The Web

What a building!

Its 17 years since the World Wide Web first appeared in the wild. And it changed a lot since then. It still is changing in its technology, the services it provides and the way they are used. Meanwhile it holds so many options that a unique view on what it offers is very difficult. So the editors decided to pick some developments that they consider mainstream to give an impression of what is about to come.

• Web everywhere – the Mobile Web
• Web for everyone – the Accessible Web
• Web of structured information – the Semantic Web
• Web of participation and communication – Collections, Blogs and Fora
• Web as a service – Rich Web Applications
• Web for business – Web Services
• Web for media – Video and Voice on the web

All these should rely on a common set of technologies and standards. Functionalities of the different aspects should interoperate with each other.

Some other topics are relevant for most of these aspects. They are horizontal by nature and some are subject to different cultural views.

• privacy on the Web
• security and trust on the Web
• treatment of intellectual property
• content classification and rating
• accessibility of the Web

We will not cover here different views on subjects like privacy, intellectual property and content classification. These are still subject of political and social discussions. Web technologies should approach what ever is seen as beneficial in a given social and political context.

Most of the work on standards and technologies described in this paper is or was conducted at the W3C which was launched 1994 by Web inventor Tim Berners-Lee.

The basement

Three technologies used in combination made up the success of the Web:

• a common syntax for documents that may contain hyperlinks (HTML)
• a syntax for locating documents (or fragments of documents) and the option to use this syntax within HTML to build hyperlinks (URL)
• a protocol to send request and responses between clients (usually browsers) and servers (HTTP)

These technologies allowed to span a global web of documents that – from a user perspective – provided a graphical user interface and a very easy way to jump from one document to a related one using a link provided in the first one.

This Web of documents quickly became a victim of its success. There was a huge amount of documents and the desire to integrate information from databases. Shop offers, flight schedules… etc. required extensions of the concept. In the short term it was scripting (server side and client side) and CGI that allowed for dynamic Web pages. But there was also a need for more structured and systematic solutions. So a generalization and extension of the 3 original technologies had to be made. In terms of structuring documents and data, two concepts where introduced: The XML Infoset and RDF Graphs.

The XML Infoset is a generalization of the HTML concept (while being stricter on the syntax. As HTML, XML structures a document or other entities in a tree with named (tagged) nodes. HTML uses a predefined set of node names and rules to restrict their use within a document. The XML Infoset holds no such restrictions. Introducing specific restrictions can be used to form new data structures or mark-up languages. In fact thousands of those XML languages exist today for a huge variety of purposes. A few supporting technologies came with XML as there are:

• XML Namespaces as a means to separate one space of tags from another
• XML Schemas to allow the definition of restrictions to specify a new XML language
• Xquery/Xpath as a means to navigate through or query XML documents
• XML base to cope with relative urls
• Xpointer to allow reference to XML documents and fragments
• DOM an abstract interface to process XML

XML relies on the concept to define complex entities by composing them of simpler ones. RDF Graphs try to define entities by describing their relationship to others. An RDF graph is composed of triples of the form subject predicate object. Where subject (a resource) denotes something you want to define, predicate defines the relationship (property), and object (a resource) is something for which the relationship holds. Objects (and even predicates) in one triple can appear as subjects in others. So we may see huge networks of relations. Properties that have their origin in mathematics or logic (like “is a subset of”, “implies”…) lay the ground for complex queries and reasoning on such networks. As XML RDF comes with a set of associated technologies:

• RDFS is a Schema language that defines a vocabulary that gives meaning to some resources (like Class, Literal, Datatype…) and properties (like subClassOf, subPropertyOf, domain, range…) This opens a standardized way to express set or logic related properties. OWL supplements and extents these definitions.
• SPARQL is a query language for RDF

Though XML and RDF follow different concepts – hierarchical and relational – in structuring information, they are not completely separated from each other. RDF/XML is a special XML language to represent (serialize) RDF Graphs – admittedly not the most human-readable one. Basic data types used in RDF are from XML Schema.

Beside the technologies introduced here there is a collection of so called Web Architecture Principles that constitute rules on how to use these technologies to build new ones.

