You are on page 1of 3

What is a Fuzzy set?

The membership function value of an object in a set is either 0 or 1. The value 0 is assigned when the object
does not belong to the set while the value 1 is assigned when the object completely belongs to that set. This
is the concept of crisp set. A fuzzy set is the generalization of the classical or crisp set which has a
membership function with range all values in [0,1]. Thus, an object could partially belong to a fuzzy set with
a membership value in (0,1), entirely belong to the fuzzy set with a value of 1, or not belong to the fuzzy set
with a value of 0.
Formally, a fuzzy set A is defined by the membership function µ𝐴 = 𝑈→[0,  1] in which U is called the
( )
universe of discourse. Therefore µ𝐴 𝑢𝑖 specifies the grade of membership of 𝑢𝑖 ∈ 𝐴.

The cardinality of a fuzzy set A, defined over a finite U, is the summation of all the membership degrees

|𝐴| = ∑ µ𝐴(𝑢)
𝑢∈𝑈

A ∝ − 𝑐𝑢𝑡 of a fuzzy set is a crisp set 𝐴α that has as its members all 𝑢∈𝑈 such that µ𝐴(𝑢)≥α,  α  ∈ (0, 1].

A notation to represent a fuzzy set over a finite universe is


µ𝐴(𝑢)
𝐴={ 𝑢 |𝑢ϵ𝑈
}
Operations with fuzzy sets
● Fuzzy set 𝐴 is a subset of fuzzy set 𝐵, 𝐴 ⊆𝐵, if and only if ∀𝑢 ∈ 𝑈, μ𝐴(𝑢) ≤ µ𝐵(𝑢).

Figure 3 - A is subset of B

● Fuzzy set 𝐴 is equal to fuzzy set 𝐵, 𝐴 = 𝐵, if and only if ∀𝑢 ∈ 𝑈, μ𝐴(𝑢) = µ𝐵(𝑢).

● Fuzzy set 𝐴 is the complement of 𝐴 if and only if ∀𝑢 ∈ 𝑈, µ𝐴(𝑢) = 1 − μ𝐴(𝑢).

Figure 4 - A and A complement.

● The triangular norm (t-norm or T-norm) is a function 𝑇: [0, 1] × [0, 1] → [0, 1] that must satisfy
certain properties (commutativity, associativity, monotonicity, 1 is identity element). t-norm function
is used to generalize the concept of conjunction in fuzzy set theory
𝐴∩𝐵 = 𝑇(𝐴, 𝐵)
A simple choice is the minimum t-norm, also known as Gödel t-norm, which takes the smaller
between the two membership values
(
µ𝐴∩𝐵(𝑢) = µ𝐴(𝑢), µ𝐵(𝑢) )

● The t-conorm (also called s-norm) function is the t-norm dual function built with the generalization
of De Morgan’s laws. The t-conorm is a generalization for the concept of disjunction in fuzzy set
theory.
𝐴∪𝐵 = 𝑆(𝐴, 𝐵)
A simple choice is the maximum t-conorm, also known as Gödel t-conorm, which takes the greater
between the two membership values

(
µ𝐴∪𝐵(𝑢) = µ𝐴(𝑢), µ𝐵(𝑢) )

Boolean information retrieval system


A Boolean IR system is composed by a set of documents D and a set of index terms T. Each document d is
described by a set of terms w, eventually after the application of stemming, lemmatization, or stopwords
remotion. A user can express its information need by means of a query. In the Boolean IR system, the query
is a combination of index terms logically connected by AND, OR, and NOT. Hence, a document is
completely relevant or completely not relevant with respect to a given query.

Fuzzy information retrieval system


Fuzzy information retrieval system can be considered as a generalization or a formalization of the weighted
Boolean model, which is an extension of the Boolean model that aims at letting the user specify the
importance of index terms inside a query assigning a weight for each one. Fuzzy IR system instead, gives a
measure of relevance of a document to a query: fully, rather, poorly relevant, or not relevant at all. It can be
misled by the probabilistic IR system, but the latter decides with some heuristics whether a document is
relevant or not, without any grades of relevance, just the binary assumption.

Document representation
The index term set of the Boolean Model will be extended to a fuzzy index term set to describe a document
in a fuzzy IR system.
𝐹𝑑 = { µ𝐹 (𝑡)/𝑡 | 𝑑∈𝐷 𝑎𝑛𝑑 𝑡∈ 𝑇 } ⊂ 𝑇
𝑑

This fuzzy set can be interpreted as the fuzzy document description of document d. Sometimes the fuzzy
document description is defined as a binary fuzzy indexing relation
𝐹𝐼 = { μ𝐹 (𝑑, 𝑡)/(𝑑, 𝑡)|𝑑∈𝐷 𝑎𝑛𝑑 𝑡∈𝑇}
𝐼

You might also like