Imielinski T., Virmani A., Abdulghani A.: Datamine: Application programming interface and query language for data mining. Proc. of the 2nd KDD Conference (1996)  Home/Search   Document Not in Database   Summary   Related Articles   Check

....operators for conducting inductive and deductive inference, statistical analysis, and many supportive functions. KQL is quite di#erent from traditional high level languages for data exploration, which have generally been Prolog based. Among the exceptions to the Prolog based approach, M SQL [3] is philosophically similar to KQL in that it builds upon the SQL data query language, integrating it with one inductive operator. KQML [2] allows the querying for specific pieces of knowledge, although it does not support the abstract templates and multiple discovery operators supported by KQL. ....

Imielinski, T., Virmani, A., Abdulghani, A. (1996) DataMine: Application Programming Interface and Query Language for Data Mining. Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96), pp. 256-261

....discovered clusters) and database objects (e.g. tuples) 16] Existing KDD query languages are usually SQL like that contains extensions to handle KDD objects. A KDD object may not exist or may be pre generated and stored in a database, like a rule base that stores discovered association rules [17]. Since the number of discovered patterns can be extremely large, KDD querying should facilitate both the above two cases: the selective generation of patterns (i.e. the discovery of patterns with user defined constraints) and the management of previously generated results (i.e. the handling of ....

T. Imielinski, A. Virmani, A. Abdulghani, DataMine: Application programming interface and query language for database mining, Proceedings of International Conference on Knowledge Discovery in Databases and Data Mining (KDD'96), 1996, pp. 256 -- 262.

....property. The description of a non trivial mining process using these operations has been given and even if no concrete query language or query evaluation strategy is available yet, it is a mandatory step towards general purpose query languages for KDD applications. Query languages like M SQL [9] or MINE RULE [13] are good candidates for inductive database querying though they are dedicated to boolean and association rule mining, respectively. A simple Pattern Discovery Algebra has been proposed in [18] It supports pattern generation, pattern filtering and pattern combining operations. ....

T. Imielinski, A. Virmani, and A. Abdulghani. DataMine: Application programming interface and query language for database mining. In KDD'96, pages 256 -- 261, August 1996. AAAI Press.

....hypotheses about future datapoints, expected consequences from the data, generalized data summaries, emerging global patterns, exceptions from hypothesized patterns, suspected errors and implied inconsistencies, hypothetical plans synthesized from the data, etc. Michalski, 1983; Han et al. 1996; Imielinski, Virmani, and Abdulghani, 1996; Michalski, 1999) A general diagram of an inductive database is presented in Figure 1. An inductive database implements new types of database operators that are based on methods for inductive inference developed in the fields of machine learning, statistics, and uncertain reasoning. These ....

....and have quite limited types of knowledge generation operators available. One of the exceptions to the Prolog based approach is M SQL, which extends the SQL data query language by adding to it the ability to query for certain types of rules and to invoke an association rule generating operator (Imielinski, Virmani, and Abdulghani, 1996). KGL 1 differs from M SQL in that it is able to define complex data mining plans that may involve many different types of knowledge generation operators, and more closely resembles a programming language than a query language. A language somewhat related to KGL 1 is KQML, which provides means by ....

Imielinski, T., Virmani, A. and Abdulghani, A., "DataMine: Application Programming Interface and Query Language for Database Mining," Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, 256-261, 1996.

....uses the constraints to optimize the association rule mining process. The idea of using constraints in the rule mining process is important as it avoids generating irrelevant rules. Along the similar line, there are also a number of works based on data mining queries. For 4 example, M SQL in [8], DMQL in [7] and Metaqueries in [26] A data mining query basically defines a set of rules of a certain type (or constraints on the rule to be found) To execute a query means to find all rules that satisfy the query. All the above methods view the process of finding subjectively interesting ....

....rule mining algorithm in [28] Those redundant and or insignificant rules are removed using the pruning technique in [14] objective interestingness analysis) Since there is no existing technique that is able to perform our task, we could not carry out a comparison. Most existing methods [10, 7, 8, 18, 20, 29] only produce conforming rules but not unexpected rules. Although the system described in [23, 24] produces unexpected association rules, it is not an interactive post analysis system, and it does not handle RPC and GI specifications. As the proposed technique deals with subjective ....

. Imielinski, T., Virmani, A. and Abdulghani, A. "DataMine: application programming interface and query language for database mining." KDD-96, 1996.

....data tables, and that mining is an essentially interactive process, where a user repeatedly poses new queries based on what he found in the answers of his previous queries. In our opinion, the idea of inductive database indicates the ultimate goal of how a transparent data mining query language [4, 5, 6, 7, 12] should look like. The transparency lies in that the user never issues explicit mining commands himself; the system mines whatever and whenever necessary. Clearly the implementation of this vision presents a great challenge. In this paper, we investigate and compare two rather extreme approaches, ....

