design and implementation of a web-based ... - Semantic Scholar

Journal of The Institution of Engineers, Singapore Vol. 44 Issue 3 2004

DESIGN AND IMPLEMENTATION OF A WEB-BASED PERSONAL DIGITAL LIBRARY Sheng-Uei Guan1 and Xiaobiao Zhang2 ABSTRACT This paper proposes a mechanism called Personal Digital Library (PDL) for networkbased personal computers. PDL provides a new Web service paradigm. The Memex (“memory extender”) device proposed by Vannevar Bush aimed at solving the problems of "locating relevant information in published records and recording how that information is intellectually connected". Proposed as a new information infrastructure based on the Web and personal computers, PDL is a Memex-like entity enhanced with the ability to connect and interact with each other. Information like bookmarks in a PDL can be made shareable with other PDLs. A web-based memex is formed by linking PDLs together. Bookmarks are organised with other shareable information, rather than placed into a separate file. External PDLs can serve as memex to overcome storage limitation in the local computer and avoid information duplication. PDL integrates current Web and file services such as browsing, retrieving, publishing, organising and searching information. PDL applies the bookshelf metaphor for users to organise their books and personal folders. PDL also provides intelligent support during navigation. For instance, users can know which books in the PDL related to a particular topic are frequently visited. PDL also has efficient notification and subscription mechanisms for its bookmarks. PDL proposes a new type of collaboration, users can know the "traces" left by another visitor. This paper gives the details on PDL design and prototyping. INTRODUCTION As early as 1945, Vannevar Bush proposed a personal information machine called Memex (memory extender) [1]. Memex later became an influential idea and by the 1980s it was hailed as the inspiration for hypertext and new ways to organise and retrieve information. To deal with the explosion of scientific information, Bush’s proposal for Memex focused on the problems of "locating relevant information in the published record and recording how that information was intellectually connected" [1]. Memex was never built. The expansion of information will be more serious as the Internet grows. Researchers have to consider the problems of "locating relevant published records in the published machines and recording how those records are intellectually connected". Today's personal computer has a huge storage and surprising speed, so it is available for use as a published machine, i.e. a server. In the near future, all computers will be network-based and server-based. This means that people can publish information and 1

Department of Electrical & Computer Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260 Fax: (65)779 1103 email:[email protected]

2

School of Computer Science and Computer Engineering, La Trobe University, Vic. 3083, Australia

59


allow other people to use their information resources. They can also visit any computer around the world, use the information resources inside and carry out online discussion. This paper proposes a new service paradigm for a network-based personal computer to publish, browse, search, retrieve, and organise information on the Web. To avoid duplications with file systems, this mechanism organises information collections including application programs, but adds new features to meet the needs arising from the integration with the Internet. Information in a machine can be made shareable with other machines, and other machines could be part of a machine's "memory extender". This proposed mechanism is called a Personal Digital Library (PDL). PDL will realise the selections of books, PDLs and related persons through intellectual associations. As a server, PDL can store information for easy retrieval and search, and provide intelligent support for users to browse and find information. As a client, PDL can concurrently retrieve information from different PDLs (servers), and find a related person to communicate with during browsing. Other PDLs can serve as memory extensions to overcome storage limitation in the local computer and avoid information duplication on the Web. PDL can be applied across platforms such as Macintosh, PC and Unix for personal use. PDL has potential uses in many areas such as personal use, education, commerce, finance and entertainment. Personal users can employ PDL with an Internet connection to manipulate their distributed information from anywhere around the world. They can also apply PDL to Internet shopping and real digital libraries. This paper introduces PDL design and prototyping. The prototype implemented shows that the PDL concept is feasible. Although PDL is designed to manage personal information collections in a network-based personal computer, it will be fruitful that if the ideas in this paper can stimulate further development in Digital Library research. RELATED WORK Memex Bush described Memex as “a mechanised private file and library” and as “a device in which an individual stores his books, records, and communications, and which is mechanised so that it may be consulted with exceeding speed and flexibility”. Memex would store information on microfilms, which would be kept on a user’s desk. Memex would have a scanner to enable the user to input new material. Bush emphasised in his paper, Memex Revisited [2], that the mechanisation of “selection by association” would bring about a successful personal machine that would allow a human being to “think creatively, unencumbered by unworthy tasks,” and that allow people to “face an increasingly complex existence with hope”.

