acm-header
Sign In

Communications of the ACM

Communications of the ACM

ACM Opens Portal


Works published by ACM since its inception have been built into a special online collection known as the ACM Digital Library. Nearly half a century of pioneering concepts and fundamental research have been digitized and indexed in a variety of ways in this resource. The retrospective capture for all ACM journals, magazines, and proceedings is nearly complete; filling gaps of out-of-print issues and refining the granularity of links will continue into the future.

The user community has grown rapidly over the past three years. More than 36,000 individual members of ACM subscribe. Four hundred corporate and academic libraries and consortia serve an even wider community. Moreover, the level of activity has increased exponentially. In the last 18 months alone, there have been 13 million full-text downloads from the ACM Digital Library, an average of more than 200 downloads per article. This activity has forced ACM to expand its server capacity and bandwidth several times, well ahead of anticipated upgrades. Indeed, it has been an ongoing challenge to deal with the success of this resource.

Thousands of users have contributed to the quality of this resource by reporting problems they encounter, spotting errors, and making suggestions for improvements. Broad community participation of this sort is not only welcomed by ACM; it is an essential factor in the evolution and enhancement of online facilities. Thousands of scholars have sent their comments to feedback@acm.org over the past few years. These have been analyzed and have resulted in the release of the ACM Portal.

The fundamental components of the ACM Portal are an enhanced version of the ACM Digital Library plus an extended bibliographic database, consisting initially of more than a quarter-million citations of core works in computing. These works are of all types (journals, proceedings, books, technical reports, theses, among others), and from all the major publishers in the discipline. The ACM Portal thus provides an Online Guide to Computing Literature and a "reading room" for ACM's own literature in the ACM Digital Library.

The ACM Portal has been built upon an entirely new system architecture to address primary concerns expressed by users around the globe: full-text download times, general system availability and performance, particularly the response times for searching. While ACM cannot do anything about the "last mile" of the pipeline, it has built a more robust system of parallel processors, redundant servers, vastly increased bandwidth, and arrangements for several global, dedicated points-of-presence.

The new system architecture is accompanied by a new data model and a modified information architecture for the site. These changes are intended to reduce the level of manual intervention required to maintain previous versions of the ACM Digital Library, particularly the handling of unusual bibliographic cases (which occur with alarming frequency in the publishing world), the performance of updates, and the maintenance of links. A scalable windowing system supports higher-level browsing in one window while corresponding lower-level, specific results continually refresh the display in a companion window.

A wider search of the world's computing literature is made possible by the bibliographic database. This is only one aspect of a functioning portal. Researchers not only need to develop bibliographies, but they frequently want to investigate particular sources more deeply. Direct access to expanded bibliographic information such as full-source data, abstracts, index terms, or reviews is useful, as are pointers to precise locations of the full text.

To serve these needs, the ACM Digital Library and the Online Guide to Computing Literature have been woven together with a set of links. For each article published by ACM, references have been extracted from the PDF representations. The universe of these references from the ACM literature currently totals somewhere between 750,000 and one million (non-unique) citations. By a process utilizing proprietary, fuzzy algorithms, these unstructured text strings of OCR'd citations are matched against the structured bibliographic entries in the Online Guide to Computing Literature.

Of the total set of references, approximately one-third are found within the bibliographic database. Internal links from the article's reference list to the bibliographic database are constructed. In every case where the reference is to an article published by ACM, another internal link is provided to the full text in the ACM Digital Library. In all other cases, the metadata for that structured database representation is written out and submitted as a query to a large, collective metadata database established last year by an international cooperative of publishers known as CrossRef (see www.CrossRef.org).

When ACM submits a reference query and it is matched, a Universal Resource Name (URN) in the form of a Digital Object Identifier (DOI)1 is returned and inserted as an external link from ACM's site to the source for the material. Supplementing these internal and external reference links are forward links, that is, to the citations of chronologically subsequent articles published by ACM that cite the one in hand.

While it is currently out-of-scope for the Portal to capture the references in works put out by other publishers, it is intended to create a comprehensive bibliographic database by capturing all the references in ACM articles that cite works from other publishers and for which there is no matching bibliographic entry in the database. This should go a very long way to establishing a near-complete bibliography, since it is a reasonable presumption that if not one of ACM's many authors ever cited a given work, that work is likely to lie beyond the periphery of core computing literature.

As a window into the world's core computing literature, the ACM Portal also expands in another direction: direct access to more full-text content. Users have already found publications from other publishers included in the ACM Digital Library. Some of these, such as the Linux Journal, receive a great deal of attention. In addition to the occasional journals that another publisher may want to include within the ACM Digital Library, entire collections from other societies may be now be hosted within the Portal. Hosted collections will appear as inter-society arrangements are concluded.

Back to Top

Browsing and Searching

Both components of the ACM Portal—the ACM Digital Library and the Online Guide to Computing Literature—are now indexed by the Verity Search engine. This change addresses the community's stated need for much faster searches. It also facilitates the quick development of a finite but potentially large number of browsable "views" into the literature.

The ACM Portal opens with several important views established. A View by Author allows the user to drill down to a page that might be called an "author's virtual bibliographic home page," listing all the works by that author known to the system. A future enhancement to this listing will enable the author to interact with it and add links to additional works and projects. The listing contains links to full source information and to full texts whenever known. Like any other listing, it may be sorted by the user in a number of different ways.

