To some, portals represent communities; to others, they are trading hubs or e-marketplaces; and to many, they are integrated desktop environments. Putting aside the hype, a common theme of substance underlies portals -- a theme of greater levels of integration. From a unifying technology perspective, a portal is a single integrated point of comprehensive, ubiquitous, and useful access to information (data), applications, and people. This definition encompasses all the different views of the purpose and functionality of portals. But more importantly, strong pursuit of satisfying this portal definition will help evolve the next generation of integrated services and business processes.
The Application Framework for e-business provides a full range of services
for developing, deploying, and managing e-business applications. Portals
are driving many different requirements that extend and enhance almost
every layer of the Framework as illustrated by the check marks in the figure
below. The key focus of portals is integrated access to both data and applications
and greater levels of integration between the two.
Portal applications, or portals, support online communities of users, employees, customers, partners, and businesses in a variety of combinations. Enabling Web self service is a key objective of many portals. The portal infrastructure helps create, integrate, manage, and personalize a comprehensive set of information and applications for specific portals that cater to the needs and interests of a particular community. The real value proposition for portals is:
- Improved productivity for individuals and businesses resulting from secure, integrated access to relevant information and applications, as well as ubiquitous access to all individuals, businesses, and institutions in the world from anywhere, any place, and any time
- Improved business processes resulting from better information flow between people and business applications, and from the collaborative environments that help reduce time to transform raw information to knowledge and expertise that feed such business processes
- Reduction in deployment and management overhead for information and application services in an enterprise
There are many different types of portals, but at a high level all portals can viewed as an instance or a combination of one of the following types:
portals or corporate desktops provide employees with access
to information and applications. One can envision that such portals will
evolve to become the next generation of integrated and intelligent desktops
that can be accessed from anywhere and from any device. Many information-centric
applications such as knowledge management, business intelligence, and customer-relationship
management software are being integrated and deployed via similar portals.
portals are owned and managed by a single enterprise but support business
processes such as supply-chain solutions and procurement among customers,
partners, and vendors across different enterprises. Supply-chain portals
aggregate parts, inventory, and pricing information from a large number
of suppliers, support interaction, collaboration, and dynamic partnerships
and help manage/coordinate the end-to-end flow of a business process.
are portals or trading hubs that connect buyers and sellers in virtual
marketplaces. Such portals are cross-enterprise and can cater to specific
industries such as the steel industry or semiconductor industry. In addition
to an infrastructure that is common to most portals, e-marketplaces may
need specific service -- for example, bidding and auction services.
portals, provided by the likes of Yahoo, Excite, and Netscape, provide
users with a customized, first point of access to the Web. The business
intent is to grow the portal's user base by providing useful information
and functions that in turn drive advertisement revenues. Community portals
are a variant that cater to specific communities of interest by providing
portals provide integrated access to data and applications and are
supplemented by a set of additional services to support the business models
of Application Service Providers, or ASPs. The new business model of ASPs
expects renting of hosted applications to be a viable and economic alternative
for users. Issues and considerations pertinent to ASPs are discussed in
the Framework's Application
hosting services white paper.
Portals are being considered for the next generation of integrated application development platforms. Instead of disconnected pieces of applications that must be sequenced manually, application vendors are exploiting the power of the integrated, process-centric view of the design and development process. Common services for almost all design and development efforts include:
- Access to information or data
- Collaboration between parties both within and outside the enterprise
- Information flow between applications, application to people, and people to people.
Many envision significant opportunities for growth in Web-based re-deployment of existing applications that tightly integrates the above components under a portal environment.
Portal infrastructures provide a set of integrated services to facilitate the
flow of information (data) between applications, people, and processes.
IBM's strategy is to provide the infrastructure and components for creating
portal applications today while preserving a growth path to future levels
of comprehensiveness, ubiquity, and integration. These services are common
to the different portal types and usage scenarios discussed above, and
are organized into six broad categories as shown in the figure below. Information
services, application services, and collaboration services
constitute core services that provide access to information, applications,
and people and are coordinated and managed by the access and integration
services layer. System management services provide control of
the environment through instrumented applications, while presentation
services control the user interaction and user experience of the portal.