60


The proposed PDL is a network-based Memex, based on a personal computer. Information including bookmarks in a PDL can be made shareable with other PDLs. A Web-based “memory extender” is formed by linking PDLs together. Another feature of PDL is to allow people to locate resources quickly through intelligent associations. Digital Library PDL has similar goals as most digital libraries (DLs). In terms of service and resource provision in a DL, a number of service levels should be considered [3]: services or resources provided, resource provision, self-assisted services, interactive services, videoon-demand services [10], and knowledge-based services. PDL focuses on self-assisted services and resource sharing. However, there are reasons that PDL is distinct from most DLs. First, it is fully self-assisted because any PDL uses the same way to organise its collections, and has the same user interface, therefore, users do not need to be assisted. Second, as to interactive services, a PDL user might discuss the use of a particular book or PDL with other users or PDL owners. Third, PDL has knowledge-based services. Examples include regularly checking changes of the favourite books and then automatically informing the owner; providing access details of each book; and automatically forwarding a query to other PDLs. The following introduces two DL research projects, funded under the NSF/NASA/ARPA Digital Libraries Initiative [4]. The U.C. Berkeley Digital Library Project [5] This project produced a prototype digital library with a focus on environmental information. The goal is to develop technologies that support "work-centred" digital library services, oriented to address the mission of work groups. The work-centred approach is user-oriented, emphasising technological developments directed towards users needs. First, users require information systems that can perform powerful retrieval and analysis of heterogeneous objects. Second, users must be able to access their own collections of various types, in addition to external data collections. This requires flexible authoring, structuring, and delivery mechanisms. Third, users must be able to integrate an information system into their established work practices, even as the system augments those practices. System interoperability is essential and may require custom interfaces [6]. PDL is a work-centred digital information system, because the services provided by PDL also concentrate on the user needs as mentioned above. PDL improves the way to organise information for easy retrieval and search. PDL users can browse and even use different types of collections (books) not only from the local PDL but also from external PDLs. In addition, PDL provides users with a universal, customisable interface to perform information organising, browsing, retrieving, searching and publishing.

61


The Stanford Digital Library Project [7] The goal of this project is to develop the enabling technologies for an integrated “virtual" library providing uniform access to the large number of networked information sources and collections. Another goal is to provide high level concepts and protocols that can allow users to access information through interfaces that hide the unimportant details of diversity of materials and provide ways to navigate and manage the "information space". Therefore, the project focuses on integration and interoperation [8]. The Integrated Digital Library creates a shared environment that links everything from personal information collections, to collections found in conventional libraries, to large data collections shared by scientists. PDL is also an integrated information system designed for network-based personal computers. Information including bookmarks in a PDL can be made shareable with other PDLs. Since bookmarks become shared resources, the information collections from other PDLs can be linked together. Microsoft Internet Explorer Compared to previous releases, Microsoft Internet Explorer has improvements in four key areas: Web browsing, Communication and collaboration, Webcasting of sites and channels, and Web integration. Other than being a browser, IE can be used to navigate and manage the local file system, and can provide communication tools. The tight integration of the browser, communication components and even application tools with the operating system is one of the important purposes of presenting PDL. To replace browser on functionality, PDL is to inherit functions such as Webcasting and Web Integration. However, PDL adds some features, for example, users can navigate either local PDL or other PDLs on the Web with the same interface. Bookmarks are organised with other shareable information rather than being placed into a separate folder. Therefore, PDL will be more powerful than most existing browsers. PDL DESIGN Architecture Fig. 1 shows a typical PDL architecture and information flow. A PDL consists of the following components: • Library Explorer The Library Explorer is used to establish a connection with another PDL and to navigate the connected PDLs. The hierarchical structure of each connected PDL will be represented in the Library Explorer, which serves as a tool to integrate the local PDL with the connected PDLs in a tree and to make the stored information available and visible for visitors. The Library Explorer can also show which books are frequently accessed through the Introduction Panel and provide a Bookshelf Panel to show books being used.

