Mining Interesting Possibilistic Set-Valued Rules

Disjunctive and conjunctive normal forms of pseudo-Boolean functions

We consider the problem of efficiently indexing Disjunctive Normal Form (DNF) and Conjunctive Normal Form (CNF) Boolean expressions over a high-dimensional multi-valued attribute space. The goal is to rapidly find the set of Boolean expressions that evaluate to true for a given assignment of values to attributes. A solution to this problem has applications in online advertising (where a Boolean expression represents an advertiser's user targeting requirements, and an assignment of values to attributes represents the characteristics of a user visiting an online page) and in general any publish/subscribe system (where a Boolean expression represents a subscription, and an assignment of values to attributes represents an event). All existing solutions that we are aware of can only index a specialized sub-set of conjunctive and/or disjunctive expressions, and cannot efficiently handle general DNF and CNF expressions (including NOTs) over multi-valued attributes. In this paper, we present a novel solution based on the inverted list data structure that enables us to index arbitrarily complex DNF and CNF Boolean expressions over multi-valued attributes. An interesting aspect of our solution is that, by virtue of leveraging inverted lists traditionally used for ranked information retrieval, we can efficiently return the top-N matching Boolean expressions. This capability enables emerging applications such as ranked publish/subscribe systems [16], where only the top subscriptions that match an event are desired. For example, in online advertising there is a limit on the number of advertisements that can be shown on a given page and only the "best" advertisements can be displayed. We have evaluated our proposed technique based on data from an online advertising application, and the results show a dramatic performance improvement over prior techniques.

The problem is the transformation of a conjunctive normal form (CNF) into a minimized (for the inclusion operator) disjunctive normal form (DNF) and vice versa. This operation is called the unionist product. For a CNF (resp. DNF), one pass of the unionist product provides the prime implicants (resp. implicates); two passes provide the prime implicates (resp. implicants). An algorithm built upon the classical Davis and Putnam procedure is presented for calculating, without the explicit minimization for the inclusion, this unionist product.

We describe the problem of mining possibilistic set-valued rules in large relational tables containing categorical attributes taking a finite number of values. An example of such a rule might be “IF HOUSEHOLDSIZE={Two OR Tree} AND OCCUPATION={Professional OR Clerical} THEN PAYMENT_METHOD={CashCheck (Max=249) OR DebitCard (Max=175)}. The table semantics is supposed to be represented by a frequency distribution, which is interpreted with the help of minimum and maximum operations as a possibility distribution over the corresponding finite multidimensional space. This distribution is approximated by a number of possibilistic prime disjunctions, which represent the strongest patterns. We present an original formal framework generalising the conventional boolean approach on the case of (i) finite-valued variables and (ii) continuos-valued semantics, and propose a new algorithm, called Optimist, for the computationally difficult dual transformation which generates all the strongest prime disjunctions (possibilistic patterns) given a table of data. The algorithm consists of generation, absorption and filtration parts. The generated prime disjunctions can then be used to build rules or for prediction purposes.KeywordsRule InductionPossibility DistributionPrediction PurposeElementary InductionInteresting RuleThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Normal forms for fuzzy logic functions and their approximation ability

Fuzzy set theories allow us to represent our knowledge under various interpretations and axiomatic foundations from linguistic to computational representations. There are at least four levels of knowledge representation: (i) linguistic, (ii) meta-linguistic, (iii) propositional, and (iv) computational. There are three transformations, which depend on the particular interpretations put on a knowledge representation schema. There are various choices, corresponding to one's interpretation of: (a) type of set representation, (fuzzy or crisp), (b) type of propositional connectives and normal forms, and (c) type of computational connectives, i.e., weak or strong t-norms and co-norms. In the light of these selections, fuzzy disjunctive and conjunctive normal forms (FDNF, FCNF) are derived from fuzzy truth tables. It is shown that classical expressions such as excluded middle etc., when fuzzified, ought to be reinterpreted with a type II, second order, semantic uncertainty. The classical expressions should not be interpreted as to whether they are valid or not. One can only state that the well-known tautologies of classical logic are valid to many degrees specified by an interval defined by their FDNF and FCNF. FDNF and FCNF boundaries identify the nonspecificity measure associated with type II, second order, semantic uncertainty. Thus, those researchers who are not familiar or who are not concerned with type II semantic uncertainty work with a myopic understanding of fuzzy set and logic theories. >

A checking test for a monotone read-once function f depending essentially on all its n variables is a set of vectors M distinguishing f from all other monotone read-once functions of the same variables. We describe an inductive procedure for obtaining individual lower and upper bounds on the minimal number of vectors T(f) in a checking test for any function f. The task of deriving the exact value of T(f) is reduced to a combinatorial optimization problem related to graph connectivity. We show that for almost all functions f expressible by read-once conjunctive or disjunctive normal forms, T(f) ~n / ln n. For several classes of functions our results give the exact value of T(f).

