Users, User lnterfaces, and Objects:
`Envision, a Digital Library
`
`Edward A. Fox, Deborah Hix, Lucy T. Nowell, Dennis J. Brueni, Wililam C. Wake, and Lenwood S. Heath
`Department of Computer Science, Virginia Tech, Blacksburg, VA 24061-0106
`
`Durgesh Rao
`National Center for Software Technology, Juhu, Bombay 400049, India
`
`Project Envision aims to build a "user-centered data-
`base from the computer science literature," initially
`using the publications of the Association for Com-
`puting Machinery (ACM). Accordingly, we have inter-
`viewed potential users, as well as experts in library,
`information, and computer science —to understand
`their needs, to become aware of their perception of
`existing information systems, and to collect their rec-
`ommendations. Design and tormative usability eval-
`uation of our interface have been based on those
`interviews, leading to innovative query formulation
`and search results screens that work well accord-
`ing to our usability testing. Our development of the
`Envision database, system software, and protocol
`for client-server communication builds upon work to
`identify and represent "objects" that will facilitate
`reuse and high-level communication of information
`from author to reader (user). All these eftorts are
`leading not only to a usable prototype digital library
`but also to a set of nine principles for digital Iibraries,
`which we have tried to follow, covering issues of
`representation, architecture, and interfacing.
`
`lntrod uction
`
`Computer and information scientists should be among
`the first to experiment with digital libraries. In the spirit
`of this recommendation, the Association for Computing
`Machinery (ACM), as well as other associations and pub-
`lishers, are becoming involved in Project Envision, a re-
`search effort supported by the National Science Foundation
`to build "a user-centered database from the computer sci-
`ence literature" (Brueni et al., 1993). Starting with users of
`Project Envision at Virginia Tech and spreading to Norfolk
`State University and other groups and individuals across
`the Internet, testing will proceed regarding the applicability
`of digital library methods to Envision's scientific domain
`of computer science literature.
`A goal of Project Envision is to solve some of the
`important research problems relating to digital libraries, es-
`pecially those relating to information storage and retrievat,
`
`© 1993 John Wiley & Sons, Inc.
`
`human–computer interaction, and electronic publishing
`(Fox & Lunin, 1993). Accordingly, we have identified and
`tried to apply a set of principles that we believe should be
`the basis for future national, and later international, digital
`libraries. The next section explains these principles.
`From the proposal stages through its current prototypes,
`Envision is being created as a user-centered system, as
`specified later in Principle 8. Therefore, users are closely
`involved in the development of Envision, through a struc-
`tured interviewing process that guided decisions about
`system functionality as well as through formative usability
`evaluation. In the third section below, "Interviews with
`Users," we discuss some of the more interesting aspects
`of our task analysis (Principle 7), based on user interviews.
`The fourth section describes the innovative Envision user
`interface design that evolved from this task analysis, and the
`results of usability evaluation of our user interface design
`for both the Envision query screen and search results screen.
`In the fifth section, "Objects and Document Type Defi-
`nition Development," we consider how working with "ob-
`jects" (see Principles 2 and 9) can help improve the over-
`alt scientific communication process, and encourage reuse
`of the fruits of scholarship. This has real implications
`regarding representation (Principles 1-3), system archi-
`tecture (Principles 4-6), archiving, and use of digital li-
`brary information. Finaily, we conclude by highlighting
`some important challenges, and summarize our plans for
`future work.
`
`Principles for Digital Libraries
`
`In reviewing early work on etectronic libraries, we noted
`the influence of current practices in traditional tibraries and
`publishing operations. In particular, if we consider the spec-
`trum of representations illustrated in Figure 1, we see that
`common practice (that is, using paper-like page images as in
`Elsevier's TULIP project) may be the least useful approach
`for the next generation of digital libraries. Page images
`have all of the limitations of regular paper (problems with
`resizing, arbitrariness of "chunking" into pages, and so on),
`
`JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE. 44(8):480-491, 1993
`
`CCC 0002-8231/93/080480-12
`
`001
`
`Facebook Ex. 1006
`
`

`
`Page
`LImage
`
`PostScript
`
`WYSIWYG
`Acrobat 1ffd Processor
`
`Style
`Sheets
`
`LaTeX
`
`SGML ]
`
`Closestto paper
`
`Least reusable
`
`Most storage ace
`g p
`
`Visually riented
`Y
`
`mazk
`
`u
`
`P
`
`:<>s::>s>s>:;>::s:»>:;>:<>:<sa::::::>;::>:;:>:«:<a>:<s ::.............................,..........:...,..................................