62


A bookmark can be made on a PDL, a folder, a book or a bookmark. It will be classified and then published in the hierarchical tree. All these bookmark-related processes are carried out in the Library Explorer. The user interface of the PDL search engine is embedded in this component, so it is convenient for the PDL owner to do a search in either local PDL or any connected PDL. In addition to a search engine, the Library Explorer includes other mechanisms for processes such as subscribing to books and keeping bookmarks updated. Meanwhile, it allows users to customise different views to see the selected PDL. In summary, the Library Explorer is an integrated tool for information organising, browsing, retrieving, publishing and representing. It is also a control panel to manipulate other mechanisms and application tools, and can give assistance to visitors during navigation through a series of views and facilities. Thus the Library Explorer is a service centre in a PDL. •

Reading Area Books collected for reading are placed in the Reading Area together with the corresponding reading tools. The Book Browser is the tool for reading hypermedia books, and has traditional browser features. PDL owners are allowed to browse books from different PDLs at the same time. They can also return to the page lastly visited of a hypermedia book, before switching to another book. To implement this feature, PDL provides a history list for every hypermedia book browsed, which is separated from the Book Browser. In order to utilise hardware resources and improve performance, the Book Browser can download the rest of the pages while browsing a page. Thus, it supports offline reading.

•

Communication Area This area contains communication and collaboration tools such as e-mail, online chat and net conferencing. Using these tools might interrupt other people. For instance, users will be alerted when a mail comes in, or will be alerted by a ring for chat or conference. Through the Remote Bookshelf panel, the PDL owner can check what other people are doing at a remote site without interruption.

•

Interoperability and Protocol PDL can be embedded in either a personal computer or a mainframe account, which can be network-based and can run a Web server. PDLs in the world can be connected over the Internet. The communication between any two PDLs is peerto-peer via HTTP. It is achieved by the Web servers that both PDLs are running. In this way PDL is platform independent. Using peer-to-peer communication, a PDL may initiate information exchange with any other PDL at any time.

•

Local Storage The Local Storage is a repository in which information collections are made shareable and organised in a topic-based hierarchy. For a personal computer, the hard drives serve as the local storage; in a mainframe, the local storage can be the whole capacity of a personal account.

63


Figure 1: PDL architecture

Features PDL is a new Internet client/server desktop offering a number of features in seven key areas: information organising, information browsing, information publishing, information searching, communication and collaboration, intelligent support and security protection. Table 1 summarises the PDL features. Once these features are implemented, PDL will give Internet users a completely new browsing experience on the Web. We detail some of them in the following sections. Information Organising In PDL, a book is a collection of relevant information for a specific purpose, for example, a newspaper, a biography, a tutorial and even a software package. Books are classified by subjects, and placed into related folders. PDL is structured as a hierarchical tree. This is illustrated in Fig. 2, which shows how books and folders are hierarchically organised. In addition to a title field, PDL provides both books and folders with a description field, which gives a brief introduction or the current author’s comments. Such a topic-related association makes it convenient for PDL visitors to get what they want, especially for those who just have a topic in mind. In Fig. 2 it can be seen that bookmarks are located in a hierarchical tree. This is an innovation, compared to traditional bookmark management. As one of information resources, bookmarks should be made shareable so that visitors can find useful information not only from the owner’s published books but also from his or her favourite information. On the other hand, if more and more bookmarks are added, the traditional method of putting all bookmarks into a separate file (or folder) will be difficult to handle

64


bookmark classification and searching. Therefore, PDL uses a tree to organise all bookmarks. Books, folders and bookmarks are all integrated into one tree. More importantly, other PDLs can be easily linked together through sharable bookmarks to form the local PDL’s ‘memory extenders’. Table 1: PDL feature summary Area

Features

Information Organising

Books, folders and bookmarks are organised in a topic-based hierarchy. Bookmarks are made shareable together with books and folders. Books, folders, and bookmarks have the same data structure (properties). Properties can summarise an item and keep themselves updated and consistent.

Information Browsing

The Windows Explorer-like user interface and behaviours are applied. There are no differences between the owner and visitors. Users can examine the properties of each item. A special area shows the books collected for reading. True offline browsing is offered. No separate tool is required to publish information. Published items are immediately available to visitors. An item’s properties are accessible.

Information Publishing

Information Searching

The search interface, query format and result format are uniform. The search engine is used to find bookmarks. The query can be forwarded to other PDLs. The owner has the same search means as visitors.

Communication and Collaboration

All tools can be supported. A new tool is for observing what other people are doing.

Intelligent Support

An area is provided to show hotter books. Notifications will be made automatically and accurately when bookmarked items are changed. Efficient subscription is provided for bookmarked hypermedia books. A PDL can be customised by using different views provided There are three classes of protection and three types of access.

