Design of Intelligent layer for flexible querying in databases

Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 30 Design of Intelligent layer for flexible querying in databases Mrs. Neelu Nihalani Dr. Sanjay Silakari Dr. Mahesh Motwani Reader, Computer Applications, Prof. & Head, Dept. of CSE, Reader, Dept. Of CSE UIT, RGPV, UIT, RGPV, JEC, Jabalpur Bhopal, MP India Bho pal, MP, India Jabalpur, MP, India neelurgpv@ yahoo. co.in ssilakari@yahoo. com mahesh_9@hotmai l.com Abstract - Computer-based info rmation technologies have been extensively used to help many organizations, private companies, and academic and education institutions manage their processes and information system s hereby become t heir nervous centre. Th e explosion of ma ssive data sets created by businesses, science and governments necessitates intelligent and more powerful computing paradigms so that users can benefit from this data . Therefore most new-generation d atabase applications demand intelligent information management to en hance efficient interactions between database and the users. Database systems support only a Boolean query model. A selection query on SQL database returns all those tuples that satisfy the conditions in the query. But lately, there is an overwhelmin g need for non-expert users to query relational databas es in their natural language using linguistic variables and terms instead of work in g with the values of the attributes. As a result, intelligent database s have emerg ed, which prov ides expanded and more flexible optio ns for manipulating queries. In this paper, we propose an intelligent layer for database which is responsible for manipulating flexible queries. Initially, the flexible queries from users in t heir natural langua ge are submitted to intelligent layer and this layer converts th e amorphous query into a structured SQL query. The shaped q uery is executed and the results are pr esented to the user. Afterw ards, on the basis of results, feedback and the acceptance or reje ction of the results are request ed from the us er. It enables the design of a knowledge based self learning system based the values obtained from user, which will aid the selection of appropriate SQL query, when a same flexible query is issued in the future. The experimental results demonstrate the effectiveness of the proposed intelligent databa se syste m. Keyword s- Data bases , Databa se Management System ( DBMS); Structured Query Language (SQL); Artificial Intelligence (AI);Intelligent database (IDB); Intelligent Database System (IDBS); Flexible Q uerying; Intelligent Layer . I. INTRODUCTION Databases are gaining prime importance in a huge variety of application areas employing private and public information systems. A general information management sy stem that is capable of managing several kinds of dat a, stored in the database is known as Databa se M anagem ent System (DBMS) [1]. The D BMS grants support for logical views of data that are separate from the physical v iews, i.e. how the d ata is actually stored in the da tabase. By permitting applications to de fine, access, and update data through a Data Definition Langua ge (DDL) and Data Mani pulat ion Language (DML) com bined into a ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 31 declarative query language such as the relational query language SQL [2] the separation is acco mplished. Structured Query Language (SQL) is an ANSI standard for accessing and manipulating the information stor ed in relati onal database s. It i s comp rehens ivel y employed in industry and is supported by major database management systems (DBMS) [24]. Most of the languages used for manipulating relational database systems are based on the norms of SQL. They work on the basis of Boolean interpretation of the queries: a logical expression is the only accepted se lection criterion and the response always encompasses only these tuples for what th e expression results in a true value [23]. But some user requirements may not be answered explicitly by a classi c querying system. It is due to the fact that the requirements’ characteristics cannot be expressed by regular query languages. Many novel-generation database applications stipulate int elligent inform ati on ma nageme nt necessi tati ng efficient interactions between th e users and database [13]. In recen t times, there is a rising demands for non-expert users to query relational databases in a more natural language encompassing ling uistic variables and ter ms, instead of operating on the values of the att ributes. Intelligent database systems, a promi sing approach, enhance the users in performing flexible querying in databases. The r esearch and advancement of intelligent databases have lately emerged as a br and new discipline and have fascinated the attention of numbers. II. LITERATURE REVIEW Our work has been inspired by a number of works available in the literature related t o intelli gent aspects of d atabase systems. The field of intell igent databas e and information systems has achieved remarkable growth in the last few decades. Advancements in Intelligent d atabases focus on two vital issues namely –  Intelligent information processing in databases.  