Li-Shiang Tsay

Association Action Rules

Action rules describe possible transitions of objects from one state to another with respect to a distinguished attribute. Previous research on action rule discovery usually required the extraction of classification rules before constructing any action rule. This paper gives anew approach for generating association-type action rules. The notion of frequent action sets and Apriori-like strategy generating them is proposed. We introduce the notion of a representative action rules and give an algorithm to construct them directly from frequent action sets. Finally, we introduce the notion of a simple association action rule, the cost of association action rule, and give a strategy to construct simple association action rules of a lowest cost.

Discovering Extended Action-Rules (System DEAR)

Action rules introduced in [3] and investigated further in [4] assume that attributes in a database are divided into two groups: stable and flexible. In general, an action rule can be constructed from two rules extracted earlier from the same database. Furthermore, we assume that these two rules describe two different decision classes and that our goal is to re-classify some objects from one of these decision classes to the other one. Flexible attributes provide a tool for making hints to a business user what changes within some values of flexible attributes are needed. for a given object to re-classify this object to another decision class. In [3], we suggested what changes are needed to classification attributes listed in both rules but we did not consider situations when such an attribute is listed only in one of these rules. Also, neither in [3] nor [4] we provide a way to compute support and confidence of action rules. ! In this paper, we show how system DEAR is discovering extended action rules which give better strategies for re-classifying objects than strategies provided by action rules. Also, the confidence of extended action rules is much higher than confidence of corresponding action rules. System DEAR, implemented in KDD Laboratory at UNC-Charlotte, requires Windows 95 or higher. It does not discretize numerical attributes which means some discretization algorithm has to applied before DEAR is used

Action rules discovery: system DEAR2, method and experiments

Subjective measures, used to model interestingness of rules, are user-driven, domain-dependent, and include unexpectedness, novelty and actionability (Adomavicius and Tuzhilin 1997, Liu et al. 1997, Silberschatz and Tuzhilin 1995). Liu et al. (1997) define a rule as actionable, if a user can do an action to his/her advantage based on that rule. Their notion of actionability is too vague and leaves the door open to a number of different interpretations. Raś and Wieczorkowska (2000) assume that actionability has to be expressed in terms of attributes that are present in the database. They have introduced a new class of rules (called action rules) that are constructed from certain pairs of association rules extracted from that database. A conceptually similar definition of an action rule was proposed independently by Geffner and Wainer (1998). Action rules have been investigated further in Raś and Gupta (2002), Raś and Tsay (2003), Raś et al. (2005) and Tzacheva and Raś (2004). In order to construct action rules, it is required that attributes in a database are divided into two groups: stable and flexible. Flexible attributes are used in a decision rule as a tool for making hints to a user what changes within some of their values are needed to reclassify a group of objects from one decision class into another one. Two strategies for generating action rules are presented. The first one, implemented as system DEAR, generates action rules from certain pairs of association rules. The second one, implemented as system DEAR2, is based on a tree structure that partitions the set of rules, having the same decision value, into equivalence classes each labelled by values of stable attributes (two rules belong to the same equivalence class, if values of their stable attributes are not conflicting each other). Now, instead of comparing all pairs of rules, only pairs of rules belonging to some of these equivalence classes are compared to construct action rules. This strategy significantly reduces the number of steps needed to generate action rules in comparison to DEAR system.

Advances in Intelligent Information Systems

Intelligent Information Systems (IIS) can be defined as the next generation of Information Systems (IS) developed as a result of integration of AI and database (DB) technologies. IIS embody knowledge that allows them to exhibit intelligent behavior, allows them to cooperate with users and other systems in problem solving, discovery, retrieval, and manipulation of data and knowledge. For any IIS to serve its purpose, the information must be available when it is needed. This means that the computing systems used to store data and process the information, and the security controls used to protect it must be functioning correctly. This book covers some of the above topics and it is divided into four sections: Classification, Approximation and Data Security, Knowledge Management, and Application of IIS to medical and music domains.

Tree-based Algorithm for Discovering Extended Action-Rules (System DEAR2)

Action rules introduced in [3] and investigated further in [5] assume that attributes in a database are divided into two groups: stable and flexible. In general, an action rule can be constructed from two rules extracted earlier from the same database. Furthermore, we assume that these two rules describe two different decision classes and that our goal is to re-classify some objects from one of these decision classes to the other one. Flexible attributes provide a tool for making hints to a user what changes within some values of flexible attributes are needed for a given group of objects to re-classify these objects to another decision class. In [4], to build action rules, all pairs of rules defining different decision classes have been considered by the algorithm. In this paper we propose a new algorithm which will significantly decrease the number of pairs of rules needed to be checked for action rules construction and the same speed up the whole process.