Security Protection

In such a shareable resources environment, bookmarks are classified and published together with books in the same tree. In Fig. 2, it can be seen that those books and bookmarks with the same topic are grouped together. From the PDL owner’s view, he or she can know which books in the local PDL are on a particular topic, and also know where to locate those books and folders related to this topic in external PDLs. Visitors will have a wider range of choices: when visiting one PDL, they can retrieve relevant information from other PDLs. This would save visitors a significant amount of time

65


during navigation. Thus, PDL offers three ways to view the tree for customisation, “show books”, “show bookmarks” and “show both”.

Figure 2: PDL hierarchical tree After introducing these PDL features, it is necessary to review three terms: • Book A book is a collection of relevant information and is a leaf node in the PDL tree. A hypermedia book is similar to a Web site containing hypermedia pages, and any application software can be regarded as a book. •

Folder A folder is a container of related books and sub-folders in a PDL. It is a non-leaf node in the PDL hierarchical tree.

•

Bookmark PDLs, books, folders and bookmarks can be bookmarked. When a bookmark is made for an item, its properties excluding the Access Times will be copied to the corresponding fields in the bookmark. In this way the information that summarises the item can be kept. This information will be used to update the bookmark and to make it consistent through comparing to the item later. Bookmarks can also work as a reference for an item to be used in the local PDL.

66


Information Browsing As the Windows Explorer-like user interface and behaviours are applied to information browsing, the PDL appearance and manipulations are familiar to most users. Once a PDL is connected, a PDL icon with the connected PDL’s name is shown on the Library Explorer. Through opening folders, visitors can easily navigate this PDL. After a selection is made on a book, the book will be moved to the Reading Area (See Fig. 1) in which the corresponding application is run depending on the book type. Visitors can also use the same interface to browse their own PDLs and read the books. Note that some books, folders, or bookmarks are not accessible to visitors. This could be so due to privacy concerns. Such an access control can be imposed by having an access-control list in place for each item (e.g. books, folders, or bookmarks) in concern. An access-control list can list the users who have the right to access the particular item or alternatively, it can list all the users who do not have the right to access the item. Access right can be further differentiated, i.e. ‘read’ or ‘write’ access. For PDL, usually only ‘read’ access is granted.

Figure 3: GUI design of the Library Explorer Fig. 3 is the GUI of the Library Explorer and shows the view of Fig. 2. The Library Explorer provides four display panels to show information in a PDL. The Library Panel

67


shows the currently connected PDLs and the local PDL with a tree-view structure. The Main Panel shows folders, books and bookmarks in the currently opened folder with a list-view structure. The books that are frequently accessed are displayed in the Introduction Panel together with the description of a selected item. The Bookshelf Panel contains the books selected from different PDLs (including the local PDL) by the PDL owner. In the Main Panel, a flag or colour is used to identify those frequently accessed items in the opened folder. This panel is also designed for showing a list of search results, a guest list of a selected item and the content files of a selected book. Information Searching As an information management system, PDL must offer search services for visitors and the owner. The search interface, the query and the result format currently depend on the search engine applied, so users have to learn how to use the search engine before starting. In contrast, the same search engine is applied to every PDL. This benefit is brought by the same information structure in all PDLs. In addition, apart from books, a PDL owner can use this search engine to find bookmarks in his PDL. This is why PDL can handle a great number of bookmarks. PDL offers title and keyword searches as in a traditional Web search engine, both the Title and the Description fields of each related item will be searched in a keyword search process. Therefore, there would be more matched results than only searching the Title field. The Library Explorer provides users with a unique interface linking to the PDL search engine. Fig. 4 shows this interface. The first three boxes decide what type of item is to be found. The others tell the search engine how to search.

Figure 4: PDL search engine GUI Users do not have to worry if the information required is not in the currently selected PDL because they still might get the information from external PDLs by checking the Search External PDLs box. Under the default settings shown in Fig. 4, the scale of this

68


type of search is the largest. Fig. 5 illustrates how PDL handles a query either from the external PDL or the local PDL.

