T. Sellis, C.C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: concepts and algorithms. In ACM SIGMOD Int. Conf. on Management of Data, pages 404--412, 1988.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... the intercondition tests are join operations on the relations which were selected [20] The problem of efficiency in the match phase of rule based system execution led to the development of the Rete algorithm [2] 3] the TREAT algorithm [11] and several related optimizations [1] 15] 16] [17]. The Rete algorithm uses a dataflow network compiled from the preconditions of all the rules. Newly added or removed working memory elements are processed through this pattern matching network resulting in activations at the leaf nodes of new instantiations for the conflict set. Figure 1 shows a ....

T. Sellis, C. Lin, and L. Raschid, Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms, ACM-SIGMOD International Conference on the Management of Data, pages 404-412, 1988.

....the requirements of new demanding applications, such as CAD, multimedia, robotics and expert systems. Along this line, several researchers have proposed the integration of production rule languages (traditionally used in expert system shells) within database environments [8] 11] 19] 22] [23], 25] 26] 30] while others have studied the possibility of adding deductive capabilities to database systems [1] 2] 4] 5] 20] 21] 28] As a result of these studies, advanced research prototypes have been produced (e.g. LDL , Coral, Starburst, Postgres) Moreover, recent ....

T. Sellis, C.C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: concepts and algorithms. In ACM SIGMOD Int. Conf. on Management of Data, pages 404--412, 1988.

....(Year) John Wiley Sons, Inc. two independent research trends have emerged: on one hand the use of production rule languages (traditionally used in expert system shells) for expressing active computations, that is, manipulations of data to be executed automatically whenever certain events occur [5, 6, 7, 10, 13, 14, 15, 16, 17]; on the other hand the use of logic based rule languages for expressing, in a declarative way, complex database queries and deductive computations [1, 2, 3, 4, 11] From a practical point of view, the integration of these paradigms (active and deductive) into a unique homogeneous framework would ....

T. Sellis, C.C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: concepts and algorithms. In ACM SIGMOD Int. Conf. on Management of Data, pages 404--412, 1988.

.... can be found in [ Schmolze and Neiman, 1992 ] The concept of locking elements to prevent interactions due to concurrent modifications is widely used in database systems and a similar scheme to the one just described was implemented in a DBMS based production system by Sellis,et al. Sellis et al. 1987 ] This implementation uses region locks to prevent interactions due to negative conditions. A region lock typically prohibits access to a class of working memory elements, possibly restricted by value and, depending on the precision with which the region can be identified, may prove unduly ....

....2.4. 3 Set Functions Recent work on incorporating rule based systems into database systems has indicated that increased efficiency and ease of programming can be obtained through the use of set based rules rather than instance based rules [ Gordin and Pasik, 1991, Widom and Finkelstein, 1990, Sellis et al. 1987 ] In the instance based implementation of a rule based system, one and only one rule instantiation is created for each set of working memory elements which match a lefthand side. That is, if the lefthand side of a rule has N positive condition elements, then the resulting instantiation will ....

Timos Sellis, Chih-Chen Lin, and Louiqa Raschid. Implementing large production systems in a dbms environment: Concepts and algorithms. Technical Report CS-TR-1960, Dept. of Computer Science, University of Maryland at College Park, 1987.

....rules, it is responsible for the automatic enforcement of data consistency by detecting the occurrences of the events. When the event is met, it evaluates the condition, and executes the action if the condition evaluates to true. Much work has been done on the rule processing in DBMSs[15, 16, 17]. However, the event component of rules has received attention only recently [4, 8, 10, 11, 12, 14] and perhaps is the least understood compared to the condition and action components. Effectively detecting an event, possibly composite one, may be a crucial problem in processing the ECA rules, ....

T. Sellis, C.-C. Lin and L. Raschid, "Implementing large Production Systems in a DBMS environment: Concepts and Algorithms," Proceedings of the ACM SIGMOD International Conference on Management of Data, Chicago, Illinois, pp. 404-412, Jun., 1988.