Intelligent as pects of databases A. Issues of Intelligent i nformation Processing in Databases Intelligent information processing has emerged as one of the major considerations on our course to achieve a knowledge society. Recen t research in intelligent information processing has paved way for the evolution of thrilling technologies that will mould our future. Intelligent Informat ion Processing is defined as a study on fundamental theory and advanced technology of intelligence and knowledge for information processing. Knowledge-based systems have qualified to offer services in a well-founded ontological fra mework and there are a number of tools available to support intelligent know ledge management. Th e techniques of Artificial Intelligence can serve as effective tools in this context. The intelligent sy stems have a wide rang e of applications ranging from surfing through the Internet and data-mining, interpreting Internet-derived material, the human We b interface, remote condition monitoring and many other regions [11]. In spite of these applications there are a number of noticeable issues related to intelligent information processing in databases: A good number of algorithms and approaches on data mining [30, 31], especially association rule m ining based algorithms work on the assumption that all items are positively correlated and it considers only those ite ms that remain ed at last in a shopping basket. Kouris, Makris, and Tsakalidis [32, 33] investigated th e mining process by taking valuable information from reject ed item s and h ave proposed a number of alternatives for taking the spe cific item s into ac count ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 32 efficiently. Another important area of research in Data mining is Outlier detection. Zhao, Bao, Sun, and Yu [12] found the existence of a number of empty cells that are ineffective to outlier detection. They ca me up with a novel index structure, called CD-Tree which stores only the non empty cells and adopts a clustering technique to store the data objects in the same cell into linked disk pages. Sampaio et al. [14 ] proposed an integrated architecture for a Spatial Data warehouse (SDW), including a formalized data model for SDW, a SQL extension query language which facilitates spatial roll-up and drill- down, optimization techniques to improve the performance of co mplex spatia l queries by pre-storing spatial aggregates, and a prototype, Map Warehouse, which validates the ideas proposed. Zarri [15] demonstrated the ubiquity of the “narrative information” and stresses the importance of the same by showing that the traditional ontological tools cannot repr esent and exploit the narrative information to provide comple te and re liable solutions. He also describes the NKRL (Narrative Knowledge Representation Langu age), a fully-implemented knowledge representation and an i nference environment specificall y created for an “inte llige nt” expl oita tion of narra tive knowledge. The primary goal of most database researches is to incorporate new and related sem antics t o the data model. Most traditional data models suffer from their inabi lity to mani pulate impreci se and vague information that occur in most real world applications. So, we employ the fuzzy set theory in distinct dat a models and has seen effective s olutions for relational and its related models. To comply with modeling of co mplex objects with imprecision and uncertainty, recent researchers have turned their focus on fuzzy semantic (conceptual) and object- oriented data models. Ma [16] reviewed the fuzzy database modeling technologies, including fuzzy conceptual data models and dat abase model s. Rega rding fuz zy database models, a brief discussion on fuzzy relational databases and fuzzy object-oriented databases is done. B. Intelligent Aspects of Database A brief review of some of the works related to intel ligent aspects of database systems is illustrated below: Wolff [1 7] em ploys th e SP theor y of com putin g and co gnitio n to describe som e different kinds of “intelligence” exhibited by an intellig ent database sy stem. The author introduces the SP theory and its main credits is what that forms the basis for an intelligent database sy stem: that it uses a simple format for diverse kinds of knowledge which is versatile, that it integrates and simplifies a series of AI functions, and that it supports the already esta blishe d data base m odels whe n required. The author illustrates the various aspects of “ intelligenc e” in the system: pattern recognition [18] and information retrieval, several forms of probabili stic reasoning, the analysis and production of nat ural la nguage , and t he uns upervi sed learning of new knowledge, based