Cornerstones

We want the Web to be universal. This is a noble claim with quite a few implications. In the end it means that everybody should be able to use the Web with a wide range of devices in any language he knows. It is quite obvious that this has its limitations (e.g. the author of these lines will probably never be able to understand an article on microbiologic in Chinese). This is no surprise. The surprising thing is how much universality can be obtained by following a few guidelines. The areas for which these guidelines are developed are: accessibility, device independence, internationalization.

W3C is the developer of some of these guidelines. In the end there are a few general rules that are refined and complemented:

• use standards wherever they exist
• add redundancy in the way you present your information
• structure your information properly, present it recognizable
• separate content from presentation

There is an overlap in the technical requirements to achieve this. A page that separates content from presentation can be easier adapted to various devices but also to accessibility needs. A picture of a building in document that is accompanied by a text might help the blind to understand it but it also may help somebody who e.g. never saw the building before to identify it or a search engine to find it. Texts reach more people by adding redundancy by providing them multilingual. Using standardized character sets helps, among other things, machine translation…

But the implications go even beyond the subjects mentioned here. E.g. a Web site that is accessible is usually better searchable. (Or as Karsten M.Self stated in Arachnophobia: “Google is, for all intents, a blind user. A billionaire blind user with tens of millions of friends, all of whom hang on his every word. I suspect Google will have a stronger impact than the ADA in building accessible Web sites.“) In general: going for accessibility, device independence, and internationalization – as a very welcome side effect – also significantly contributes to usability and the value of a site.

Supporting Pillars

Having all these nice technologies and the knowledge what could we expect from the Web? Every technology is there to make our life easier. It shall reduce the effort we have to spend, to reach our goals. As information technology in general the Web offers a broad spectrum of application areas. Unlike information technology the Web offers these applications networked on a global scale. Broadband connections and nearly unlimited storage and compute power together with interoperable technologies open the field for multimedia and huge databases. Business models raise and fall, killer applications pop-up. Social networking, privacy, sharing knowledge and intelligence, managing intellectual property, participation in decision processes, marketing of ideas and products, the importance of the long tail,… have to be rethought or became only relevant based on the Web.

We see a convergence of Web technology with mobile phone technology. Digital mobile phones appeared in the public during the first half of the 90s – at the same time as the Web. Since then the growth of this market even outperforms the growth of the Internet. Mobile phones play a crucial role for the economy in many developing countries. Today every mobile phone can access the Web. So both technologies converge.

On the other hand there are some things that are still a mess. Searching can easily become a nightmare. Phishing and the abuse of browser vulnerabilities threaten users. Automatic translation is rather not ready yet to access information in languages the user does not speak. So there is still a lot to be fixed and researched, a lot of technological, social and political decisions are to be made.

A classification of the technologies that are available or under development might be by the following topics:

• Web applications
• Semantic Web
• Web Services
• Mobile Web
• Voice Communication
• Privacy and security

Web applications

Plain HTML is all you need to build a Web application. At the other end of the spectrum are fully fledged applications e.g. for the office work or picture processing. Video, audio, sophisticated graphics, presentational software for mathematics or chemistry can be included. Plugins and Interpreters for programming languages allow nearly any content or processing that works on a desktop to be used within a Web page. Quite a few technologies are available to support this. Some are standardized, some are proprietary. Some are “open" (e.g. by providing Open Source) some are “locked".

Some very important technologies and standards in this context are provided by W3C – free of charge and open for everyone. Among these standards, HTML is the one most widely used. It is also the standard that provides means to embed objects that follow other standards to do graphics, video, images… For good or ill there are two HTMLs that are under standardization by W3C today named HTML 5 and XHTML 2. This might not be the place to go into details. On the risk of oversimplifying one might characterize these two directions as follows:

• the main focus of XHTML 2 is to better use the advantages of XML within HTML. At the same time one wants to get rid of constructs (elements) whose functionality might better be implemented in cleaner and more general way. Allowing integration of other technologies or requirements (compound documents, semantic Web, accessibility) is also a design goal.
• HTML 5 is more on advancing the existing HTML (HTML 4 or XHTML 1). It is not necessarily relying on XML, though it supports its use, but upward compatibility from earlier versions of HTML is very important.

Of course there is quite some overlap in goals and principles of both HTMLs but the differences in their design goals are significant.