....to the databases. It may be a natural consequence that such an alternative entails a seamless integration of our framework into the relational databases. In other words, the design alternative can make all the approaches providing persistence to production systems also available to us [13] 20][17][21] Another design alternative of our work is to enhance the semantic expressiveness of conventional production rule based languages by providing fuzzy match facility. As a uniform construct to handle fuzzy linguistic variables as well as fuzzy numbers, it enables users to describe the semantics ....

T. Sellis, C.-C. Lin and L. Raschid, Implementing large production systems in a DBMS environment: concepts and algorithms, In proc. of the ACM SIGMOD International Conference on Management of Data, Chicago, Illinois, (Jun., 1988) 404-412.

....shows much difference from these two facilities both in semantic and syntax, which makes the integration difficult to achieve. Many advantages have been reported to be gained when conforming the styles of their rule processing to those of relational framework [15] 20] 23] Some works such as [5][18][21] feature such attempts to implement relational production languages, relocating facts into the relational databases from working memory in discrimination networks [6] 8] 16] Currently, production languages such as OPS5 [6] and ECLPS [9] are being used to implement the relational production ....

....implementation of production languages. However, differently from such a main memory implementation, the structure of our information including facts is defined in a way to have the least semantic gap with relations. Since there are many database implementations of the production languages [15][18][21] 23] our design criterion adopted in the structure may leave our information persistent by enabling our framework to fully exploit the implementations. We now begin this section by defining a fuzzy set. Definition 1 A fuzzy set F in a classical universe of discourse U is characterized by ....

T. Sellis, C.-C. Lin and L. Raschid, Implementing large production systems in a DBMS environment: concepts and algorithms, In proc. of the ACM SIGMOD International Conference on Management of Data, Chicago, Illinois, (Jun., 1988) 404-412.

....one seeks to optimize one processing of a rule program at a time, 1 Almost all production rule systems developped in AI implement such a scheme with variants of RETE or TREAT RR n 2533 4 F. Fabret, E. Simon and several implementations of this scheme have been proposed for database rule systems [SLR88], DWE89] SZ91] Han92] In this paper, we argue that this scheme is however not appropriate for optimizing rules in active database systems. In these systems (see [WCD95] for a survey) a rule is triggered by the occurence of some specific triggering event associated with the rule. Events are ....

T. Sellis, C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In Proc. International Conference SIGMOD, Chicago, May 1988.

....systems, such as OPS5, use production rules as the programming paradigm. All rules are active all the time. Whenever the conditions for a rule are satisfied, it fires, and executes its action part. Some good work has been done towards designing large production systems in a database context [23, 27]. The difference between these systems and ours is that in these systems the rules become ends in themselves: the entire program is written in terms of rules. In O , procedural descriptions may be used where they are appropriate, and rules where they are. Also, in these systems, all rules are ....

....A major difficulty with production systems is that they are very hard to debug when there are situations in which multiple rules can fire, since each has its conditions satisfied. Priority levels, whether explicitly stated as in [14, 24] or determined by criteria such as specificity [23, 27], we believe are a bad idea, since their use is against the declarative spirit of constraint specification and can decrease the potential for concurrent execution. Since we do not specify the order in which the rules will fire, the programmer is forced to make a conservative assumption, and we ....

T. Sellis, C. Lin and L. Raschid, "Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms", Proc. of the ACM-SIGMOD Int'l Conf. on the Management of Data, Chicago, Illinois, 1988. - 30 -

....and triggers will be processed using a query based approach, which will not scale up to a large number of triggers and constraints. We speculate that it may be possible to work around this assumption. A predicate index like the one proposed in this paper potentially could be used. The DIPS system [Sell88] uses a set of special relations called COND relations for each condition element (tuple variable) in a rule. These COND relations are queried and updated to perform testing of both selection and join conditions of rules. Embedding all selection predicate testing into a process that must query ....

Sellis, T., C.C. Lin and L. Raschid, "Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms," Proceedings of the 1988 ACM SIGMOD Conference.