on a number of examples. One of the essential chara cteristic of intelligent database management syste ms is the ability to provide auto mated support to users to maintain the seman tic correctness of data in co mpliance to th e integrity constraints. These integrity constraints are a vital means to characterize the well-for medness and semantics of the information stored in databases. Martinenghi, Christiansen, and Decker [19] gave an ov erview of the field of efficient integr ity checking and maintenance for relational and deductive databases [20]. The work describ es both theoretical and practical asp ects of integrity control, including int egrity ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 33 maintenance via active rules. The authors delineate novel scopes of research, particularly with regard to: integrity in XML document [22] collections and in distributed databases, where a strong impact for future de ve lopments can be expected. These lin es of research pose a number of new and highly relevant research challenges to the d atabase community [21]. III. INTELLIGENT DATAB ASE SYSTEM (IDBS) Int ellig ent da tabase (IDB) syst ems integrate the resources of both RDBMS s and KBSs to offer a natural way to deal with information, making it easy to store, access and apply [4], [5]. The book “Intelligent D atabases” by Kamran Parsaye, Mark Chignell, Setrag Khoshafian and Harry Wong in 1989 was the first to refer the term Intelligent Database [4]. It reco mmended three levels of intelligence for database systems: High level tools, the user interface and the data base e ngine. The hi gh le vel tool s are meant fo r admi niste ring dat a qual ity and unraveling useful relevant patterns automati cally by employ ing a process called Data mining. This layer is highly dependent on the use of artificial intelligence techniques. Intellig ent databases encompasses of artifici al intelligence components that aids in the intellectual operation of the search, provide means of representing knowledge, and are based on connectionist neural network models [6]. The tasks to be addressed by the intelligent databases are highly complicated, if not impossible, for a human mind to cope with. The ta sks involve searching for and extracting meaningful information across a huge data set. . It would be extremely impossible for human m inds t o deduc e, in duce or i nfer any significant new data from the vast data repositories with the efficien cy or speed that machine intellig ences in the shape of "intelligent" databases achieve [7]. Artificial intelligence is very much able at addressing the difficulties that people are very bad at and perhaps in this context, we consider the "intelligen t" databases [7]. Recent researchers in the fie ld of intelligent databases namely Bertino, Catania and Zarri [3] proposed a me ans to incorporate two technologies explicit; "Intelligent database sy stems (IDBS) built from the integration of database (DB) technology with techniques developed in the field of artificial i ntelligence (AI)". Their work also analyzes the inherent weaknesses of the technologies when used in isolation, the traditional databases lacking any semantic value and the inability of artificial int elligence meth ods to deal with large-data sets. It has been stated by Kamran Parsaye and Mark Chignell [25] that the int elligent databases represent the evolu tion and merg er of several technologies including automatic discovery, Hypermedia, object orientation, expert systems and traditiona l databases. An intelligent database affords expanded and more flexible options for query ing. For example, a user is all owed to type in a question as an imperative sentence. The database then provides a list of hits arranged according to the likelihood (from highest to lowest) that the resulting data contains a useful answer to that question. The Artific ial Inte lligence (AI) may correct the suspected errors (such as inaccurate spelling) in th e input provided by the user [8]. A smal l numbe r of intell igent databases presen t synony ms (items with similar meanings) or antonyms (items with opposite or negative meanings) for keywords and phrases. In order to exploit maximum benefit from an intelligent database, the us er must formulate queries with forethought, phrasing them with care, just as it is necessary when interrogating a person [8]. As noted in [9], AI/DB integration is crucial for next generation computing, the continued development of ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 34 DBMS technology and for the effective application of much of AI technology [10]. The motivations driving the integration of these two technologies includes, the need for (a) access to hefty amounts of shared data for knowledge processing, (b) effective manage ment of data as well as knowledge, and (c) intelligent processing of data. In addition to these driving factors, the design of Intelligent Database Interfac e (IDI) was m otivated by