T. Imielinski, A. Virmani, and A. Abdulghani. DataMine: Application programming interface and query language for database mining. In Simoudis et al. [14], pages 256--261.

....data tables, and that mining is an essentially interactive process, where a user repeatedly poses new queries based on what he found in the answers of his previous queries. In our opinion, the idea of inductive database indicates the ultimate goal of how a transparent data mining query language [5, 4, 6, 7, 12] should look like. The transparency lies in that the user never issues explicit mining commands himself; the system mines whatever and whenever necessary. Clearly the implementation of this vision presents a great challenge. In this paper, we investigate and compare two rather extreme approaches, ....

T. Imielinski, A. Virmani, and A. Abdulghani. DataMine: Application programming interface and query language for database mining. In Simoudis et al. [14], pages 256-261.

....reasonable because association rules and sequential patterns are very often mined in the same datasets. MineSQL is designed as a query language for advanced users but it can also serve as an Application Programming Interface (API) for building business applications dealing with knowledge discovery [7]. MineSQL provides mechanisms for storing patterns in relational tables by offering new complex data types. MineSQL allows a user to specify various constraints defining the requested class of patterns (statistical constraints, time constraints, and item constraints) Current algorithms do not ....

Imielinski T., Virmani A., Abdulghani A.: Datamine: Application programming interface and query language for data mining. Proc. of the 2nd Int'l Conference on Knowledge Discovery and Data Mining (1996).

....for an attribute provides a tree which can be ascended until an apropriate level of generality is found. DBMINER [13] A system that implements a a wide spectrum of data mining functions incorporating several interesting data mining techniques remarking attribute induction. DATAMINE [14]: It is the first prototype of a system which provides an API to facilitate building complex data mining applications. RECON [22] It performs top down data mining using a deductive database. On the other hand, bottom up data mining is performed using a rule induction system and a data ....

T. Imielinski et al., DataMine: Application Programming Interface and Query Language for Database Mining, In Proceedings The Second International Conference on Knowledge Discovery and Data Mining, pp. 256 - 261, August 1996.

....on top of a database system, we believe this approach is superior to incorporating post processing features in KDD systems from scratch every time in an ad hoc manner. Particular proposals which are naturally subsumed by our approach include the rule querying feature of the DataMine system [10], the mining conditions of Meo et al. 13] the rule templates of Klemettinen et al. 11] and the rule covers and groupings of Toivonen et al. 16] Furthermore, we show that also specific post processing needs of particular applications are covered by our approach, by presenting a concrete ....

....Note that this query actually involves both the rule table and the data table. As such it is not a purely structural filter, going beyond strict post processing of generated rules. The possibility of post processing rules by relating them back to the data was also hinted at by Imielinski et al. [10]. Expressing such operations does not pose a problem in our approach, since we store the rule table in the same database as the data tables. Performance considerations Performance wise, global filtering operations, such as the last two OQL queries shown, are rather heavy. A powerful database ....

[Article contains additional citation context not shown here]

T. Imielinski, A. Virmani, and A. Abdulghani. DataMine: Application programming interface and query language for database mining. In Simoudis et al. [15], pages 256--261.

....with the best performance obtained in cache mine which caches and mines the query results on a local disk. SQL like operators for mining association rules have also been developed (Meo et al. 1996) Further, proposals for data mining query language (Han et al. 1996; Imielinski and Mannila 1996; Imielinski et al. 1996; Siebes 1995) have emerged. We note that most of this work is targeted for serial environments. PKDD e orts will bene t from this research, but the optimization problems will of course be di erent in a parallel setting. Some exceptions include the parallel generic primitives proposed in (Freitas ....

T. Imielinski, A. Virmani, and A. Abdulghani. DataMine: Application programming interface and query language for database mining. In Int'l Conf. Knowledge Discovery and Data Mining, August 1996.

....property. The description of a non trivial mining process using these operations has been given and even if no concrete query language or query evaluation strategy is available yet, it is a mandatory step towards general purpose query languages for KDD applications. Query languages like M SQL [12] or MINE RULE [17] are good candidates for inductive database querying though they are dedicated to boolean and association rule mining, respectively. A simple Pattern Discovery Algebra has been proposed in [23] It supports pattern generation, pattern filtering and pattern combining operations. ....

T. Imielinski, A. Virmani, and A. Abdulghani. DataMine: Application programming interface and query language for database mining. In KDD'96, pages 256 -- 261, August 1996. AAAI Press.

....discovered knowledge and identify some of the attributes of the knowledge itself. Earlier efforts to develop a language to assist in knowledge discovery tasks have been almost exclusively logic based, using Prolog style queries (e.g. LDL) One exception is an extension to SQL, called M SQL (Imielinski, Virmani and Abdulghani, 1996), which allows a user to query for certain types of rules and invoke a rule generating operator. KGL differs from M SQL in that it is able to call upon many different types of knowledge generation operators, and also in that it is designed to be less tightly coupled with SQL (although an SQL ....