....Then, any processing of the rule base uses the optimized versions of the rules. This scheme works well for production systems 1 where one seeks to optimize one processing of a rule program at a time, and several implementations of this scheme have been proposed for database rule systems [SLR88], DWE89] SZ91] Han92] In this paper, we argue that this scheme is however not appropriate for optimizing rules in active database systems. In these systems (see [WCD95] for a survey) a rule is triggered by the occurence of some specific triggering event 1 Almost all production rule ....

T. Sellis, C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In Proc. International Conference SIGMOD, Chicago, May 1988.

....such as OPS5 [BFKM85] to a conventional DBMS; this approach is taken in, e.g. Tzv88] More recently the prevalent approach has been to build rule processing directly into the database system. Examples of recent or ongoing projects in expert database systems are [BM93,DE89,DOS 92,GP91,SLR88] Note that some systems described as active database systems This work was partially performed while the authors were at the IBM Almaden Research Center, San Jose, CA. At Stanford this work was supported by equipment grants from Digital Equipment Corporation and IBM Corporation. y Permanent ....

T. Sellis, C.-C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 404--412, Chicago, Illinois, June 1988.

....the introduction of deductive databases as a research area [Bocc86a, Bocc86b, KoGM87, Mink88, MUvG86, TsZa86] there has been a corresponding interest in extending the functionality of database management systems (DBMS) to provide support for rules. Some of this research has been described in [CeWi90, DeEt88, MaSi88, SeLR88, SeLR93, SiMa88, Wido91, WiFi90]. Whereas rules in deductive database systems have been primarily perceived as supporting retrieval queries against the database, most DBMS research has focused on supporting rules whose execution can update the database and trigger the further execution of rules. This paradigm is similar to the ....

Sellis, T., Lin, C-C. and Raschid, L. Implementing large production systems in a DBMS environment: concepts and algorithms. Proceedings of the ACM Sigmod International Conference on the Management of Data, pages 34-45, 1988.

....the language proposal is quite influential. It demonstrates the potential benefits and feasibility of integrating database and expert system technology. It is also a good example of showing the advantage of having a formal model underlying a language design. 2.2. 2 DIPS Like RPL, the DIPS system [120, 130, 131] represents another example of using database technology in supporting production rules functionality. Two special data structures are used for the processing of OPS5 rules in a database environment: the Working Memory Relations (WM) and the Condition Relations (COND) Each class of WME s is ....

Timos Sellis, Chih-Chen Lin, and Louiqa Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In ACM SIGMOD Intl. Conf. on Management of Data, pages 404--412, 1988.

....rule execution, taking into account externally generated operations, selftriggering rules, and simultaneous triggering of multiple rules. 1 Introduction Recently, there has been considerable interest in integrating production rules systems and database management systems. Some work, such as [DE89,SLR88, Tzv88], focuses on using database technology to efficiently support OPS like production rules languages [BFKM85] Other work including ours focuses on extending database systems to include a production rules facility [Coh89, dMS88, Esw76, Han89, KDM88, MD89, RS89, SHP88, SJGP90] Generally, ....

T. Sellis, C.-C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 404--412, Chicago, Illinois, June 1988.

....semantics (see Section 4) A first prototype of Chimera has been implemented, employing some techniques adapted from Starburst [12] There are several other relational active database systems, not as closely related to Starburst as the systems described above. Two projects, DATEX [8] and DIPS [38], implement the OPS5 rule language using an underlying database system and special indexing techniques to support efficient processing of large rule and data sets. The PARADISER project also uses a database system for efficient processing of expert system rules, but in PARADISER the focus is on ....

T. Sellis, C.-C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 404--412, Chicago, Illinois, June 1988.

.... that addresses the problem at the schema level [17] 33] There are a number of similarities with the database and knowledge base community and the proposed work, e.g. the concept of articulation [9] 15] translating heterogeneous information into a meta level model [22] 30] active database [29] and associating constraints and triggers with objects [11] etc. hence the adaptation of methods from the heterogeneous database literature, mediation and integration aspects to the problem of disciplined manipulation of information sources across networks, languages and platforms. Considering ....

T. Sellis, C. Lin and L. Rashid, "Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms"; in Proceedings of the ACM SIGMOD International Conference on Management of Data, Chicago, Illinois, 1991.