The theory of Boolean functions, especially in respect to representing these functions in the disjunctive or conjunctive normal forms, is extended in this chapter onto the case of finite predicates. Finite predicates are decomposed by that into some binary units, which will correspond to components of Boolean vectors and matrices and are represented as combinations of these units. Further, the main concepts used for solving pattern recognition problems are defined, namely world model, data, and knowledge. The data presenting information about the existence of some objects with definite combinations of properties is considered, as well as the knowledge presenting information about the existence of regular relationships between attributes. These relationships prohibit some combinations of properties. In this way, the knowledge gives the information about the non-existence of objects with some definite (prohibited) combinations of attribute values. A special form of regularity representation, called implicative regularities, is introduced. Any implicative regularity generates an empty interval in the Boolean space of object descriptions, which do not contradict the data. The problem of plausibility evaluation of induced implicative regularities should be solved by that. The pattern recognition problem is solved by two steps. First, regularities are extracted from the database (inductive inference); second, the obtained knowledge is used for the object recognition (deductive inference).

One of the tasks of synthesis and analysis of the functioning of combinational devices is the representation of a Boolean function in various bases. The absolute completeness of the use and implementation of various transformations of Boolean functions is impossible without the implementation of transformations into universal bases of Schaeffer (AND-NOT) or Webb (OR-NOT). The Logic library of the Maple computer algebra system allows you to use only the operations &nand (Schaeffer basis) and &nor (Webb basis) to build a model of a digital device. Therefore, there is a need to expand the functionality of the Logic library of the Maple computer algebra system by representing Boolean functions in Schaeffer and/or Pierce (Webb) bases. In this regard, the authors of the article have developed procedures that allow the representation of disjunctive or conjunctive normal forms in the Schaeffer basis or in the Webb basis. The development was carried out taking into account the existing data processing functions and procedures. The article describes the technology of extending the Logic library of the Maple computer algebra system by developed procedures that implement the transformation of a given or received Boolean function in Schaeffer or Webb bases.

An Algorithm is given which, as input, accepts a formula φ in prenex normal form, with prime formulas P=o, P≠o, for P ε Z [x 1 ,...,x 1 ], and some r ε N. It produces a quantifierfree formula ψ in disjunctive normal form (DNF) such that, in any algebraically closed field, φ ψ is true. The time required for a formula of length l with q quantifiers and r-q free variables is bounded by (c 1 ·l) c r 2 r q , for some constants c 1 , c 2 . If no ∀-quantifiers occur and the matrix of φ is in DNF, then the time is bounded by (c 1 ·l)rc q 2 .

We exhibit an unambiguous -DNF (disjunctive normal form) formula that requires conjunctive normal form width , which is optimal up to logarithmic factors. As a consequence, we get a near-optimal solution to the Alon–Saks–Seymour problem in graph theory (posed in 1991), which asks, How large a gap can there be between the chromatic number of a graph and its biclique partition number? Our result is also known to imply several other improved separations in query/communication complexity, learning theory, and automata theory.

We present some compositions of clones of Boolean functions that imply factorizations of Ω, the clone of all Boolean functions, into minimal clones. These can be interpreted as representation theorems, providing representations of Boolean functions analogous to the disjunctive normal form, the conjunctive normal form, and the Zhegalkin polynomial representations.

In this paper, we put forward a weak conjunctive normal form for G3, which is similar to the conjunctive normal form for classical propositional logic. We also give two algorithms to reduce a formula to weak conjunctive normal form, where one is based on charactering countermodels in model-based-formulas, and the other is based on rewriting formulas according to transformation rules. In the end, we show that there is no polynomial time algorithm to reduce an arbitrary formula to the weak conjunctive normal form.

We study the aggregation of individuals’ beliefs into their collective judgment on a certain proposition, called the conclusion , which is logically related to other preliminary propositions. While the literature has highlighted the crucial dependency of voting outcomes on the selection of an agenda —i.e., the set of propositions voted on—the structure of this dependency is not well understood. We introduce two novel premise-based rules, defined through the conjunctive normal form and disjunctive normal form of the conclusion. Our theorem shows that for any profile of voters’ beliefs, these rules respectively yield the highest and lowest voting outcomes, with respect to a given total order over the set of truth values, within the class of all premise-based rules. This result clarifies the extent to which a voting outcome can be manipulated through the selection of an agenda.

Feature terms are a common denominator of basic data-structures in knowledge representation and computational linguistics. The adaptation of the usual unification algorithms for first order terms is not straightforward, because feature terms may contain logical disjunction. Expansion into disjunctive normal form would reduce the problem more or less to unification of first order terms. But for reasons of efficiency, rewriting into disjunctive normal form should not be compulsory. In this paper, a sequent calculus is defined which gives a clear formal basis for proof optimizations: inference steps which require more than one subproof, i.e. which lead towards a “disjunctive normal form”, are only performed when they are no longer unavoidable. It is shown that the calculus is sound and complete with respect to the so-called feature structure interpretations of feature terms.