`
`:,::<::
`?s:::::<'^::'<::i.>:::#:«<::::«:::#:::«::::<::<::^::%#:^:::::i::'i'-::?::.-::-: .::::::.:::::::::::._.^::::^.:::::^:.::::.:::._::::::::.::,^::.::
`
`Closest to computer
`
`Most reusable
`
`Least storage space
`
`:#:is:#:t`•
`
`Declarative mazkuP
`'v
`
`FIG. 1. Spectrum of document representations.
`
`can only be reused by copying, require enormous amounts
`of storage space, consume natural resources, and are marked
`up in a way that virtually excludes computer exploitation
`of a document's organization to help in searching.
`Once we go beyond our historical focus on pages, an
`enormous range of electronic products and se rvices become
`feasible. Beginning in 1988, we worked with ACM to enlist
`the involvement and creativity of many individuals and
`groups interested in electronic publishing (Fox, 1988a). A
`variety of CD-ROM, hypertext, book, videotape, and online
`database products emerged. Then the time seemed right to
`take the next step, specifically, designing and building a
`prototype electronic archive or digital library (Fox, 1990).
`Consequently, we began to reconceptualize the idea
`of digital libraries, as we envision their next generation.
`We aimed to harmonize and integrate concepts from a
`variety of interrelated fields: Artificial Intelligence (Al); dis-
`tributed systems; electronic publishing; human—computer
`interaction; hypertext; hypermedia; information storage and
`retrieval; object-oriented approaches (analysis, databases,
`design, development, programming); and open systems.
`This focus led to the following set of nine principles for
`constructing electronic archives, arranged in three areas:
`Representation, Architecture, and User Interface. An ear-
`lier explanation of these principles was presented in Fox
`(1992).
`
`Representation
`
`Principle 1: Declarative representations of documents
`should be used. Linguistics and communication theory
`teach us to be concerned both with the content and
`form of documents. Document form is represented using
`one or more "markup" schemes, and the most usable
`scheme for electronic publishing is called "declarative"
`or "descriptive" markup (Coombs et al., 1987), which is
`supported by an ISO standard, the Standard Generalized
`Markup Language of SGML (Goldfarb, 1991). This
`approach lets us model documents as a collection of ordered
`hierarchies of content objects (OHCOs) (DeRose et al.,
`1990). Thus, the guidelines developed as part of the
`Text Encoding Initiative suggest markup conventions for
`
`old manuscripts, poetry, dictionaries, and other literary
`works. Often there are multiple OHCOs, such as one for
`chapter and verse, and another for page and paragraph.
`Note that in Biblical scholarship, for example, the former,
`rather than the latter page-oriented, approach is preferred.
`Further, the many important links (see also Principle 3)
`inside or among documents can be flexibly captured for
`increased portability using the declarative ISO standard
`HyTime, which is a hypermedia standard based on
`SGML (Newcomb et al., 1991). In summary, declarative
`representations of documents are feasible, and standards
`now exist which facilitate easy interchange.
`
`Principle 2: Document components should be represented
`using natural forms, namely "objects" that can be ma-
`nipulated by users familiar with those objects. When we
`think of documents in their most general form, specifically
`as multimedia "bundles" of information, it becomes clear
`that object-oriented representations are essential. String
`matching systems like PAT view documents as substrings,
`and basic retrieval (e.g., simple Boolean or vector or prob-
`abilistic) systems concentrate on vectors of features, so it is
`infeasible to ask context-dependent questions or to inquire
`about structure as might arise in a question about inclusion
`relationships. As we move to multimedia documents, which
`are becoming more common as multimedia technologies
`are refined and multimedia systems become more avail-
`able (Fox, 1991), the weaknesses of these models become
`even more evident. In particular, multiple media must be
`synchronized or coordinated as well as interrelated. An
`ISO committee, the Multimedia Hypermedia information
`coding Experts Group (MHEG), deals with input, output,
`and interaction objects and their relationships in real-
`time multimedia systems. Describing and processing these
`documents becomes so complicated that object-oriented
`programming, where savings arise through inheritance,
`is essential. User interaction is also complicated, unless
`various document parts each can be manipulated as a
`separate object: Video or audio is played or stopped; ani-
`mations are run; spreadsheets are executed with new data;
`simulations are tried with different parameters; algorithms
`are executed or animated; and three-dimensional images
`are rotated to provide different perspectives. Mathematical
`
`JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE—September 1993 481
`
`002
`
`Facebook Ex. 1006
`
`
`
`
`

`
`objects present unique challenges and possibilities, so that
`derivation or proof objects can be analyzed step by step,
`or formula objects can be visualized in various ways
`(Wolfram, 1991). Clearly, representations using objects that
`are convenient for users will allow authors to communicate
`more directly with readers, rather than going through the
`awkward, low-level medium of paper.