the desire to preserve the substantial assets represented by most existing databases. Several general approach es to AI/DB integration have been investigated and stated in th e literature [26, 27, 28, 29]. In this research, we propose to develop an intelligent layer which can be incorporated with the exis ting database system, which is responsible for the inte llige nt info rmat ion pr oces sing a nd performing flexible queries. The user queries are given in a m ore conversing language using linguis tic variables and terms. The intelligent la yer designed in our scheme processes the unqualified user query and constructs a S tandard SQL query from it. Initially, the conj unctive clauses are identified in th e user query with the help of the conjunctive training set. Afterwards, on the basis of the identified conjunction, the flexible query is divided into two parts: Subjective/Display part and Criteria part. Th e subjective/display part contains information about tables and criteria part contains the information about conditions and field names. The expression mapping is carried out in criteria part, which converts expressions into corresponding mathematical symbols. Following expression mapping, the stop words ar e removed from both the parts of t he query. Now, the subjective part contains the table name a nd the c riteri a part cont ains field names and conditions. The next step is to locate the associ ated tables and fields from the database. First, we scan through the metadata set for tables to identif y the corresponding tables. If the search is unsuccessful, we go for the Ontology based semantic matching, or else the computation of Levenshtein distance, which is a metric employ ed for measuring the difference between two sequences. The aforesaid procedure is also employed to identify the corresponding field names. From the above data, the SQL query is constructed. The results of the formed SQL query are presented to the user. Subsequently, feedb ack and the acceptance or re jections of the results are obtained from the user based on the presented results. A knowledg e based self learning system is designed with the values obtained from user, which will aid the selection of appropriate SQL query, when a same flexible qu ery is is sued in the future. The experimental results demonstrate the effectiveness of th e presented intelligent dat abase system. Th e rest of the paper is organized a s follows: A brief review of some of the works in t he literature related to intellig ent database system is given in Sect ion 2. The proposed intelligent database sy stems for flexible querying and the know ledge based self learning system are explained in Section 3. The results of the proposed IDBS are presented in Section 4. Finally, the conclusions are su mmed up in Sectio n 5. IV. THE PROPOSED INTELIGENT LAYER FOR DATABASES Wi th a dvance s and i n-dee p applications of computer technologies, in particular, the extensive applications of Web technology in various areas, databases have become the repositories of large volumes of data.. It is known th at databases respond only to standard S QL queries and it is highly impossible for a comm on person t o be we ll vers ed in SQL querying. Moreover they may be unaware of the database structures namely table formats, their fields with corresponding types, primary keys and more. On account ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 35 of these we design an intelligent layer which accepts common user’s imperative sentences as input and converts them in to standard SQL queries to retrieve data from relational databases based on knowledge base. The primary advantage of the system is that it conceals the inheren t complexity involved in information retrieval based on unqualified user queries. In general, a database ( D ) is termed as set of tables organized in some common struc ture . The vita l inform ati on that b riefl y desc ribes t he ta bles i n the database is organized into a metadata set ( M ). The metadata set holds entries for all the ‘ n ’ tables in the relational databas e with all their corresponding fields and thei r uniq ue prim ary ke y. The proposed approach e mploys a set of predefined training structures. The primary benefit of these training sets is that they can be expanded or appended when the intelligent i nformation system discovers so me new knowledge. The significant training sets used are: Expression mapping ( map E ), Conjunction set ( T C ), Semantic set ( S ) for tables and fields are T S and F S respectively, Stop- words set ( W S ). The Expression mapping set ( map E ) contains the list of commonly used conditional clauses and their associated mathematical sym bols. It acts as a look up table t o locate the SQL defi ned mathematical operators. The Conjunction training set ( T C ) consists of the list of generally used Conjun ctive clauses lik e where, who etc. These conjunctiv e clauses determine the exact Query definition. When the syste m