Portals differ, but most portals will include some combination of some of the services listed above. The remainder of the paper elaborates on each service category and describes the relationship among services.
A portal must provide aggregation services giving users a single point of access to multiple heterogeneous data sources, both structured and unstructured -- including relational database, multidimensional databases, document management systems, ERP databases such as SAP, e-mail systems, Web servers, news feeds, and various file systems and servers -- in a transparent manner. The system must support all document types and formats -- audio, video, image, text, postscript, and so on. The aggregation services connect all these data sources in a federated manner across multiple locations and companies. From a user's perspective, access must be transparent and seamless.
Portals need to be supported by a web content management platform consisting of well defined processes and tools that support the collaborative activities of content creators and site administrators. These activities include web based life-cycle management, authoring and designing Web based assets, controlling content delivery to the user and increasing value through appropriate content reuse and personalization." The current 1st paragraph now becomes the 2nd paragraph and the lead-in sentence to the bullet list should read "To maintain, manage, and provide access to various document types, portals also require an integrated set of document management services including:
- Document check in and out
- Document version control
- Document security for specific users or groups of users
- Extensive audit trails to identify who did what to which document and when it was done
- Lifecycle management of documents, including retention management and permanent archiving/destruction
Search technology helps users filter a growing morass of information to find information that is useful and relevant. Many different search techniques exist, but elaborate discussion on such techniques is beyond the scope of this paper. However, one important search technique is parametric search, which associates name/attribute pairs to search keywords. This technique is especially relevant to electronic commerce where, for example, a customer needs to find dress pants that are black, size 34, and cost under $40.
All search techniques rely on meta-data to satisfy a search request. Many search infrastructures rely on Web crawlers to create the search meta-data. A Web crawler retrieves Web pages for use by a search engine by recursively scanning and extracting URLs from a starting document or Web page. Every page that is scanned is indexed, sometimes ranked, summarized, and analyzed for its contents -- the information stored as meta-data. Due to the huge and growing number of pages on the Web, and the constantly changing nature of some pages, Web crawling can be a serious challenge and resource drain. Meta-data can also be created using other techniques or directly imported from data warehouses.
Many portal applications require business intelligence tools integrated into the infrastructure. Data mining services provide a rich set of content analyzers supported by visualization tools to identify patterns and relationships in data and interactions. Inferences from such relationships drive business decisions. Another emerging trend is logging of click-stream data so that it can be mined to enhance e-commerce revenues.
Portals must also provision syndicated content from a wide range of leading content providers. Such content includes news headlines and feeds, weather, stock quotes, financial and vertical industry news, as well as sports and entertainment news.
Application services for portals support two major functions -- application access services and application hosting services. Application access services are supported by Web application servers, which provide seamless access to both Web-enabled and legacy applications. Access to legacy applications is supported by a rich set of connectors running on the Web application server. Application hosting services simplify management and deployment of applications that are hosted externally by an ASP. Details of application hosting services can be found in the Framework's Application hosting white paper.
Portals may need to support real-time collaboration so that it is easy for users to find colleagues, partners, customers, and suppliers online, and communicate with them regardless of where they're working. These services allow users to chat, share live documents and applications, and create an instant shared workspace where team members can centrally communicate. Mail and shared calendar services are also instances of collaboration tools that are increasingly finding their place as integrated portal services. A unified messaging system for fax, mail, and telephone that converts text to speech is another growing portal requirement.
Another useful collaboration service for some portal applications is expertise location, which maintains profiles that can be queried directly by users to locate experts by skill, experience, project, education, job type, and many other attributes. The profiles get loaded through a variety of measures. Skill data is acquired through a metrics tool in the discovery engine, which mines topics out of documents and cross references them with user activity to infer interest and skill.
To effectively deal with the ever increasing amounts of data, applications and content can be organized into categories creating a taxonomy that can be easily navigated by users. Categories can be user- or administrator-defined. Self-organized categories are generated automatically via classification techniques. Often tools support is necessary for managing categories.