`
`Principle 3: Links should be recorded, preserved, organized,
`and generalized. As we integrate documents into very
`large collections covering an entire scientific domain or
`professional area, links among those documents become
`increasingly important to help with search and browsing.
`Groupings of those links into paths, threads, tours, and webs
`are essential for organizing, personalizing, sharing, and
`preserving the structural, interpretational, and evolutionary
`connections that develop. We are beginning to see the emer-
`gence of wide area hypertext systems (Yankelovich, 1990)
`like the WorldWideWeb (WWW), that carry this concept
`forward into a distributed environment. Clearly, we must
`coordinate hypertext and hypermedia linking with the var-
`ious approaches to search and retrieval (Fox et al., 1991b).
`One approach is the idea of information graphs (including
`hypergraphs), where objects of all types are interrelated
`by links or arcs that capture not only citation (reference)
`but also inheritance, inclusion, association, synchroniza-
`tion, sequencing, and other relationships. By specializing
`object-oriented databases to this task, we are building a
`foundation for next-generation integrated retrieval systems
`(Chen, 1992). Our work with the Large object-oriented
`External Network Database (LEND) system and methods
`for querying information graphs (Betrabet et al., 1993) is
`along these lines, as are other efforts to build systems
`for managing information graphs (Giyssens et al., 1990;
`Paredaens et al., 1992). Clearly, adaptations of hypertext
`(link) and semantic network (Al) concepts are essential for
`digital Iibraries.
`
`Architecture
`
`Principle 4: There should be a separation between the
`digital library and user interfaces 10 it. To serve millions
`of users, with their diversity of backgrounds, talents, and
`needs (see "Interviews with Users"), a variety of user
`interfaces will be needed for digital libraries. With hardware
`limitations and variations, there are a host of other reasons
`for building user interlaces that are particularly suited to
`common environments. Thus, in Project Envision, we have
`development efforts underway for Macintosh (specifically,
`both 13-inch and megapixel displays), X/Motif, and NeXT -
`step user interfaces. Earlier reports (Nowell & Hix, 1992,
`1993a, b) and the discussion in the section on interface
`design below explicate these issues. With all this necessary
`tailoring of interfaces, it is clearly much easier if the system
`architecture is such that the digital library itself can be
`decoupled and developed separately. Common parlance
`
`refers to a client system running the user interface, a
`server system managing access to the digital library itself,
`and a well-defined protocol organizing the communications
`required between the two. ln the case of digital libraries
`it makes sense to begin with the Z39.50 protocol that
`was originally devised for communication between library
`catalog and bibliographic database systems. That is the
`approach taken in the popular Wide Area Information
`Server (WAIS) system (Kahle et al., 1993). We believe that
`further generalization is needed, so that information objects
`and their links can also be communicated, and we have
`been developing an Envision protocol to test that idea. In
`our case, then, we have Envision client software to manage
`the user interfaces, an Envision protocol, and the main
`(distributed) Envision system.
`
`Principle 5: Searching should make use of advanced re-
`trieval methods. At the heart of digital library systems like
`Envision, there must be support for searching, browsing,
`foliowing links, presenting selected information, and other
`services. Regarding searching, our experimental studies,
`and others recently completed in connection with the 1992
`Text REtrieval Conference (TREC), indicate that advanced
`retrieval methods can be more effective then conventional
`Boolean approaches. Our work with hundreds of thousands
`of library catalog records indicates that users prefer vec-
`tor and feedback methods to standard Boolean searching
`(Fox, 1988b; Fox et al., 1993). These approaches can be
`further extended through the use of frames (Weaver et al.,
`1989) and other representations to get closer to "concept
`searching." On the efficiency side, advances in hashing
`(Wartik et al., 1992) can improve performance in ordered
`dictionaries (Fox et al., 1991a). In many indexing, linking,
`and other situations, guaranteed direct access to large
`collections, given a desired key, can be supported by
`rapidly finding minimal perfect hash functions (Fox et al.,
`1992a, b). With all these possible benefits, future digital
`libraries should certainly be designed to use the most
`advanced retrieval methods possible.