Figure 5: PDL processing query The following description presents the search procedure under default settings shown in Fig. 4. Depending on these settings, a matched item can be a book, a folder or a bookmark. This search is also keyword-based, so both the Title and Description fields of each item will be checked. The search engine will use the input keywords to search for books, folders and bookmarks in the PDL. If no matches are found, the same query with the settings will be created and forwarded to some preferred PDLs. The same procedure will then be applied in these PDLs, except that they need not forward this query if no matches are found. Thereafter, they will send the search results to the original PDL. Either the PDL owner or visitors can do a search in a PDL through the same search interface. It is useful that the PDL owner is provided with the same search means as visitors. There are two advantages of this approach. One advantage is that it is convenient, because the owner needs to use the search function when there are a number of items in his or her PDL. This is especially true for bookmarks. As it is not possible to remember all the bookmarks, the PDL owner just needs to type in some keywords and then can quickly get the desired bookmark by checking the Bookmark box. The other advantage is that it is efficient, because the query can be forwarded to more than one external PDL without any other particular operations. If the Search External PDLs box is selected, PDL should check bookmarks first before doing a search in some preferred PDLs. This process is efficient as most bookmarks represent the items of external PDLs.

69


Intelligent Support PDL offers intelligent support during navigation. PDL provides an area (i.e. Introduction Panel) to inform visitors which books are more frequently visited by other people. Visitors can also know which items are hot or new in a folder through some flag or colour. Together with the selected item description shown in this area, it seems that there is an on-line librarian to help visitors to navigate in a PDL. PDL is able to inform the owner automatically when those bookmarked items have been changed. Most properties of an item depend on the related properties of its immediate children, and a bookmark has the same data structure to record properties for the bookmarked item. In this way the changes can be detected through comparing each property. In addition, PDL provides two options keeping bookmarks updated and consistent: it checks regularly and checks when there is little or no workload. Both checks are done in the background.

Figure 6: Book subscription process PDL also provides subscription for bookmarked hypermedia books. In each book there is a property, Update Period, to let visitors know when it is regularly updated. This property can be used as a reference for subscription. PDL would make a subscription on 70


time, depending on the Update Period and the Modification Date in the bookmark. Fig. 6 illustrates the process to subscribe to a book. PDL starts this subscription process regularly in the background for each bookmark marked by the PDL owner, and places the subscribed books in a special folder. The PDL owner can go to this folder and read those books when desired. IMPLEMENTATION PDL Prototype A runtime PDL prototype has been developed under Windows using Microsoft Visual Basic. To demonstrate that PDL can be available across platforms, a PDL is set up manually in Sun OS Unix machine running a Web server. By using this PDL as a server role, the communication between two PDLs can be tested. Any Web server can be used. The current selection is Apache [13]. Using the prototype, a user can construct his local PDL and navigate either the local PDL or an external PDL. The hypermedia books and property files in a PDL can be browsed. Finally, the Library Explorer can demonstrate PDL features and services. Fig. 7 shows the Library Explorer with a couple of connections with other PDLs in the prototype. The Internet folder in Steven’s Library is being navigated. In this folder there are two books opened and shown in the Bookshelf Panel. The description of current selection is shown in the Introduction Panel. Fig. 8 is the view of Show Bookmarks and looks like a bookmark system.