To dynamically assemble personalized documents, portals require content integration tools to manage discrete content objects, which can be combined on the fly to create different target formats. The target format may be PDF, HTML, CD-ROM, or print media. Language translation of discrete content objects can be handled automatically during the assembly of the target document. By providing the appropriate granularity level, a content management system can easily generate the necessary target documents without having to maintain redundant information.
Personalization encompasses the ability to provide the user with the right content both from the user's and Web site owner's perspective. It is based on gathering information about the user or communities of users and delivering the right content at the right time based on current context. A personalization algorithm determines whether content is presented to the user, and if so, in what order of priority. There are several different types of personalization techniques:
based, where user information is profiled in a directory and used for
making recommendations. The information may be provided by the user or
set on the user's behalf.
filtering is a personalization scheme to effectively identify individuals
who share similar tastes. Users are asked to rank alternatives; a rank
order of matches with other users determines proper recommendations.
recommendation learns and predicts a user's preferences by observing
usage patterns and from rankings obtained by methods similar to collaborative
- Content-based filtering is a scheme for automated personalization where suitability of a document is based on analysis of keywords and can be used when the selection criteria is not subjective.
A portal will provide access to many different applications often at many different sites. Access to these applications must be transparent and seamless without compromising security. Because portals must support access from both inside and outside the firewall, the usual set of security services are required: virtual private networks, intrusion detection, secure sockets layer (SSL), public key infrastructure (PKI), etc. Refer to the Framework's security services white paper for more information.
A key portal requirement is single sign-on, in which a user logs on once to gain access to all portal resources. Another key portal requirement is access management that supports filtration of access based on user profiles and an organization's content access policies. Because many different applications are made accessible to many different users or user groups in an almost real-time manner, access management must be role-based and hierarchical. For example, a person who assumes a new managerial role must have instant access to a set of applications derived from updated profile information.
In some portal scenarios such as the supply-chain portal described earlier, a key portal function is the ability to build and participate in a workflow process. Business processes across enterprises can function in an integrated manner with the help of business integration servers that broker information across multiple loosely-coupled applications that need to work together. The business integration server maintains workflow process definitions and routes business documents or messages via a robust messaging infrastructure. In this scenario, the portal functions as a workflow client handling messages or e-mail as they arrive.
In addition to its role as a workflow client, the portal can support a process manager that coordinates interactions from many different sources including collaboration servers, mail servers, and business integration servers in an integrated manner. Such interactions may be managed as a hierarchy of processes in which process state is persistent. The portal server ensures consistent data and view for roaming users, performs user task management functions, and coordinates the different services.
Portals must support access to data, applications, and people via a variety of devices -- including phone (wired and wireless), pagers, fax, handhelds, laptops, desktops, workstations, and servers. Transcoding services are required to filter or convert content to match the form factor and capabilities of the target device. Alternatively, appropriate type of content may be selected to best suit a particular device type. Services for disconnected use help users transmit and receive information reliably when they establish a network connection. Refer to the Framework's pervasive computing white paper for more information.
Search, aggregation services, categorization, and content integration are related via the underlying meta-data used or created by each. Proliferation of different types of meta-data associated with different types of portal resources coupled with strong demands to support a variety of personalization services will make it necessary to manage meta-data in an integrated and cohesive manner. Applications use meta-data to implement enhanced service functions. It is often difficult to determine a priori how data should be organized and what meta-data should be accumulated. As a result, authoring tools are required that support dynamic creation, editing, and rearranging of meta-data and associated parameters. Catalog tools provide a table of contents for the meta-data stored in the data warehouse. Most of these tools are designed for end user use.
A portal must be supported by a set of presentation services that manages the look and feel of the portal interface and enhances the user experience. Customization of a portal occurs at various levels. Users may wish to tailor the look and feel of the user interface including the taxonomy of applications that are integrated into the portal desktop. Users may desire a different look and feel from a specific portal based on the task at hand. For example, the user interface for recreational browsing at home may be different than the one used at work for business purposes.
Portals offer a desktop look and feel where the browser is partitioned into frames and real-estate carved out and allocated to applications. Application and content sources are wrappered into components (commonly referred to as portlets) that can be positioned on the screen in a plug and play manner. A portal desktop will need to support task management, which is not intrinsically supported by the current generation of Web browsers. Such task management may be extended to pervasive devices as well.