encounters a relatively new conjunctive clause, it is appended to the existing training set. The trained stop word set ( W S ) contains the list of all common stop words that are lik ely to occur i n a user type d qu ery. T he se m antic set ( S ) contains the list of al l possible semantics related to table names and fields in t he databa se. A. USER QUERY CONTRIBU TION The following subsection gives a vivid description of how the user query is transformed to be used for data retrieval from databases. In our proposed approach, we define a universal set u Q which holds all the individual tokens in the user typed query. Every token represents a unique element in the universal set u Q . Every user query is likely to contain a display or subjective part which specifies t he intended result, the Conjunction part which determines the SQL definition Clause and the Criteria part which describes the cond ition or constrai nt. All these parts of the user qu ery will be represented as three distinc t sets d P , j P and c P . Here the sets d P , j P and c P contain tokens that represen t the subject, the Conjunctive clause and the Criteria respectively. u Q  User Query d P  Subjective/Display part of u Q j P  Conjunctive part of u Q c P  Crite ria/c onditi on pa rt of u Q By using the set of predefined trai ning st ruct ures ,w e have determ ine d the Sets d P , j P and c P respectively. Then, we analyze the set c P to locate the conditional clauses provid ed by the user. The elements of the set c P are intersected wi th the trained expression mapping set to map t he corresponding mathematical symbols. Following the expression mapping process, we go for the stop words removal process. This process is meant for eliminating all those words in the user ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 36 query that are ineffective for SQL quer y construction. Intersection of the sets d P and c P with the trained stop word set w S will yield a non-empty set of stop words to be eliminated from user query. The above process removes all the stop words from the sets d P and c P . Now probably, the set d P contains the name of the table, the set c P contains the field name and the condition and th e set j P contains the conjunctiv e clause from the user query. B. SUB JECTIVE PART RET RIEVAL The next step is to locate the tabl e mentioned in th e set d P from M . The set d P is intersected with the metadata set M. If it yields a non-em pty set, d P itself is chosen as the table. Else if it y ields an empty set, then th e element of singleton set j P is intersected with the set T S to retrieve the appropriate table name associated with the match ing semantics. Else the dis tance measure is co mputed between the ele ment of j P and every individual element of the set T S . The element in T S that cor rela tes w ith minimu m Euclidean distance is chosen as the semantics. Finally, the table name associated with the semantics is chosen. T S  Tables in S F S  Fie lds i n S  map S T S Mapping t d P  Table ) ( d P Subsequently, we will have to locate th e set of related tables which might contain the field name mentio ned in the user query. Initially, we scan the set of fields found in t d P . The procedure is illustrated as below: f c P  Fie lds i n c P c t  Table of f c P Case: 1    )) ( ( t d f c P f P t d c P t  Case: 2   c t Semantic Map of   ) ( ), ( t d F f c P S P otherwise   c t Distance m easure of   ) ( ), ( f c t d F P P S If the above procedure results in a non- empty set c t , then to find the related tables, we select the pri mary key k P of the table t d P from the m etadata set and intersect it with all el ements in the set F S . If the intersection yields a non-e mpty set, then the associated tables are appended to a set t . Wher e t is set of all tables having p K as a field; where T t  . Obviously, the nu mber of related tables found for large databases will be enormous. Now to locate a particular field from the se relat ed tabl es is a tedi ous task, as some of the tables found may be irrelevant to the user query. So, in our proposed approach we introduce a nove l step to refine the table set t found. Here, we compute a set V which contains the set of all values corresponding to the field p K in t . If the value set V found for tables in t is a subset of the value set of p K in t d P , then append t to t . Else, eliminate i t . Subsequently, we go in for the locating the particular fiel d from the set of relate d tables t . The same procedure e mployed to locate the tables is utilized. First, th e field ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 37 name T F obtained from the user query is checked in the M followed by the application of semantics and calc ulation of Euclidean distance. Finally, we have deduced the appropriat e table name t , field name c t and the conditional clause q C from the user query. With all the above information we construct the SQL query to retrieve the requested data from the Database