, the result diversification definitions found in the literature [2,5,6,12,14,16,17] were always applied to unary operations, and none of them explored combining two relations, such as in a join operation.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Diversity in Similarity Joins

Santos

Carvalho

Oliveira

et al. 2015

Similarity Search and Applications

Self Cite

View full text Add to dashboard Cite

Abstract. With the increasing ability of current applications to produce and consume more complex data, such as images and geographic information, the similarity join has attracted considerable attention. However, this operator does not consider the relationship among the elements in the answer, generating results with many pairs similar among themselves, which does not add value to the final answer. Result diversification methods are intended to retrieve elements similar enough to satisfy the similarity conditions, but also considering the diversity among the elements in the answer, producing a more heterogeneous result with smaller cardinality, which improves the meaning of the answer. Still, diversity have been studied only when applied to unary operations. In this paper, we introduce the concept of diverse similarity joins: a similarity join operator that ensures a smaller, more diversified and useful answers. The experiments performed on real and synthetic datasets show that our proposal allows exploiting diversity in similarity joins without diminish their performance whereas providing elements that cover the same data space distribution of the non-diverse answers.

show abstract

Section: Search Results Diversificationmentioning

confidence: 99%

Section: Search Results Diversificationmentioning

confidence: 99%

Section: Search Results Diversificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Diversity in Similarity Joins

Santos

Carvalho

Oliveira

et al. 2015

Similarity Search and Applications

Self Cite

View full text Add to dashboard Cite

show abstract

Some Branches May Bear Rotten Fruits: Diversity Browsing VP-Trees

Jasbick¹,

Santos²,

Oliveira³

et al. 2020

Similarity Search and Applications

View full text Add to dashboard Cite

Relational Conditional Set Operations

Lescano

Cordeiro

2021

New Trends in Database and Information Systems

View full text Add to dashboard Cite

A set is a collection of different objects. Some basic operations from the Theory of Sets are the set membership (∈), subset (⊆), intersection (∩), and difference (−). The relational Algebra adapts the set operations to work with relations. However, as we show in this work, the set operations have limitations because of the implicit use of the identity predicate. That is, a tuple is a member of a set if it is identical to any tuple in the set. For example, let's consider two relations. The first one is a list of products that a person wants to buy. The second one is a list of products that one store has. Now, we could get any item from the desired products list and query "can we buy this item in the store?" with the set membership operator (∈), being true if the item is a member of the second set or false if not. With the set membership operator as a basis, we can also perform other queries such as subset, intersection, and difference. The subset (⊆) query would answer to "can I buy all the desired products in the store?". The intersection (∩) would answer to "what products can I buy in the store?" And finally, the difference (−) would answer to "what are the desired products that I cannot buy in the store?". Still, many applications need other comparison predicates that are not limited to identity. For example, if we add quantity and price to the sets of desired products and store's products, comparing the tuples by identity won't have much sense, since a product in the store with stock greater than the required should be valid, and it is also valid a product with a price lower than the user's maximum budget for that product. This MSc work presents the new Relational Conditional Set Operations. The novel operators encapsulate the idea of set operations with conditional queries, facilitating specific operators for them, and allowing their optimization. For example, they are potentially useful in applications of product sales with units and prices, job promotions with skills that have enough experience or certification level, and internships with minimum grades. We validate our proposal's semantics and scalability by studying the first of these applications. Also, we open path for future works such as: to implement the operators in a DBMS; to propose SQL queries able to answer these kind of queries and compare it with our current approach; to extend the idea for bag algebra; to explore a whole new path of optimization for our algorithms; to add support for complex data, allowing similarity comparisons in the predicate; and, to study the use of these operators as basis for other operations that currently use the traditional set operation as basis; among others.

show abstract

Parameter-free and domain-independent similarity search with diversity

Cited by 27 publications

References 18 publications

Diversity in Similarity Joins

Diversity in Similarity Joins

Some Branches May Bear Rotten Fruits: Diversity Browsing VP-Trees

Relational Conditional Set Operations

Contact Info

Product

Resources

About