In order to allow to separate the presentation from the content of HTML or XML documents, a technology named CSS was developed. It provides flexible means to define for every element how it is to be presented within the document. CSS is a very powerful technology that allows Web designers to adopt their pages to a wide range of devices or window sizes. A more general approach is XSL. XSL allows not only to format existing documents it also allows to transform their content. So the elements of a document can be rearranged, compacted or extended.

In order to allow the reader of a Web page to reply (e.g. to place an order in a job or to register for an event) HTML documents may contain forms. Forms are available as part of the HTML standard. Much more powerful - and properly integrated in the XML world – is Xforms that can be used with any XML language.

A very powerful XML language for graphics is SVG. It allows vector based two dimensional graphics which – other than pixel based images – can be scaled up and down without losing quality. SVG allows also vector based animations. Many modern browsers support SVG or a subset of it.

As was said before, (X)HTML allows to include non (X)HTML objects. These are either referred by a link or directly included – often as XML documents with another namespace (e.g. SVG or the mathematical mark-up language MathML). This should normally work fine unless there is supposed to be some user interaction or synchronization of animations. There is series of compound document standards on its way (WICD, CDF…) that will allow better integration of parts of a document using different technologies.

The Web – especially when combined with powerful programming languages and interfaces – can act as a platform for sophisticated applications. So we see a lot of applets, widgets and applications that rely on Web technologies. Games, maps, image processing or complete office suites… can be based on Browsers using technologies like Compound Document Formats, XML, the DOM (and DOM events), Xforms, XHTML, SVG and some more.

In order to build presentations with media from various sources in a synchronized manner SMIL might be used. SMIL was published in 1998 and is an XML language that allows to arrange media in time and place.

Semantic Web

Finding things that are on the Web is still one of the big challenges. There are a lot of search engines around – text based, based on subject catalogs and specialized engines like lowest price finder etc. Nevertheless finding e.g. all music events in Berlin that take place in April next year might be a challenge if you don't find a specialized database. The reason for this is that Web documents usually do not tell “this is about an Event that takes place in April” in a way that is understood by search engines. A human reader can usually easily conclude this when she reads the page. Not so a search engine. It finds the word “April” but has no idea that this might be the month in which an event takes place it does not even know the page is about an event nor what an “event” might be and the same applies to the word “music”.

In order to solve this we have to “teach” computers. This is done by defining relationships between the things we are talking about. We have (to find ways) to tell computers that music is a performing art. That a performing art is an art. That everything of type music has a duration. That music can be a live performance. That a life performance is an event. That every event has a start time etc. Second we need a way to let computers know that a certain piece of data is e. g. a start time. And we have to be able to specify the event this start time belongs to.

We mentioned above that RDF is a language that allows us to talk about relations. But in order to be usable for computers we also need a unique way to name properties and resources. Here the concept of the URL comes in handy. The URL constitutes a way to uniquely identify Web documents. But there is no technical reason to constrain the mechanism of URLs to Web pages. By removing this restriction we obtain the URI which is used by RDF to uniquely identify resources and properties.

Properties are resources, too. So we can make statements about properties. E.g. we could define that a property “isSubClassOf” allows us to conclude that if “car isSubClassOf vehicles” and “myCar isMemberOf car” than “myCar isMemberOf vehicles”. This allows to conclude new statements from existing ones by applying logic. There are different RDF vocabularies (and rules) defining the meaning of such properties. Apart from the simplest, RDFS, there are (today) three levels standardized under the name OWL: OWL Lite, OWL DL and OWL Full. Each level offers different options for specifying and concluding relations. There are algorithms that help to find such conclusions (and irritatingly enough: in general it is not sure that algorithms processing some OWL Full will ever finish).

What do these technologies mean for the Web? We could imagine an “overlay” Web where e.g. every start date of an event identifies itself as start date, refers to the event it belongs to, which identifies itself as being a performance, and identifies that it is music… There is a standard named RDFa which allows to include RDF statements in an XHTML document.

Beside searching, another use of linked data has emerged in the Web: the mash-up. Though today often based on proprietary APIs the idea behind it is to use data from one source and merge it with some information from another source. Using Semantic Web technologies to link the data would ease the construction and maintenance of mash-ups a lot.

