Chih-Chen Lin

Implementing large production systems in a DBMS environment: concepts and algorithms

It has been widely recognized that many future database applications, including engineering processes, manufacturing and communications, will require some kind of rule based reasoning. In this paper we study methods for storing and manipulating large rule bases using relational database management systems. First, we provide a matching algorithm which can be used to efficiently identify applicable rules. The second contribution of this paper, is our proposal for concurrent execution strategies which surpass, in terms of performance, the sequential OPS5 execution algorithm. The proposed method is fully parallelizable, which makes its use even more attractive, as it can be used in parallel computing environments.

Exploiting concurrency in a DBMS implementation for production systems

An important aspect of integration of AI and DBMS technology is identifying functional similarities in database processing and reasoning with rules. In this paper, we focus on applying concurrency techniques to rule-based production systems. We tailor DBMS concurrent execution to a production system environment and investigate the resulting concurrent execution strategies for productions. This research is in conjunction with a novel DBMS mechanism for testing if the left hand side conditions of productions are satisfied. This set-oriented mechanism uses a special data structure implemented using relational tables. To demonstrate the equivalence of a serial and a concurrent (interleaved) execution strategy, for a set of productions, we assume a locking scheme to enforce serializability and specify what locks must be obtained. We define a logical commit point for a production. After this point the execution of a production is independent of all other (executing) productions. We compare the number of possible serial and parallel execution schedules.

Data intensive production systems: the DIPS approach

In this paper we highlight the basic approach taken in the design of the DIPS system, and briefly present the main contributions. These include the use of special data structures to store rule definitions; they are implemented using relations. A matching algorithm uses these structures to efficiently identify when the antecedents of productions are satisfied, making them applicable for execution. Partial match information stored in the data structures is used by the matching algorithm. We also describe a proposed concurrent execution strategy for applicable productions, which surpasses in performance, the traditional sequential OPS5 production execution algorithm. The requirements for a correct, serializable execution, based on locking, is described. An advantage of the matching technique in DIPS is that it is fully parallelizable, which makes it attractive for implementation in parallel computing environments.

Structured representation of rule-based specifications in CIM using updated Petri nets

A graphical representation tool-updated Petri nets (UPN)-has been developed to model rule-base specifications for CIM databases. UPN facilitates the modeling of relationships between operations of various manufacturing application systems and the database updates and retrievals among all the respective distributed databases. Based on this representation, a hierarchical modeling technique which includes refining and aggregating rules has also been developed. An application of the UPN is demonstrated in designing rule-based systems for controlling and integrating the information flow between manufacturing applications, including computer aided design, computer aided process planning, manufacturing resources planning, and shop floor control. >

INformation Systems for Integrated Manufacturing (INSIM): a design methodology

Abstract Full control and management of information flow in manufacturing has not yet been achieved, mainly because of the data inconsistencies and lack of established functional relationships among different manufacturing application systems. Research toward CIM has been concentrating on the computerization of individual functions, such as computer-aided design and shop floor control, and the integration of data relations, such as global database frameworks and distributed database management systems. A mechanism with the potential to control the information flow among all of the manufacturing application systems, in order to streamline factory activities based on company-specific and company-wide policies and procedures is proposed here. The goal is to achieve a fully-integrated manufacturing management system. The INformation Systems for Integrated Manufacturing (INSIM) reflects a design methodology to build a knowledge base to serve as the control mechanism. This design methodology features an enhance...

Implementation of a Prolog-INGRES interface

This report describes a working prototype of a Prolog-INGRES interface based on external semantic query simplification. Semantic query simplification employs integrity constraints enforced in a database system for reducing the number of tuple variables and terms in a relational query. This type of query simplifier is useful in providing very high level user interfaces to existing database systems. The system employes a graph theoretic approach to simplify arbitrary conjunctive queries with inequalities. One very interesting feature of the system is to provide meaningful error messages in case of an empty query result resulting from contradiction. In addition to data, rules are stored in the database as well and are retrieved automatically if the Prolog program references them but they are not defined in the Prolog rulebase.

Performance of DBMS implementations of production systems

The performance of various methods for the implementation of large production rule bases is studied. The authors first give a brief overview of the basic techniques, and then present some optimization alternatives for one of them. The results obtained indicate that both the Rete network method and the one based on COND relations are feasible in a relational database system environment. One major advantage of the method based on COND relations is that it localizes the search for applicable rules. That is, for a given update to the database only a single relation (the corresponding COND relation) needs to be searched.<<ETX>>

A Geometric Approach to Indexing Large Rule Bases

The efficiency of finding qualifying rules against updates in large production systems has always been an important research issue. In database implementations of rule systems, this problem is even more critical. Predicate Indexing is a reduced version of the above problem when only selection conditions are considered. In this paper we draw a similarity between predicate indexing and indexing of geometric data, and discuss index design issues. Results from experiments using special data structures are also presented.

Performance issues in the binary relationship model

Abstract In this paper we propose to use the Binary Relationship Model not only in the logical, but also in the physical level. At the logical level we propose and analyze a query optimization algorithm. The major optimization issue is to find an optimized join sequence for chain queries. A branch-and-bound search algorithm with dynamic programming is used in the search for such a sequence. At the physical level we propose to store the binary relations using multiattribute hashing. We develop a join algorithm which takes advantage of the clustering that multiattribute hashing achieves, and derive analytical formulas to predict its performance.

Control and Management of Information Flow for Computer Integrated Manufacturing

Abstract A mechanism to control and manage the information flow among several manufacturing application systems, in order to form one integrated manufacturing system, is proposed in this paper. The INformation System for Integrated Manufacturing (INSIM) reflects a design methodology to build a knowledge base to serve as the information control mechanism. The methodology includes the collection of rules (knowledge acquisition), their graphical modeling, systematic model validation and automated implementation to an operating production system. This design methodology features an enhanced graphic modeling tool - Updated Petri Nets (UPN) - which is capable of modeling database updates and retrievals, under specific constraints and conditions and uses a hierarchical modeling approach. For this purpose an UPN editor was developed which is used to create, explode, unfold, validate and correct the information flow model.