Figure 7: A Library Explorer for navigation

71


Figure 8: Bookmark view in PDL

Figure 9: A PDL and its mirrored PDL

72


PDL Enhancement To improve the performance of PDL, offline browsing is implemented. A PDL owner can decide whether to build a mirrored PDL for the connected PDL in the local storage during navigation or to create a cache folder for information retrieved as is done in many browsers. For example, IE has a cache folder normally called Temporary Internet Files under the Windows folder. PDL does not assign a fixed place for cache contents. To build a mirrored PDL, the “root” cache folder is created to represent the connected PDL and is labelled with its name. When a folder in the PDL is first opened, its corresponding cache folder is created using the same name under the “parent” cache folder in the local storage. Similarly, when a hypermedia book is selected for browsing, a cache copy is also created in the relevant folder, and the browsed pages can be retrieved and put into this cache copy. In this way the mirrored PDL is built by following the navigation trail; in other words, if all folders are visited in the connected PDL, the mirrored PDL will have the same hierarchical structure. Fig. 9 shows a PDL and its mirrored PDL that is built following the navigation trail. More importantly, PDL does not need to make a connection for later navigation except for updating current information. PDL provides an option to select to browse a hypermedia book either online or offline. A complete book can be downloaded into the cache folder from the connected PDL. PDL also has options to improve the browsing performance, for instance, the remaining pages can be downloaded at the same time when the PDL owner browses a page of a book, and another book can be downloaded for later use when a book is browsed. Discussions For PDL to be put into daily use, some questions may arise. For example, will there be a limit to the number of currently connected PDLs being displayed at the Library Panel? In a large organization or institution, when there are more than 50 users concurrently connected, will there be any technical problem? Ideally, there should be no constraint on the number of currently connected PDLs being displayed at the Library Panel or the number of users concurrently connected to a PDL. However, the constraint induced by limited screen size, network bandwidth, CPU or disk access speed may be in place. Note that even though a large organization may have more than 50 users concurrently connected, the workload may not be high due to the nature of user interactions, i.e. not all the users would be active at the same time. Another question that may be raised is: what is the difference between PDL and a typical personal portal service? First of all, a portal service generally provides services only to visitors, instead of the user. In contrast, PDL provides an integrated service framework to the user and visitors. In addition, tools like “library”, “trails” offer more functions to the user so that he can organize information in his own way, while these services are applicable not just to his webpages (under his portal) but also to his folders and files and even external files – accessible via linked PDLs.

73


FUTURE WORK Research on PDL is still in a prototype stage. Several further research issues are presented in the following. • Remote application use PDL owners should be able to reference an application tool and use it transparently in the local site. The manipulations and output should be the same as those being run locally. This allows those applications occasionally used to be stored in a trusted external PDL without occupying the local storage. This feature has been implemented in LAN. As the Internet grows, using remote application needs to be extended from LAN to the Web. It is believed that the PDL architecture can help this extension. •

Remote management on own PDL PDL owners should be able to access and manage their PDLs from anywhere in the world in the same way as from his machine.

•

Search protocol It is necessary to design a search protocol to fully fulfill the PDL search requirements. Following the PDL design, items in a PDL have to be checked against their Title and Description fields rather than only their Title fields to find matched items. In addition, a PDL may receive and process a number of queries from different PDLs at the same time. Therefore, a new search protocol is needed to make the search efficient. Another function of this new search protocol will be that a PDL can forward a search query to other PDLs and handle the results that will come back after travelling through a couple of PDLs.

•

Information retrieval protocol As the above three issues are all involved in exchanging information, it is necessary to improve HTTP and/or design a new one. Based on the PDL design, exchanging information in document format is not suitable. Therefore, the new protocol should enable the manipulations and output of any application in the remote site to the local site.

•

Hierarchical PDLs To overcome storage limitation and avoid information duplication, it is expected that a couple of distributed PDLs can be integrated into a PDL tree to form a virtual PDL. This means that there is still one PDL when this PDL tree is viewed from the root. In fact, each PDL virtually acts as a branch under “My Library”. Applying PDL to set up a real digital library is a good application example for hierarchical PDLs. Each integrated PDL will contribute to a particular function area. The research on hierarchical PDLs will raise a number of interesting issues such as how to organise these distributed PDLs, how to keep hierarchical, distributed information consistent and how to make search efficient.

74


•

Intelligent support Intelligent support should assist visitors to avoid wasting time on searching and manipulating PDL. The more intelligent support that PDL provides, the more assistance visitors will get during navigation. How to build intelligent agents into PDL is an area for further research. An intelligent search agent may be built into PDL. PDL should also be able to build a security agent to block illegal access so that it can concentrate on legal information processing.

•

3D PDL PDL in a 3D virtual reality friendly user interface can be considered. Such an interface will make intelligent support lively and information browsing and searching efficient. Therefore, based on this feature, VRML-based collaborative user interface for PDL is another valuable research direction.