....processor for a production rule language such as OPS5 [7] to a conventional DBMS; this approach is taken in, e.g. 23] More recently the prevalent approach has been to build rule processing directly into the database system. Examples of recent or ongoing projects in expert database systems are [6, 11, 12, 13, 21]. Note that some systems described as active database systems actually use the Condition Action rule paradigm, and hence fall into the class of expert database systems as we use the term here; examples of such systems are [14, 22] Since expert database systems evolved from production rule systems ....

T. Sellis, C.-C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: Concepts and algorithms. In Proc. ACM SIGMOD Int'l Conf. on Management of Data, pages 404--412, June 1988.

....can be incrementally maintained as user defined priorities are altered. We also discuss how the proposed scheme can be extended to build a multi level hierarchical priority system. 1 Introduction Incorporation of production rules into database systems has recently received considerable attention [6, 7, 8, 11, 19, 21, 25, 26, 27, 30]. A central issue in production rule systems is conflict resolution [20, 14] Given that two or more rules are triggered, a conflict resolution mechanism determines which rule is considered first for execution. Some rule systems (for example, Postgres Current address: Computer Science ....

T. Sellis, C.-C. Lin, and L. Raschid. Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms. In Proc. ACM-SIGMOD International Conference on Management of Data, pages 404--412, Chicago, June 1988.

....for the DATEX expert database. The Matchbox [65] algorithm has been proposed for parallel production systems, and Gridmatch [84] was proposed as a basis for integrating production rules with relational databases. CLL is the name we have given to the unnamed algorithm by Sellis, Lin, and Raschid [71], to provide efficient rule activation support in relational databases. Finally, A TREAT has been proposed as the matcher in the Ariel active database [37] We see that most of the algorithms either use CR or CS, or both. Thus, their worst case space requirements grow as a polynomial function of ....

C. Sellis, C. Lin, and L. Raschid. Implementing large production systems in a dbms environment: Concepts and algorithms. In Proceedings of the ACM SIGMOD 1989, Intl. Conf. on the Management of Data. ACM Press, 1989.

....bases. One research effort in DBMS rule processing has been to extend the syntax of production rule languages with set oriented constructs and use more efficient set oriented DBMS query processing strategies to evaluate conditions and execute actions of rules. This research has been reported in [8, 9, 10, 15, 36, 37, 44]. Parallelism has also been used to improve the efficiency in executing rule actions as well as in evaluating conditions [2, 6, 16, 17, 18, 20, 24, 27, 34, 35, 39] Research reported in [2, 6, 18, 34, 35, 39] has investigated the execution of a number of rules in parallel, and the corresponding ....

....as it need not check conditions of all rules, but instead, only a small subset of the rules are examined. For example, OPS5 uses a binary discrimination network, called the Rete Network [14] which contains redundant data and allows for incremental evaluation of conditions of rules. Similarly, in [36, 37], we proposed a condition monitoring scheme, DBCond, which uses a set of auxiliary relations, called the COND relations, one for each database relation. Many other proposals for DBMS rules also support auxiliary information stored in the database [9, 41, 42] When rule actions are executed, they ....

Sellis, T., Lin, C-C., and Raschid, L. Implementing large production systems in a DBMS environment: concepts and algorithms. Proceedings of the ACM-SIGMOD International Conference on the Management of Data, Chicago, IL (1988).

No context found.

T. Sellis, C.C. Lin, and L. Raschid. Implementing large production systems in a DBMS environment: concepts and algorithms. In ACM SIGMOD Int. Conf. on Management of Data, pages 404--412, 1988.

No context found.

Sellis T., Lin C., and Raschid L. (1988) Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms. ACM-SIGMOD International Conference on the Management of Data, pages 404-412. 8

No context found.

Sellis T., Lin C., and Raschid L. (1988) Implementing Large Production Systems in a DBMS Environment: Concepts and Algorithms. ACM-SIGMOD International Conference on the Management of Data, pages 404-412.

No context found.

T. Sellis, C.C. Lin, and Raschid L. Implementing large production systems in a DBMS environment: concepts and algorithms. In ACM SIGMOD International Conf. on Management of Data, pages 404--412, 1988.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC