Dare Obasanjo




Introduction: XML and Data

XML stands for eXtensible Markup Language. XML is a meta-markup language developed by the World Wide Web Consortium(W3C) to deal with a number of the shortcomings of HTML. As more and more functionality was added to HTML to account for the diverse needs of users of the Web, the language began to grow increasingly complex and unwieldy. The need for a way to create domain-specific markup languages that did not contain all the cruft of HTML became increasingly necessary and XML was born.

The main difference between HTML and XML is that whereas in HTML the semantics and syntax of tags is fixed, in XML the author of the document is free to create tags whose syntax and semantics are specific to the target application. Also the semantics of a tag is not tied down but is instead dependent on the context of the application that processes the document. The other significant differences between HTML and XML is that the an XML document must be well-formed.

Although the original purpose of XML was as a way to mark up content, it became clear that XML also provided a way to describe structured data thus making it important as a data storage and interchange format. XML provides many advantages as a data format over others, including:

  1. Built in support for internationalization due to the fact that it utilizes unicode.
  2. Platform independence (for instance, no need to worry about endianess).
  3. Human readable format makes it easier for developers to locate and fix errors than with previous data storage formats.
  4. Extensibility in a manner that allows developers to add extra information to a format without breaking applications that where based on older versions of the format.
  5. Large number of off-the-shelf tools for processing XML documents already exist.
The world of traditional data storage and XML have never been closer together. To better understand how data storage and retrievel works in an XML world, this paper will first discuss the past, present, and future of structuring XML documents. Then we will delve into the languages that add the ability to query an XML document similar to a traditional data store. This will be followed by an exploration of how the most popular RDBMSs have recognized the importance of this new data storage format and have integrated XML into their latest releases. Finally the rise of new data storage and retrieval systems specifically designed for handling XML will be shown.

Structuring XML: DTDs and XML Schemas

Since XML is a way to describe structured data there should be a means to specify the structure of an XML document. Document Type Definitions (DTDs) and XML Schemas are different mechanisms that are used to specify valid elements that can occur in a document, the order in which they can occur and constrain certain aspects of these elements. An XML document that conforms to a DTD or schema is considered to be valid. Below is listing of the different means of constraining the contents of an XML document.

  <gatech_student gtnum="gt000x">
   <name>George Burdell</name>

  1. Document Type Definitions (DTD): DTDs were the original means of specifying the structure of an XML document and a holdover from XML's roots as a subset of the Standardized and General Markup Language(SGML). DTDs have a different syntax from XML and are used to specify the order and occurence of elements in an XML document. Below is a DTD for the above XML fragment.
      <!ELEMENT gatech_student (name, age)>
      <!ATTLIST gatech_student gtnum CDATA>
      <!ELEMENT name (#PCDATA)>
      <!ELEMENT age (#PCDATA)>
    The DTD specifies that the gatech_student element has two child elements, name and age, that contain character data as well as a gtnum attribute that contains character data.

  2. XML Data Reduced (XDR): DTDs proved to be inadequate for the needs of users of XML due to to a number of reasons. The main reasons behind the criticisms of DTDs were the fact that they used a different syntax than XML and their non-existent support for datatypes. XDR, a recommendation for XML schemas, was submitted to the W3C by the Microsoft Corporation as a potential XML schema standard which but was eventually rejected. XDR tackled some of the problems of DTDs by being XML based as well as supporting a number of datatypes analogous to those used in relational database management systems and popular programming languages. Below is an XML schema, using XDR, for the above XML fragment.
      <Schema name="myschema" xmlns="urn:schemas-microsoft-com:xml-data"
       <ElementType name="age" dt:type="ui1" />
       <ElementType name="name" dt:type="string" />
       <AttributeType name="gtnum" dt:type="string" />
       <ElementType name="gatech_student" order="seq">
        <element type="name" minOccurs="1" maxOccurs="1"/>
        <element type="age" minOccurs="1" maxOccurs="1"/>  
        <attribute type="gtnum" />
    The above schema specifies types for a name element that contains a string as its content, an age element that contains an unsigned integer value of size one byte (i.e. btw 0 and 255), and a gtnum attribute that is a string value. It also specifies a gatech_student element that has one occurence each of a name and an age element in sequence as well as a gtnum attribute.

  3. XML Schema Definitions (XSD) : The W3C XML schema recommendation provides a sophisticated means of describing the structure and constraints on the content model of XML documents. W3C XML schema support more datatypes than XDR, allow for the creation of custom data types, and support object oriented programming concepts like inheritance and polymorphism. Currently XDR is used more widely than than W3C XML schema but this is primarily because the XML Schema recommendation is fairly new and will thus take time to become accepted by the software industry.
      <schema xmlns="" > 
       <element name="gatech_student">
          <element name="name" type="string"/>
          <element name="age" type="unsignedInt"/>    
         <attribute name="gtnum">
           <restriction base="string">
            <pattern value="gt\d{3}[A-Za-z]{1}"/>
    The above schema specifies a gatech_student complex type (meaning it can have elements as children) that contains a name and an age element in sequence as well as a gtnum attribute. The name element has to have a string as content, the age attribute has an unsigned integer value while the gtnum element has to be matched by a regular expression that matches the letters "gt" followed by 3 digits and a letter.

The above examples show that DTDs give the least control over how one can constrain and structure data within an XML document while W3C XML schemas give the most.

XML Querying: XPath and XQuery

It is sometimes necessary to extract subsets of the data stored within an XML document. A number of languages have been created for querying XML documents including Lorel, Quilt, UnQL, XDuce, XML-QL, XPath, XQL, XQuery and YaTL. Since XPath is already a W3C recommendation while XQuery is on its way to becoming one, the focus of this section will be on both these languages. Both languages can be used to retrieve and manipulate data from an XML document.
  1. XML Path Language (XPath): XPath is a language for addressing parts of an XML document that utilizes a syntax that resembles hierarchical paths used to address parts of a filesystem or URL. XPath also supports the use of functions for interacting with the selected data from the document. It provides functions for the accessing information about document nodes as well as for the manipulation of strings, numbers and booleans. XPath is extensible with regards to functions which allows developers to add functions that manipulate the data retrieved by an XPath query to the library of functions available by default. XPath uses a compact, non-XML syntax in order to facilitate the use of XPath within URIs and XML attribute values (this is important for other W3C recommendations like XML schema and XSLT that use XPath within attributes).

    XPath operates on the abstract, logical structure of an XML document, rather than its surface syntax. XPath is designed to operate on a single XML document which it views as a tree of nodes and the values returned by an XPath query are considered conceptually to be nodes. The types of nodes that exist in the XPath data model of a document are text nodes, element nodes, attribute nodes, root nodes, namespace nodes, processing instruction nodes, and comment nodes.

    Sample XPath Queries Against Sample XML Fragment

    1. /gatech_student/name
      Selects all name elements that are children of the root element gatech_student.

    2. //age
      Selects all age elements in the document.

    3. /gatech_student/*
      Selects all child elements of the root element gatech_student.

    4. /gatech_student[@gtnum]
      Selects all gtnum attributes of the gatech_student elements in the document.

    5. //*[name()='age']
      Selects all elements that are named "age".

    6. /gatech_student/age/ancestor::*
      Selects all ancestors of all the age elements that are children of the gatech_student element (which should select the gatech_student element).

  2. XML Query Language (XQuery): XQuery is an attempt to provide a query language that provides the same breadth of functionality and underlying formalism as SQL does for relational databases. XQuery is a functional language where each query is an expression. XQuery expressions fall into seven broad types; path expressions, element constructors, FLWR expressions, expressions involving operators and functions, conditional expressions, quantified expressions or expressions that test or modify datatypes. The syntax and semantics of the different kinds of XQuery expressions vary significantly which is a testament to the numerous influences in the design of XQuery.

    XQuery has a sophisticated type system based on XML schema datatypes and supports the manipulation of the document nodes unlike XPath. Also the data model of XQuery is not only designed to operate on a single XML document but also a well-formed fragment of a document, a sequence of documents, or a sequence of document fragments.

    W3C is also working towards creating an alternate version of XQuery that has the same semantics but uses XML based syntax instead called XQueryX.

    Sample XQuery Queries and Expressions Taken From W3C Working Draft

    1. path expressions: XQuery supports path expressions that are a superset of those currently being proposed for the next version of XPath.

      1. //emp[name="Fred"]/salary * 12

        From a document that contains employees and their monthly salaries, extract the annual salary of the employee named "Fred".

      2. document("zoo.xml")//chapter[2 TO 5]//figure

        Find all the figures in chapters 2 through 5 of the document named "zoo.xml."

    2. element constructors: In some situations, it is necessary for a query to create or generate elements. Such elements can be embeded directly into a query in an expression called an element constructor.

      1.       <emp empid = {$id}>
        Generate an <emp> element that has an "empid" attribute. The value of the attribute and the content of the element are specified by variables that are bound in other parts of the query.

    3. FLWR expressions: A FLWR (pronounced "flower") expression is a query construct composed of FOR, LET, WHERE, and a RETURN clauses. A FOR clause is an iteration construct that binds a variable to a sequence of values returned by a query (typically a path expression). A LET clause similarly binds variables to values but instead of a series of bindings only one occurs similar to an assignment statement in a programming language. A WHERE clause contains one or more predicates that are used on the nodes returned by preceding LET or FOR clauses. The RETURN clause generates the output of the FLWR expression, which may be any sequence of nodes or primitive values. The RETURN clause is executed once for each node returned by the FOR and LET clauses that passes the WHERE clause. The results of these multiple executions is concatenated and returned as the result of the expression.

      1.        FOR $b IN document("bib.xml")//book
                WHERE $b/publisher = "Morgan Kaufmann"
                AND $b/year = "1998"
                RETURN $b/title
        List the titles of books published by Morgan Kaufmann in 1998.

      2.        <big_publishers>
                     FOR $p IN distinct(document("bib.xml")//publisher)
                     LET $b := document("bib.xml")//book[publisher = $p]
                     WHERE count($b) > 100
                     RETURN $p
        List the publishers who have published more than 100 books.

    4. conditional expressions: A conditional expression evaluates a test expression and then returns one of two result expressions. If the value of the test expression is true, the value of the first result expression is returned otherwise, the value of the second result expression is returned.

      1.        FOR $h IN //holding
                     IF ($h/@type = "Journal")
                     THEN $h/editor
                     ELSE $h/author
                 SORTBY (title)
        Make a list of holdings, ordered by title. For journals, include the editor, and for all other holdings, include the author.

    5. quantified expressions: XQuery has constructs that are equivalent to quantifiers used in mathematics and logic. The SOME clause is an existential quantifier used for testing to see if a series of values contains at least one node that satisfies a predicate. The EVERY clause is a universal quantifier used to test to see if all nodes in a series of values satisfy a predicate.

      1.        FOR $b IN //book
                WHERE SOME $p IN $b//para SATISFIES
                   (contains($p, "sailing") AND contains($p, "windsurfing"))
                RETURN $b/title
        Find titles of books in which both sailing and windsurfing are mentioned in the same paragraph.

      2.        FOR $b IN //book
                WHERE EVERY $p IN $b//para SATISFIES
                   contains($p, "sailing") 
                RETURN $b/title
        Find titles of books in where sailing is mentioned in every paragraph.

    6. expressions involving user defined functions: Besides providing a core library of functions similar to those in XPath, XQuery also allows user defined functions to be used to extend the core function library.
      1.       NAMESPACE xsd = ""
               DEFINE FUNCTION depth($e) RETURNS xsd:integer 
                 # An empty element has depth 1
                 # Otherwise, add 1 to max depth of children
                IF (empty($e/*)) THEN 1
                ELSE max(depth($e/*)) + 1
        Find the maximum depth of the document named "partlist.xml."

    XML and Databases

    As was mentioned in the introduction, there is a dichotomy in how XML is used in industry. On one hand there is the document-centric model of XML where XML is typically used as a means to creating semi-structured documents with irregular content that are meant for human consumption. An example of document-centric usage of XML is XHTML which is the XML based successor to HTML.

          <html xmlns ="">
        <title>Sample Web Page</title>
            <h1>My Sample Web Page</h1>
            <p> All XHTML documents must be well-formed and valid. </p>
            <img src="" height ="50" width = "25"/>
            <br />
            <br />        

    The other primary usage of XML is in a data-centric model. In a data-centric model, XML is used as a storage or interchange format for data that is structured, appears in a regular order and is most likely to be machine processed instead of read by a human. In a data-centric model, the fact that the data is stored or transferred as XML is typically incidental since it could be stored or transferred in a number of other formats which may or may not be better suited for the task depending on the data and how it is used. An example of a data-centric usage of XML is SOAP. SOAP is an XML based protocol used for exchanging information in a decentralized, distributed environment. A SOAP message consists of three parts: an envelope that defines a framework for describing what is in a message and how to process it, a set of encoding rules for expressing instances of application-defined datatypes, and a convention for representing remote procedure calls and responses.

           <SOAP-ENV:Envelope  xmlns:SOAP-ENV=""
             <m:GetLastTradePrice xmlns:m="Some-URI">

    In both models where XML is used, it is sometimes necessary to store the XML in some sort of repository or database that allows for more sophisticated storage and retrieval of the data especially if the XML is to be accessed by multiple users. Below is a description of storage options based on what model of XML usage is required.

    1. Data-centric model: In a data-centric model where data is stored in a relational database or similar repository; one may want to extract data from a database as XML, store XML into a database or both. For situations where one only needs to extract XML from the database one may use a middleware application or component that retrieves data from the database and returns it as XML. Middleware components that transform relational data to XML and back vary widely in the functionality they provide and how they provide it. For instance, Microsoft's ADO.NET provides XML integration to such a degree that results from queries on XML documents or SQL databases can be accessed identically via the same API. Some like Merant's jxTransformer require the user to specify how the results of a SQL query should be converted to XML via a custom query while others like IBM's Database DOM require the user to create a template file that contains the SQL to XML mappings for the query to be performed. Another approach is the one taken by DB2XML where a default mapping of SQL results to XML data exists that cannot be altered by the user. Middleware components also vary in how the sophistication of their user interface which may vary from practically non-existent (interaction done via programmatically using APIs) to interaction being via a sophisticated graphical user interfaces.

      The alternative to using middleware components to retrieve or store XML in a database is to use an XML-enabled database that understands how to convert relational data to XML and back. Currently, the Big 3 relational database products all support retrieving and storing XML in one form or another. IBM's DB2 uses the DB2 XML Extender. The DB2 extender gives one the option to store an entire XML document and its DTD as a user-defined column [of type XMLCLOB,XMLVARCHAR or XMLFile] or to shred the document into multiple tables and columns. XML documents can then be queried with syntax that is compliant with W3C XPath recommendation. Updating of XML data is also possible using stored procedures.
           TABLE mail_user
             user_name    VARCHAR(20) NOT NULL PRIMARY KEY
             passwd       VARCHAR(10)
             mailbox      XMLVARCHAR
             SELECT user_name FROM  mail_user WHERE extractVarchar(mailbox,"/Mailbox/Inbox/Email/Subject") LIKE "%XML%" 
         The above query returns the names of all the users that have any email in their inbox that 
         contains the string "XML" in its subject. To improve the performance of the XPath query it is 
         necessary to index the mailbox XMLVARCHAR.

      Oracle has completely integrated XML into it's Oracle 9i database as well as the rest of its family of products. XML documents can be stored as whole documents in user-defined columns [of type XMLType or CLOB/BLOB] where they can be extracted using XMLType functions such as Extract() or they can be stored as decomposed XML documents that are stored in object relational form which can be recontituted using the XML SQL Utility (XSU) or SQL functions and packages. For searching XML, Oracle provides Oracle Text which can be used to index and search XML stored in VARCHAR2 or BLOB variables within a table via the CONTAINS and WITHIN operators used in collusion with SQL SELECT queries. XMLType columns can be queried by selecting them through a programming interface (e.g. SQL, PL/SQL, C, or Java), by querying them directly and using extract() and/or existsNode() or by using Oracle Text operators to query the XML content. The extract() and existsNode() functions uses XPath expressions for querying XML data. Oracle 9i also allows one to create relational views on XML documents stored in XMLType columns which can then be queried using SQL. The columns in the table are mapped to XPath expressions that query the document in the XMLType column.

           CREATE TABLE mail_user(
           user_name VARCHAR2(20),
           passwd    VARCHAR2(10), 
           mailbox   SYS.XMLTYPE );
           SELECT user_name FROM mail_user m WHERE m.mailbox.extract('/Mailbox/Inbox/Email/Subject/text()').getStringVal() like '%XML%'
         The above query returns the names of all the users that have any email in their inbox that 
         contains the string "XML" in its subject.  To improve the performance of the XPath query it is 
         necessary to index the mailbox XMLType.
      Microsoft's SQL Server 2000 also supports XML operations being performed on relational data . XML data can be retrieved from relational tables using the FOR XML clause. The FOR XML clause has three modes: RAW, AUTO and EXPLICIT. RAW mode sends each row of data in the resultset back as a XML element named "row" and with each column being an attribute of the "row" element. AUTO mode returns query results in a nested XML tree where each element returned is named after the table it was extracted from and each column is an attribute of the returned elements. The hierarchy is determined based on the order of the tables identified by the columns of the SELECT statement. With EXPLICIT mode the hierarchy of the XML returned is completely controlled by the query which can be rather complex. SQL Server also provides the OPENXML clause which to provide a relational view on XML data. OPENXML allows XML documents placed in memory to be used as parameters to SQL statements or stored procedures. Thus OPENXML is used to query data from XML, join XML data with existing relational tables, and insert XML data into the database by "shredding" it into tables. Also W3C XML schema to can be used to provide mappings between XML and relational structures. These mappings are called XML views and allow relational data in tables to be viewed as XML which can be queried using XPath.

      As can be seen from the above descriptions, there is currently no standard way to access XML from relational databases. This may change with the development of the SQL/XML standard currently being developed by the SQLX group.

    2. Document-centric model: Content management systems are typically the tool of choice when considering storing, updating and retrieving various XML documents in a shared repository. A content management system typically consists of a repository that stores a variety of XML documents, an editor and an engine that provides one or more of the following features:

      • version, revison and access control
      • ability to reuse documents in different formats
      • collaboration
      • web publishing facilities
      • support for a variety of text editors (e.g. Microsoft Word, Adobe Framemaker, etc)
      • indexing and search capabilities

      Content management systems have been primarily of benefit for workflow management in corporate environments where information sharing is vital and as a way to manage the creation of web content in a modular fashion allowing web developers and content creators to perform their tasks with less interdependence than exists in a traditional web authoring environment. Examples of XML based content management systems are SyCOMAX, Content@, Frontier, Entrepid, XDisect, and SiberSafe.