Although the PDL prototype has been implemented using existing technologies, the implementation of a full-fledged PDL still has some hindrances to overcome. In addition to improving the PDL design, support from hardware, operating systems, communication protocols and network resources is also important. First, a personal computer used for PDL must have fast speed and large storage, and be network-based with an assigned address and name (e.g. IP and Domain Name). Second, operating systems must be able to allow PDL services to replace (or sit on top of) their file services. Thirdly, an efficient, reliable and safe communication protocol will be required to exchange information. Finally, PDL software will have to be made available (and FREE of charge) by some trusted party. The PDL software will need to be properly maintained and upgraded regularly. And a sufficient number of users will have to use it to make PDL united not a dream but a feasible routine task. CONCLUSIONS As a mechanism for providing a new service paradigm on the Web, the Personal Digital Library goes further to address the “trails” problems from hierarchically organising information and also making it shareable. By linking PDLs together, a web-based “memory extender” is formed. With PDL, a user will be able to have information at his fingertips by: following trails, browsing across PDLs, or using the PDL searching mechanism. With information readily available, a user may then focus his energy on thinking or being creative, without the need to muddle through heaps of folders, searching for information which may end up nowhere to find, or browsing across huge volumes of books painfully when trying to recollect his prior “trails” just to piece together some loose ideas. With PDL, the effort of retrieving relevant information has been made minimal. Retrieving information will no longer be a hurdle to the road of creation. In conclusion, PDL is a Memex-like entity enhanced with the ability to connect to other PDLs. A fully implemented PDL can fulfil Bush’s vision in more aspects than Memex. This Memex-like entity will become a successful personal machine that allows a human

75


being to “think creatively and wisely, unencumbered by unworthy tasks,” and that allows people to “face an increasingly complex existence with hope” on the Internet. REFERENCES Vannevar Bush. As we may think. Atlantic Monthly, 176, 1, July 1945, pp. 101-108. Vannevar Bush. Memex Revisited. In V. Bush, editor, Science is not Enough, William Morrow and Co. Federal Information Processing Standards Publication (1998). Analogue to Digital Conversion of Voice by 2400 bits/Second Mixed Excitation Linear Prediction (Draft). Sarah Ormes and Charles R. McClure. A Comparison of Public Library Internet Connectivity in the USA and UK. In Sarah Ormes and Lorcan Dempsey, editors, The Internet, Networking and the Public Library, Library Association Publishing, 1997, pp. 35. NSF/DARPA/NASA Digital Libraries Initiative Projects. Access information available at: http://www.cise.nsf.gov/iis/dli_home.html Robert Wilensky. The U.C. Berkeley Digital Library Project. Communications of the ACM 38(4), April 1995, pp 60. Project Homepage: http://elib.cs.berkeley.edu/ Robert Wilensky. Toward Work-Centred Digital Information Services. IEEE Computer 29(5), May 1996, pp37-44. Stanford Digital Libraries Group. The Stanford Digital Library Project. Communications of the ACM 38(4), April 1995, pp 59-60. Project Homepage: http://www-diglib.stanford.edu/ Andreas Paepcke, Steve B. Cousins, Hector Garcia-Molina, Scott W. Hassan, Steven P. Ketchpel, Martin Röscheisen and Terry Winograd. Using Distributed Objects for Digital Library Interoperability. IEEE Computer 29(5), May 1996, pp61-68. Scott W. Hassan and Andreas Paepcke. Stanford Digital Library Interoperability Protocol. Technical Report SIDL-WP-1997-0054, Stanford University, 1997. WenSyan Li, Quoc Vu, D.Agrawal, Y.Hara, and H. Takano, “PowerBookmarks: A System for Personalizable Web Information Organization, Sharing and Management”, Computer Networks, 31, May 1999. Chang-Li Lin and Sheng-Uei Guan, "The Design and Architecture of a Video Library System", IEEE Communications (Featured Topic Issue on Enterprise Networking", 8619, Vol. 34, No. 1, Jan. 1996. Luis Gravano, Hector Garcia-Molina and Anthony Tomasic: The Effectiveness of

76


GlOSS for the Text Database Discovery Problem. In Proceedings of the 1994 ACM SIGMOD Conference, May 1994, pp.126-137. Sheng-Uei Guan, Hsiao-Yeh Yu, Jen-Shun Yang, "A Prioritized Petri Net Model and Its Application in Distributed Multimedia Systems", IEEE Transactions on Computers, 477-481, Vol. 47, No. 4, Apr. 1998. Microsoft Corporation. Microsoft Internet Explorer 4.0: The Web the Way You Want It Reviewers Guide to Final Release. January 1998. Available at: Apache 1.3. Access information available at: http://www.apache.org/ Sheng-Uei Guan and Sok-Seng Lim, "Modeling Interactive Distributed Multimedia Application by a New Class of Reconfigurable Petri Nets", Proceedings International ICSC Congress on Intelligent Systems and Applications, University of Wollongong, Australia, Dec. 12-15, 2000.

77