A view by the ACM Computing Classification Scheme (CCS) allows the user to drill down through ACM's own hierarchically structured subject tree from more general categories such as "operating systems" to narrower subjects like "process management," and then to specific topics like "threads." At all stages of these drill-downs, context is explicitly maintained for the user; the underlying database query is constructed and displayed; and the option of manually modifying it is always present.

The view by CCS is supplemented with a view by Subjects, which presents the user with an alphabetical listing of the Permuted Word Index to the Classification Scheme. This is especially useful for those less familiar with the formal structure of the CCS.

One other noteworthy preconstructed view is by Technical Interest. This view utilizes a user-selected and editable profile. As such, it is one of several customizable services available within the ACM Portal. In the absence of such a profile, the view defaults to the full set of predefined Technical Interests.

New views can be readily generated and they will be constructed according to user feedback received by ACM. Views are "trainable" and "tunable" by adjusting a sophisticated set of parameters and indices.

Views exist side by side with Search. A quick Search is available on all Results Pages. Any word or phrase entered will run across all available indices. Result sets always tell how many were found out of how many possibilities. At the level of a specific Citation Page, a search can be launched to find Similar articles. Results are returned in order of relevance as ranked by the system. The algorithms used for Similar searches are not obvious, but involve the use of dictionaries, thesauri, and cluster analysis of full-text indices. Experiments with the Similar search function have returned some unexpected but very interesting results.

Advanced Search provides for Boolean operations across specific fields with limiting constraints. These searches may be written directly by users familiar with the syntax, or entered into a template and constructed by the system. Librarians especially may find interesting possibilities here relevant to their holdings. One can see within the Portal today, for example, that John Wiley & Sons has published 2,953 books in the last 16 years, 245 of which have "software engineering" as their primary subject.

Back to Top

Personalized Electronic Community Services

The contents of both major components of the ACM Portal (the Online Guide to Computing Literature and the ACM Digital Library), can be manipulated with a set of customizable tools. Individual users may personalize their space, creating their own "carrel" within the Library, where they organize and store copies of the articles they want to read or build bibliographies for their own future research.

These custom collections have several properties. Their contents may be manually chosen and inserted, or they may be accreted by the system based upon user-selected criteria or saved searches. Users may choose to be alerted by email if new material is added to a customized collection, or simply view what is new by periodic inspection. Owners of such personal collections may choose to share them, with specific individuals or with an entire predefined group, like a specific ACM SIG. No one automatically receives such a personalized collection. Rather, one is invited to share by the creator, and may opt to use the collection or ignore the invitation.

Primitive versions (although often malfunctioning) of these services existed before the release of the ACM Portal. Today, there is a great deal more power and richness in these services and they are also easy to learn and fun to use. These personalized tools can serve to facilitate cooperative work, course-pak development, even electronic reserves for a class. The groundwork has been laid for building additional electronic community services, such as uploading user-created documents like syllabi or lecture notes into a shared collection. Private conversations, group discussions, and public forums will be supported; online reviews of the literature are a feature of Online Computing Reviews.

Back to Top

Pricing and Access

Open House for the Online Guide to Computing Literature is occurring from now through September. In October, it will become an independent product for everyone except individual members of ACM who subscribe to the ACM Digital Library. For these subscribers, it will be bundled with the ACM Digital Library into the ACM Portal. Their subscriptions will be upgraded; they will automatically become Portal subscribers. Their first renewal will be priced at $110. The following year, beginning in July 2002, renewals will be at the full price of $125. The full price for students will be $45 in July 2002. Until then, upgrades to the Portal are billed at $40. Students may elect to receive print copies of Communications in lieu of the Portal, or pay separately for the print.

ACM members who do not wish to subscribe to the Portal may subscribe to the Online Guide to Computing Literature for $55. If they wish to download any full texts, they will pay a per-article transaction fee. Individuals who are not members of ACM may subscribe just to the Online Guide to Computing Literature at a price of $175.

All institutions may buy access to the Online Guide to Computing Literature or to the ACM Digital Library or to both. Pricing varies for academic and corporate institutions, beginning at $1,200 for academic institutions currently subscribing to the ACM Digital Library; steep discounts apply for academic consortia and corporate-wide licenses. For more information, please visit www.acm.org/membership/library/.

For the first time, patrons of an institution who are not members of ACM will be able to acquire personalized services, by buying their own individual accounts. Institutions wishing to purchase a block of personal accounts for their most intensive users, may do so.

The new Online Computing Reviews (see www.Reviews.com) will be treated in a similar fashion. Access to it will be bundled into the ACM Portal for individual members of ACM. It will be a separate, unbundled product for non-members and institutions. Individual patrons of institutions who subscribe to Online Computing Reviews will be able to create their own personal profiles to make use of this journal and its service. The printed edition of Computing Reviews will remain a part of the ACM Institutional Core Print Package.

Back to Top

Author

John White is the CEO of ACM in New York City.

Back to Top

Footnotes

1 Arms, W. Uniform resource names: Handles, PURLS, and digital object identifiers. Commun. ACM 44, 5 (May 2001); www.acm.org/pubs/citations/journals/cacm/2001-44-5/p68-arms/

Back to Top

Figures

UF1Figure. System overview.

Back to top


©2001 ACM  0002-0782/01/0700  $5.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2001 ACM, Inc.


 

No entries found