Imielinski, T., Virmani, A. and Abdulghani, A., "DataMine: Application Programming Interface and Query Language for Database Mining," Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, Portland, OR, 1996, pp. 256-261.

....set has 770 rules) Without the proposed system, it would be very hard for us to analyze these large numbers of rules. 5. Related Work Traditionally, a query based approach is used to help the user identify or generate interesting rules. The approach takes many forms, e.g. templates [6] M SQL [5], DMQL [4] and action hierarchy [1] Although query languages can be quite different, a query basically defines a set of rules of a certain type (or constraints on the rules to be found) To execute a query means to find all rules that satisfy the query. We believe that the query based approach ....

Imielinski, T., Virmani, A. and Abdulghani, A. "DataMine: application programming interface and query language for database mining." KDD-96, 1996.

....version of KGL is a command based language whose constructs take on the form verb object parameters . Most previous efforts to create a high level language for multistrategy learning and knowledge discovery have taken a Prolog style logic programming approach. One notable exception is M SQL (Imielinski, Virmani and Abdulghani, 1996), which extends the SQL data query language by adding the ability to query for certain types of rules and invoke a rule generating operator. The KGL knowledge generation language differs from M SQL in that it is able to call upon many different types of knowledge generation operators, and also ....

Imielinski, T., Virmani, A. and Abdulghani, A., "DataMine: Application Programming Interface and Query Language for Database Mining," Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, Portland, OR, pp. 256-261, 1996.

....DM is an interdisciplinary subject, whose core lies at the intersection of machine learning and databases. Four desirable characteristics of a DM system are: 1) the discovery of comprehensible knowledge, typically expressed by high level rules; 2) integration with databases [Han et al. 96] Imielinski et al. 96] 3) a high degree of autonomy, necessary to discover knowledge previously unknown by the user [Zytkow 93] 4) the efficiency of the knowledge discovery process, necessary to cope with large databases. This paper proposes a genetic programming (GP) framework for DM that addresses, to some ....

T. Imielinski, A. Virmani and A. Abdulghani. DataMine: application programming interface and query language for database mining. Proc. 2nd Int. Conf. Knowledge Discovery & Data Mining, 256-261. AAAI Press, 1996.

....discovery assistant developed by Klosgen [52] IMACS is a data mining system developed at AT T Laboratory by Brachman et al. 13] using sophisticated knowledge representation techniques. DataMine is system exploring interactive ad hoc query directed data mining, developed by Imielinski, et al. [45]. IDEA, developed at AT T Laboratory by Selfridge, et al. 74] performs interactive data explorations and analysis. There have been many other data mining systems reported by machine learning and statistics researchers. Moreover, data warehousing systems have been seeking data mining tools for ....

T. Imielinski and A. Virmani. DataMine -- application programming interface and query language for kdd applications. In Proc. 1996 Int'l Conf. on Data Mining and Knowledge Discovery (KDD'96), Portland, Oregon, August 1996.

No context found.

T. Imielinski, A. Virmani, and A. Abdulghani. Datamine: Application programming interface and query language for database mining. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD'96), Portland, Oregon, August 1996.

....rule generation involves trivially flattening an attribute A with k values into k boolean attributes, each being a predicate of the form (A = a i ) However, this approach further multiplies the number of attributes, increasing the problem size substantially. The the DISCOVERY BOARD system [5] generates such generalized association rules directly from relational data, which can be persistently stored and later filtered using rule queries. We attempt to parallelize this algorithm, and present our observations and results in this paper. The goal of our research is cost effective ....

T. Imielinski, A. Virmani, and A. Abdulghani. Datamine: Application programming interface and query language for database mining. In Proceedings of the Second International Conferenceon KnowledgeDiscoveryand Data Mining (KDD'96), Portland, Oregon, August 1996.

No context found.

Imielinski T., Virmani A., Abdulghani A.: Datamine: Application programming interface and query language for data mining. Proc. of the 2nd KDD Conference (1996)

No context found.

T. Imielinski, A. Virmani, A. Abdulghani. Datamine: Application programming interface and query language for data mining. In Proc. of the 2nd KDD Conference, 1996.

No context found.

Imielinski T., Virmani A., Abdulghani A.: Datamine: Application programming interface and query language for data mining. Proc. of the 2nd KDD Conference (1996)

No context found.

Tomasz Imielinski, Aashu Virmani, and Amin Abdulghani. "DataMine: Application Programming Interface and Query Language for Database Mining." In Evangelos Simoudis, Jia Wei Han, and Usama Fayyad, editors, Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96), p. 256. AAAI Press, 1996.

No context found.

Tomasz Imielinski, Aashu Virmani, and Amin Abdulghani. DataMine: application programming interface and query language for database mining. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96), pages 256--262. AAAI Press, 1996.

No context found.

Imielinski T., Virmani A., Abdulghani A.: Datamine: Application programming interface and query language for data mining. Proc. of the 2nd KDD Conference (1996)

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