Portals need to support a variety of system management services including:
- Notification services. This service notifies a user when changes are made to the portal; for example, when a new application is made available or removed.
- Logging services. Logs can be maintained for every resource at a portal, including users, groups, applications, and content. The information may be used for a variety of purposes such as billing, debug, and trace.
- User registration services. This service registers first-time users at the portal, and collects and manages user information.
- Subscription services. This service supports subscription to specific types of content and notifies users when such content becomes available. Alternatively, the content can be routed to the user automatically.
- Administration services. This service provides a set of tools for performing a variety of site administration functions.
- Workspace. This service allocates and manages resources for portal users such as user workspace, disk quota, etc.
The portal services discussed above build on and extend the underlying Framework
infrastructure and services with functions required to support portal applications.
As illustrated in the figure below, the portal services affect the application
server software, application integration, network infrastructure, and systems
management layers of the Framework.
Many standards are of key importance to portals. IBM's portal infrastructure supports numerous standards including HTTP, SSL, XML, LDAP, TCP/IP, Java, EJB, SQL, ODBC, JDBC, and CWMI. Portals support integration of a diverse set of data, applications, and processes; and eXtensible Markup Language (XML) is a key enabler.
a content management point of view, XML provides a flexible source data
format for Web delivery and content integration, permitting multiple presentation
formats and/or media types (Web, CD, paper) to be supported from a single
source repository. XML allows smaller fragments of text, graphics, etc.
to be managed as reusable information objects that can be dynamically assembled
into final form documents or Web pages.
XML-based query language (XQL) and database schema standards such as XML-schema
will enable a common query infrastructure for search.
based meta-data standards allow flexible interchange of document meta-data
across systems. Such interchange supports federated searches as well as
integration of document meta-data. CWMI, for example, enables inter-operability
with other systems and extensibility of stored information.
enables interchange of information across complex, application-specific
business-to-business e-commerce applications (for example, airline technical
data and semiconductor data sheets). XML enables application integration
as well as integration of data into applications and processes.
the portal client, XML affords rich presentation formats that can be individually
tailored and controlled by the information consumer.
pervasive devices, XML-based markup languages will help transcode content
from one form to another. VxML, for example, will enable integrated speech
support in portals.
IBM already provides or will provide many of the portal services described in this paper. IBM's WebSphere Portal Server provides a comphrensive and integrated set of portal services that allows companies to build their own custom portal Web site that serves the needs of employees, business partners and customers. Users can sign-on to the portal and receive a personalized Web page providing access to the information and Web applications they need. With support for a wide range of syndicated content such as newsfeeds and weather channels, users have access to a variety of up-to-date information from a single screen.
Integrated with WebSphere Portal Server is IBM's Enterprise Information Portal which features a new assembly of APIs built mostly out of pre-existing technology that provide access to heterogeneous sources of data in a federated manner. It also features a newly enhanced search engine for structured and unstructured content in federated data stores that returns text, video, audio, and other information from a variety of data sources.
A new knowledge-management portal suite offering from Lotus code-named "Raven" includes application integration, content tracking and analysis, user profiling, and expertise location all integrated into a portal to manage personal and community information and activity. A key component of Raven, the Lotus K-Station that delivers a comphrensive presentation framework and rich collaboration capabilities, is currently available.
In addition to providing a unified point of access to information, applications, and people, portal services facilitate the development of integrated portal applications and business processes. The set of services described in this paper cater to many different types of portals and usage scenarios. Today, IBM provides many of the services described in this paper, some as disparate entities. Over time, the services will be supported by our portal infrastructure in an integrated manner.
Dr. Avi Saha is a member of Software Strategy team in IBM and has been responsible for co-ordinating IBM's portal strategy. He has worked in diverse areas such as multiprocessor system design, computer-aided design and verification tools, neural networks, learning systems, data mining, and graphics hardware design. Many IBM patents are based on his work. Dr. Saha has M.S. and Ph.D. degrees in Electrical and Computer Engineering from the University of Texas at Austin.