D . C. SEMANTIC MATCHING Semantic matchi ng is a tec hniq ue used to categorize information which is semantically related. S emantic matching is employed as a fundamental technique in many applications areas such as resource discovery, data integration , data migration, query translation, peer to peer networks, agent co mmunica tion, schema and ontology merging. In fact, it has be en accepted as a valid solution to the semanti c heterogeneity problem, namely managing the diversity in knowledge. Interoperability among people of different cultures and languages, having different viewpoints and using different terminology has been a massive problem for years. Especially with the advent of th e Web and the consequential information explosion, the problem seems to be more emphasized. D. DISTANCE MEASURE The Levenshtein distan ce is a metric employed for measuring the difference between two sequences (i.e., the so called edit distan ce). The Levens htein distance between two strings is defined by the minimum number of operations needed to transform from one string to the other, where an operations can be insertion, deletion, or substitution of a single character. A generalization of the Levenshtein distance (Damerau– Levenshtein distance) permits the transposition of two characters as an operation. V. EXPERIMENTAL RESULTS In this section, we have presented the experimental results of the proposed IDBS. The present ed IDBS has been implemented in JAVA wit h MySQL and MS-Access as d atabases. The flexibl e user queries and the results obtained in the designed IDBS are as follow s: A. Flexibl e User query: List orders details where unitprice should be greater than 200 Gener at ed SQ L Q uer y: Select * from orders AS A, orderdetails AS B where A.Orde rID=B.OrderID and B.UnitPrice > 200 B. Flexible User query: List su pplier details where city is equal to London . Generated SQL Query: Select * from suppliers AS A where A. city=London Results of query A: Results of query B: ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 38 From the above pair of examples, we understand that when the SQL query complies with user query the results ar e expected to be correct. VI. CONCLUSION Relati onal Data Model is universally employed tool for construction of database systems and applications. A dominant technology for data storage and retrieval was deve loped by Relational Database Management Systems. Yet, these systems struggle the problem of rigid ity. Every user requirement cannot b e solved by the cl assic query ing system directly because of the requirements’ characteris tics that are not expressi ble by standard query languages. In this paper, we have presented an innovative approach for the design of an intelligent da tabase system for performing flexi ble queries in databases. An intelligent layer has been designed and incorpo rated into the existing database systems. The presented system accepts flexible user qu eries and converts them into a stand ard SQL query. Expression mapping, stop words removal, semantic matching and Levenshtein distance measure techniques hav e been utilized by t he intel ligent layer in t he formation of the SQL query . The usefulness of the presented system has been demonstrated with the aid of experim ental result s. VII. REFERENCES [1]. Elmasri R. and Navathe S. B., “Fundamentals of Databa se Systems”, Second Editi on, The Benja min/Cum mings Publishing Company, Inc., ISBN 0- 8053-1753- 8, 1994. [2] A strahan M. M., Blasgen M. W., Chamberlin D. D., Eswaran K. P., Grey J., Griffiths P. P., King W. F, Lorie R. A, Mc Jones P. R., Mehl J. W, P utzolu G. R., Tr aiger I. L., Wade B. W., and Wa tson V., “System R: Relatio nal Approach t o Database Managemen t”, ACM Transac tions on Da tabase System s, Vol.1, No. 2, June 19 76, Pages 97-137. [3] Bertino, B. Catania, G.P. Zarri, “Intell igent database sys tems”, Reading, Addsion Wesley Pr ofessional, 2001. [4] Kamran Parsaye, Mark Ch ignell, Setrag Khoshafia n and Harr y Wong, “I ntelligent databases-ob ject-orie nted, de ductive hypermedia technolog ies”, New Yor k, John Wiley& S ons, 198 9. [5] Abel, M., Silva, L.A.L., De Ros, L.F., Mastella, L.S., Campbell, J.A., e Novello, T., "Petrograp her: Ma naging pe terogra phic dat a and knowledg e using an intell igent data base applicatio n", Expert systems w ith Applicat ion, vol. 26, no. 1, S PECISS, P p. 9- 18, 2004. [6] "I ntelligen t Database" from http://en.wik ipedia.org/w iki/Intellige nt_Data base . [7] "Artifi cial Intelligen ce: Intellig ent Databases," Applied Kn owledge Research & Innovatio n (AKRI) Ltd., England, 2005. [8] "Intelligent Databa se" from http://se archsql server. techtar get.com/sD efinit ion /0,,sid8 7_gci1124415 ,00.html . [9] M. Brodie, “Future Intelli gent Infor mation Systems: AI and Databa se Technol ogies Working Toge ther” in Rea dings in Artific ial Intelligence and Databases, M organ Kaufman, S an Mateo, CA, 1988. [10] Donald P. Mc kay and Timothy W. Finin, "The Intellige