Discovering the concise set of actionable patterns

It is highly expected that knowledge discovery and data mining (KDD) methods can extract useful and understandable knowledge from large amount of data. Action rule mining presents an approach to automatically construct relevantly useful and understandable strategies by comparing the profiles of two sets of targeted objects -- those that are desirable and those that are undesirable. The discovered knowledge provides an insight of how relationships should be managed so that objects of low performance can be improved. Traditionally, it was constructed from one or two classification rules. The quality and quantity of such Action Rules depend on adopted classification methods. In this paper, we present StrategyGenerator, a new algorithm for constructing a complete set of Action Rules which satisfies specified constraints. This algorithm does not require prior extraction of classification rules. Action rules are generated directly from a database.

Reclassification Rules

The ultimate goal of knowledge discovery (KD) is to extract sets of patterns leading to useful knowledge for obtaining user desirable outcomes. The key characteristics of knowledge usefulness is that these patterns are actionable. In the last decade, KD algorithms such as mining for association rules, clustering, and classification rules, have made a tremendous progress and have been demonstrated to be of significant value in a variety of real-world data mining applications. However, the results of the existing methods require to be further processed in order to suggest actions that achieve the desired outcome, by giving only previously acquired data. To address this issue, we present a novel technique, called reclassification rules, to gather all facts, to understand their causes and effects, and to list all potential solutions and the responding effects. Algorithm, Strategy Generator-II, is proposed to discover a complete set of reclassification rules which meets pre-specified constraints.

E-Action Rules

The ability to discover useful knowledge hidden in large volumes of data and to act on that knowledge is becoming increasingly important in today’s competitive world. Action rules were proposed to help people analyze discovered patterns and develop a workable strategy for actions [10]. A formal definition of an action rule was independently proposed in [4]. These rules have been investigated further in [11, 12].

Tree-Based Algorithms for Action Rules Discovery

One of the main goals in Knowledge Discovery is to find interesting associations between values of attributes, those that are meaningful in a domain of interest. The most effective way to reduce the amount of discovered patterns is to apply two interestingness measures, subjective and objective. Subjective measures are based on the subjectivity and understandability of users examining the patterns. They are divided into actionable, unexpected, and novel. Because classical knowledge discovery algorithms are unable to determine if a rule is truly actionable for a given user [1], we focus on a new class of rules [15], called E-action rules, that can be used not only for automatic analysis of discovered classification rules but also for hints of how to reclassify some objects in a data set from one state into another more desired one. Actionability is closely linked with the availability of flexible attributes [18] used to describe data and with the feasibility and cost [23] of desired re-classifications. Some of them are easy to achieve. Some, initially seen as impossible within constraints set up by a user, still can be successfully achieved if additional attributes are available. For instance, if a system is distributed and collaborating sites agree on the ontology [5], [6] of their common attributes, the availability of additional data from remote sites can help to achieve certain re-classifications of objects at a server site [23]. Action tree algorithm, presented in this paper, requires prior extraction of classification rules similarly as the algorithms proposed in [15] and [17] but it guarantees a faster and more effective process of E-action rules discovery. It was implemented as system DEAR 2.2 and tested on several public domain databases. Support and confidence of E-action rules is introduced and used to prune a large number of generated candidates which are irrelevant, spurious, and insignificant.

Mining E-Action Rules, System DEAR

The essential problem of Knowledge Discovery in Databases is to find interesting relationships, those that are meaningful in a domain. This task may be viewed as one of searching an immense space of possible actionable concepts and relations. Because the classical knowledge discovery algorithms are not able to determine if a pattern is truly actionable for a user, we focus on a new class of action rules, called e-action rules that can be used not only for automatically analyzing discovered patterns but also for reclassifying some objects in the data from one state into another more desired state. For a quicker and more effective process of e-action rules discovery, action tree algorithm is presented. Support and confidence of the rules are proposed to prune a large number of irrelevant, spurious, and insignificant generated candidates. The algorithm is implemented as DEAR_2.2 system and it is tested on several public domain databases. The results show that actionability can be considered as a partially objective measure rather than a purely subjective one. E-Action rules are useful in many fields such as medical diagnosis and business.