Another example of using linked Data is the FOAF project. It provides an RDF vocabulary to describe a person. Part of the description can be data about who this person knows – which can link to or contain another FOAF description. Some open social Networks are based on FOAF.

Beside RDF, another technology is used to link data on the Web: Microformats. Microformats use the “class” attribute of (X)HTML to denote that an HTML element contains some data. In order to be interoperable, Microformats have to be administered centrally. Microformat specifications are available from microformats.org.

Web Services

Web Services is mostly not about the browsing part. It is more about the interoperation of software applications that run on computers interconnected by a network. A set of standards help to achieve this interoperability and to couple applications in a loose way to achieve complex operations. Especially in e-business, Web Services play a central role. The opportunity to pass data from one business step to the next ones using an XML data format appears very attractive and has the potential to automate parts of the business process. So we see a lot of concepts and buzzwords (EAI, ESB, MOM, SOA…) in this area that in the end are based on Web Services. Several requirements have to be met in order to get Web Services this working:

• you need a protocol to exchange messages between the applications
• you need to be able to address services and messages
• you need to describe services in a machine readable way
• you may want to apply policies to control the use of services

There are a few standards that lay the ground for Web Services based on these requirements:

• SOAP specifies the format of messages that can be used to communicate with a service
• WS-Addressing describes how to address Web Services and messages
• WSDL is a language to describe a service by the kind of messages it will receive or send and its functionality. There is also a standard for Semantic Annotations to WSDL (SAWSDL) which allows to use ontologies to deceive the semantics of Web Services.
• a means to specify policies for the use of a service can be specified and attached to a service by using the Web Services Policy specifications.

Mobile Web

Mobile use of the Web is characterized by devices with limited capabilities (e.g. mobile phones) in terms of connectivity, bandwidth and processing power of the client. So Mobile Web standards are about making browsing possible and a good experience while using mobile devices. Mobile Web standards focus on two aspects:

• down sizing of existing standards to match the constraints of mobile devices
• the development of guidelines especially for authors (but for authoring-tool manufacturers as well) on how to make sure that a Web site is mobile friendly
• a way for servers to learn about the capabilities of an attached client

XHTML Basic, SVG Tiny, SVG Basic, SMIL mobile, Xforms Basic and CSS mobile are standards that include those modules of the respective technologies that mobile devices should provide. In order to tailor the content delivery of a server to the capabilities of a mobile device (or the browser used on the device) the server should be able to learn about the capabilities of the device. The DDR simple API describes an interface that servers can use to query device description repositories that hold this information.

A second aspect of mobility is that location awareness might be an advantage for the user. Many mobile devices have some build-in capability to locate their position on the globe. In order to use this information for services to the user, work on a Geolocation API specification has been started.

Voice Communication

One might wonder about the minimum requirements to access the Web. Would a plain old telephone work? Sometimes? Quite some information is disseminated or communicated by phone. So it is quite reasonable to ask how this could be supported by Web technologies. Everyone who used a computer-based phone service before knows that – friendly spoken – there is space for improving the user experience. In order to pave the road for better (semi-) automated voice communications a set of standards was developed that covers the following areas:

• VoiceXML is about defining frameworks for dialogs. It is a mark-up language to collect information from a user. The flow of the dialog is controlled by the VoiceXML mark-up depending on the answers of the user. VoiceXML is an XML language with scripting capabilities.
• Two speach oriented standards support the use of VoiceXML. The SSML is an XML based language to assist the generation of synthetic speech. SRGS allows the specification of grammars for recognition of speech.

CCXML provides support for telephony call control and can be used in conjunction with dialog systems such as VoiceXML. EMMA is an XML language to describe input received on several channels (voice, handwriting, keyboard…) simultaneously.

Privacy and security

The Web has changed our life in many ways. One of the most ambivalent changes is in our privacy. While using the Web we disclose a lot of our personal data, our interests and preferences. Every question we ask, every Web site we visit can be registered. It is hard for the user to guess how this information will be used. In order to protect their citizens governments have released privacy laws. But the scope of legislation usually ends at the national border which limits its effect in the Internet. This problem cannot be resolved by technical means only. People have to handle their private data with care and socially acceptable and technically implementable new models for privacy have to be found.