    3. Hybrid model: In situations where both documentric-centric and data-centric models of XML usage will occur, the best data storage choice is usually a native XML database. What actually constitutes a native XML database has been a topic of some debate in various fora which has been compounded by the blurred lines that many see between XML-enabled databases, XML query engines, XML servers and native XML databases. The most coherrent definition so far is one that was reached by consensus amongst members of the XML:DB mailing list which defines a native XML database as a database that has an XML document as its fundamental unit of (logical) storage and defines a (logical) model for an XML document, as opposed to the data in that document, and stores and retrieves documents according to that model. At a minimum, the model must include elements, attributes, PCDATA, and document order. Described below are two examples of native XML databases with the intent of showing the breadth of functionality and variety that can be expected in the native XML database arena.

      Tamino is a native XML database management system developed by Software AG. Tamino is a relatively mature application, currently at version 2.3.1, that provides the means to store & retrieve XML documents, store & retrieve relational data, as well as interface with external applications and data sources. Tamino has a web based administration interface similar to that used by the major relational database management systems and includes GUI tools for interacting with the database and editting schemas.

      Schemas in Tamino are DTD-based and are used primarily as a way to describe how the XML data should be indexed. When storing XML documents in Tamino; one can specify a pre-existing DTD which is then converted to a Tamino schema, store a well-formed XML document without a schema which means that default indexing ensues or a schema can be created from scratch for the XML document being stored. A secondary usage of schemas is for specifying the datatypes in XML documents. The main advantage of using datatypes in Tamino is to enable type based operations within queries (e.g. numeric comparisons). The query language used by Tamino is based on XPath and is called X-Query (not to be confused with the W3C XQuery).

      Tamino also ships with a relational database management system which is called the SQL Engine. Schemas can be used to creating mappings from SQL to XML which then allow for the storage or retrieval of XML data from relational database sources either internal (i.e. the SQL Engine) or external. Schemas can also be used to represent joins across different document types. Joins allow for queries to be performed on XML documents with differing schemas. Future versions of Tamino are supposed to eliminate the need to specify joins up front in a schema and instead should allow for such joins to be done dynamically from a query.

      Tamino provides APIs for accessing the XML store in both Java and Microsoft's JScript. C programmers can interact with the SQL engine using the SQL precompiler that ships with Tamino. Interfaces that allow ODBC, OLE DB and JDBC clients to communicate with the Tamino SQL Engine are also available. Finally, Tamino ships with the X-Tensions framework which allows developers to extend the functionality of Tamino by using C++ COM objects or Java objects. Tamino operations have ACID properties (Atomicity, Consistency, Isolation and Durability) via the support of transactions in its programming interfaces.

      dbXML is an Open Source native XML database management system which is sponsored by the dbXML Group. dbXML is designed for managing collections of XML documents which are arranged hierarchically within the system in a manner similar to that of a file system. Querying the XML documents within the system is done using XPath and the documents can be indexed to improve query performance.

      dbXML is written in Java but supports access from other languages by exposing a CORBA API thus allowing interaction with any language that supports a CORBA binding. It also ships with a Java implementation of the XML:DB XML Database API which is designed to be a vendor neutral API for XML databases. A number of command line tools for managing documents and collections are also provided.

      dbXML is mostly still in development (version at time of writing was 1.0 beta 2) and does not currently support transactions or the use of schemas but these features are currently being developed for future versions.


    1. Chamberlin, Don et al. XQuery 1.0: An XML Query Language (Working Draft). 7 June 2001. < >

    2. Clark, James and Steve DeRose. XPath: XML Path Language (Version 1.0). 16 November 1999. < >

    3. Bourett, Ronald. XML and Databases. June 2001. <>

    4. Bourett, Ronald. XML Database Products. 22 October 2001. <>

    5. Turau, Volker. Making Legacy Data Accessible For XML Applications. 1999. <>

    6. Cheng, Josephine and Xu, Jane. IBM DB2 Extender. From ICDE "00 Conference, San Diego. February 2000. < >


    The following people helped in reviewing and proofreading this paper: Dr. Sham Navathe, Kimbro Staken, Dmitri Alperovitch, Sam Collins, Omri Gazitt and Dennis Lu.

    © 2001 Dare Obasanjo