nt Databa se Interfa ce: Integrating AI and Database sys tems", In Proceedings of the 19 90 Nationa l Confer ence on Artificia l Intellige nce, pp. 677- -684, 1990. [11] Muse n, Mark, Neum ann, Bernd, "Intelligen t Info rmatio n Proces sing ", Seri es: IFI P International Federation for Information Processing, Stud er, Rudi (Eds.), Vol. 93, pp. 328, 2002, I SBN: 978-1-4 020-7171- 3 [12] Huanliang Su n, Yubin Ba o, Faxin Zhao, G e Yu and Daling W ang, "CD-Trees: An ISSN : 0975-3397 Mrs. N eelu Nihalan i et al /Int ernationa l Journal on C omputer Science and En gineering Vol.1(2) , 2009, 30- 39 39 Efficient Index S tructure for Ou tlier Detection", Lecture Notes in Computer Science, Springer Berlin / Heidelber g, vol. 3129, 2004. [13] M. Koyuncu A. Yazici and R. Ge orge, “Flexible Quer ying in an Int elligent Ob ject- Oriented Dat abase Environm ent,” Proc. Fourth In t'l Conf. Flexible Query A nswerin g Systems (FQ AS '2000), pp. 75-8 4, Oct. 2000. [14] Marcus Costa S ampaio, Andr é Gomes de Sousa, Clá udio de S ouza Bap tista, "Tow ards a logical multi dimensio nal model for s patial data ware housing a nd OLA P", Da ta Warehous ing and O LAP, pp. 83- 90, 200 6. [15] Gian Piero Zarri, "Ontologi es and reasoning techniqu es for (l egal) int elligent info rmation retrieval sys tems", Artific ial Inte lligence and Law, Volume 15 , I ssue 3, Pages 251- 279, September 2007. [16] Z. M. Ma, Li Yan, “Fuzzy X ML data modeling with the UML and re lational data models”, Data & Knowledge Enginee ring, Vol.63, No. 3, pp.972-996, December , 2007. [17] J. Gerard Wolff, “Towards an intellige nt database syst em founded on the SP theory of computing a nd cogniti on”, Da ta & Knowledge E ngineeri ng, v.60 n. 3, p.59 6- 624, March, 2007 [18] J. Gerar d Wolff, “ Medical d iagnosis as pattern re cognition in a fram ework of inf orm atio n com pre ssi on by mul ti ple alignment”, unification an d search, D ecision Support Sy stems, vol. 42 no.2, pp.608-625 , November 200 6 [19] D. Martinenghi, H. Christi ansen, H. Decker, "Integrity Chec king and Ma intenance i n Relational and Deduct ive Databa ses - a nd Beyond", in the book Intel ligent Da tabases: Technologies a nd Applicati ons, chapter X, Z. Ma (Ed.), Idea Group Pub lishing, p p. 238- 285, 2006 [20] Decker, H.; Mart inenghi, D., "Ge tting Rid of Straitjackets for Fl exible Integrity Checking", 1 8th Interna tional Co nference on Database a nd Expert Sys tems App licati ons, DEXA apos;0 7, Page (s):360 - 364, 3- 7 Sept. 2007 [21] Antoni Oli vé, "Integr ity Constr aints Checking I n Deducti ve Databa ses", Very Large Data Bases, pp. 513-52 3, 1991. [22] Chr istiansen, H .; Rekouts, M. , "Integrity Checking a nd Mainte nance w ith Active Rules in XML Database s", 24th British National C onfere nce on Da tabases, BNCOD apos; 07, P age(s):59 - 67, 3-5 July 20 07. [23] Cornelia Tu dorie, Cr istian Ne acsu and I onel Manolach e,” Fuzzy Que ries in Romanian Language: A n Intell igent Inter face," The Annals of”DU NAREA D E JOS ” Universit y of Galati, Fascicle II I, 2005. [24] Dietmar Wo lfram, "Appl ications of SQL fo r Informetri c Data Processing ", Proceedings of the 33rd confe rence of the Canadian Association for Inf ormation Science, 2005 . [25] Bur cin Bosta n-Korpeo glu, Ad nan Yazici , “A fuzzy Pe tri net mode l for intellige nt databases”, Science Dir ect Data & Knowledge E ngineeri ng, vol. 62, no. 2, pp. 219–247, 20 07. [26] J. Bocca, “EDUCE a Marriage of Convenience : Prolo g and a Rela tional DBMS” Third S ymposium on Logi c Programmin g, Salt Lake City, Sept. 1 986, pp. 36 45. [27] U. Chakrava rthy, J. Minker, and D. Tran, “Interfa cing Predica te Logic Languages a nd Relation al Databases,” in Proc eedings of the First Internationa l Logic Pr ogramming Conference, pp.91 98, Septembe r 1982. [28] C. Chang, “ DEDUCE 2: F urther investiga tions of de duction in relational databases,” i n Logic and Da tabases, e d. H. Gallaire, pp.201 236, New Yor k, 1978. [29] C. Chang and A. Walker, “PROS QL: A Prolog Pr ogramming Inter face wit h SQL/DS,” Proc. Of the 1st Intl. Workshop on Expert Database Systems, Kiawah Island, South Caro lina, October 1984. [30] Tzung-Pei Hong , Ch ing-Yao Wa ng , Yu- Hui Tao, "A new incremental data minin g algorithm using pr e-large itemsets", Intelligent Da ta Analysis, vo l. 5, no. 2, Pp.111-1 29, Apri l 2001. [31] Oke, S.A. "Applica tion of Decis ion Tree as a Data mining Tool in a Manufac turing System", In Z. Ma, Intelligent Databases: Technologies and App lications (pp. 117- 136), Her shey, PA: I dea Group Pu blishin g, 2007 [32] I.N. Kouris, C.H. Makris, A. K. Tsakalidis, “An impr oved algorit hm for m ining association roles using m ultiple sup port values”, i n: Proc. of FLAIRS Internat. Conf. , St. Augus tine, FL , 2003. [33] Ioannis N. Kouris, Christos H . Makris, Athanasios K. Tsakalid is, "Efficie nt automati c discov ery of 'h ot' itemset s", Informatio n Process ing Lette rs, Volume 90 , Issue 2, Pages: 65 - 72, 2004 ISSN : 0975-3397