P3P is a standard for specifying a machine readable description of a privacy policy. It e.g. allows companies to publish their privacy policies in a way that a user agent (e.g. as part of a Web browser) can understand. Since there is no way to enforce the policy, P3P is useful in cases where the user can trust the site that issues a policy. This is usually the assumption when doing business. Some browsers (e.g. Internet Explorer 6 and later) use P3P information provided by servers to allow users to control the use of cookies.

There are institutions we trust in. But can we be sure that the Web site that pretends to be the home page of our bank is really our bank? Can we securely transfer confidential data over the Web? Can your bank make sure that a money transfer from your account was really issued by you? That it was not modified on the way?

These problems are not specific for the Web. Mathematics – especially the number theory – has provided us with the instrument of encryption that solves many problems in this area. The internet protocols provide with TLS/SSL means to apply these instruments to the transfer of data. There are some things that obviously still have to be resolved to apply this technology on the document or data level. XML Encryption, XML Signature and XML Key Management are 3 related technologies that allow to apply well known cryptographic methods on fragments of XML documents e.g. to digitally sign a part of the body of a SOAP message.

There is ongoing work on how to evaluate or inform users about the security context of an environment they are using. In the end this work is about making better use of certificates, encryption, Web site dynamics and other information provided by the server. This should enable users to better understand the security context of a Web site when making trust decisions.

Though there is a lot of progress in this area, there are still quite some unresolved issues with respect to privacy and security.

The roof

Web Sites and Killer Applications

All these technologies and standards are rather a means to an end than ends in themselves and the list is far from complete. Together with steadily increasing bandwidth and compute power they make the Web a place where people use their creativity for doing business, conduct discussions, disseminate ideas or build relationships. During the last 17 years the Web transformed the internet from a geek's toy to a powerful infrastructure that is formed by millions of individuals, business people, journalists, artists, scientists… Google, Yahoo, Flickr, Ebay, Youtube, Myspace, Wikipedia, Mozilla are only a few popular projects that live in this space.

But there are still a lot of open questions that wait for ideas, technical solutions, standards or a social agreement to be answered. In parallel to the work on millions of Web sites, Blogs, Wikis, shops and fancy applications there is ongoing work on better standards and technologies. The W3C provides a framework where vendors, authors and users cooperate to improve the supporting frame of the Web.

Abbreviations

ADA

Americans with Disabilities Act

API

Application Programming Interface

CCXML

(Voice Browser) Call Control XML

CDF

Compound Document Formats

CGI

Common Gateway Interface

CSS

Cascading Style Sheets

DDR

Device Description Repositories

DOM

Document Object Model

EAI

Enterprise Application Integration

EMMA

Extensible MultiModal Annotation markup language

ESB

Enterprise Service Bus

FOAF

Friend of a Friend

HTML

Hypertext Markup Language

HTTP

Hypertext Transfer Protocol

MathML

Mathematical Markup Language

MOM

Message Oriented Middleware

OWL

Web Ontology Language

P3P

Platform for Privacy Preferences

RDF

Resource Description Framework

RDF/XML

RDF in XML Syntax

RDFS

RDF Schema

RDFa

RDF in attributes

SAWSDL

Semantic Annotations for WSDL

SMIL

Synchronized Multimedia Integration Language

SOA

Service-Oriented Architecture

SOAP

(originally for) Simple Object Access Protocol (long version is misleading and not used anymore)

SPARQL

Protocol and RDF Query Language

SRGS

Speech Recognition Grammar Specification

SSML

Speech Synthesis Markup Language

SVG

Scalable Vector Graphics

TLS/SSL

Transport Layer Security / Secure Sockets Layer

URI

Uniform Resource Identifier

URL

Uniform Resource Locator

VoiceXML

Voice Extensible Markup Language

W3C

World Wide Web Consortium

WICD

Web Integration Compound Document

WSDL

Web Service Description Language

Xforms

XML application that represents the next generation of forms

XHTML

Extensible Hypertext Markup Language

XML

Extensible Markup Language

References

The W3C Technology Stack
<http://www.w3.org/2004/10/RecsFigure.png>

About W3C Technology
<http://www.w3.org/Consortium/technology>

W3C Technical Reports and Publications esp. all Standards
<http://www.w3.org/TR/>

About W3C Activities
<http://www.w3.org/Consortium/activities>