`
`Principle 6: Open systems that include the user, and where
`(some oJ) the functions of librarians are carried out by the
`coinputer, must be developed. As digital libraries emerge,
`and become directly available to end-users, it is important
`not only to improve the user interfaces, but also to provide
`assistance to users like that offered by experienced librari-
`ans and search intermediaries. One approach is to develop
`distributed expert-based information systems, building upon
`studies of user-intermediary protocols (Belkin et al., 1987).
`Specifying the user's information need or problem, model-
`ing the user, specifying the subject domain, and manag-
`ing the overall dialog are of particular importance. Our
`COmposite Document Expert/extended/effective Retrieval
`(CODER) system was designed along these lines (Fox
`& France, 1987; Fox, 1987). Other efforts in this re-
`gard suggest that, while development is difficult and time-
`consuming, such an approach may be of value when large
`
`482 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE—September 1993
`
`003
`
`Facebook Ex. 1006
`
`

`
`numbers of users are involved. We hope that experienced in-
`termediaries will become involved in expert system projects
`to pass on their guidance to millions of end-users.
`
`User Interface
`
`Principle 7: Task-oriented access to electronic archives must
`be supported. Current efforts to build prototype digital
`libraries are often focused on a particular subject domain,
`in part because of support provided by associations or
`publishers. Thus, the CORE project (involving the Amer-
`ican Chemical Society and Chemical Abstracts Service,
`as well as Bellcore, Cornell, and OCLC) deals with the
`chemical literature (Lesk, 1991). Part of the hope of that
`project is to have access to the chemical literature be a
`key feature of a "chemist's workstation." Supporting the
`research, referencing, writing, and educational activities of
`staff in a university chemistry department can be viewed as
`providing task-oriented access to information suitable for
`each of those types of activities. We believe that in addition
`to having user interfaces that support information access as
`a separate activity, with its aspects of searching, browsing,
`previewing, and so on, "embedded information access"
`must be enabled. For example, a chemist preparing a class
`or conference presentation should be able to escape from a
`tool like PowerPoint TM , find a description of an important
`reaction, grab the registry number and structure diagram
`for one slide, extract a table showing yield for another
`slide, and return directly to the expanded presentation.
`Similarly, a programmer accessing the Envision archive
`should be able to interrupt a programming effort to find a
`useful algorithm from Collected Algorithms, and add it as a
`subroutine, along with capturing some of its documentation
`and pointers to more information. We hope that efforts of
`this type will proceed in similar fashion to how computers
`in cars, microwave ovens, and compact disc players now
`support rather than interfere with users' tasks.
`
`Principle 8: A user-centered development approach should
`be adopted. Since workstations are often devoted to in-
`dividual users, we must make them serve those users.
`We should turn our system development efforts around
`to be centered on the users, rather than on the machine.
`Without this focus on the user, we may well produce digital
`libraries (and other interactive systems) that can compute
`perfectly and quickly, but cannot communicate effectively
`and efficiently with their users. As we learn more about
`design and development of interfaces (Hix & Hartson,
`1993), a user-centered approach becomes more feasible.
`The next two sections explain our efforts in user-centered
`design of the Envision system.
`
`Principle 9: Users should work with objects at the right
`level of generality. If we follow Principle 2, our digital
`libraries will represent information in terms of usable
`objects. With advanced search methods such as those called
`
`for in Principle 5, we can search, browse, and preview those
`objects. Further manipulation should be supportive of user
`tasks, as called for in Principle 7. We consider all of these
`issues further in the section on objects and document type
`definition development.
`The following sections discuss many of these principles
`further, focusing on users, user interfaces, and objects. For
`more general information on Project Envision, the reader is
`referred to Brueni et al. (1993).
`
`Interviews with Users
`
`In accordance with Principle 8, we began by focusing
`on potential users of a digital library of computer science
`literature, such as Envision. Over a four-month period
`we interviewed 12 professionals in the areas of computer
`science and information retrieval. Interviewees were chosen
`carefully to broadly represent the type of user we expect
`for Envision. During intensive interviews lasting from one
`to two hours, interviewees responded to questions focused
`on four topics:
`
`(1) Current information retrieval practices.
`(2) Current information dissemination practices.
`(3) Desired information retrieval and manipulation capa-
`bilities.
`(4) Demographic data.
`
`When seeking publications relevant to a particular topic,
`most of our interviewees have used electronic information
`systems of some kind. These include computerized library
`catalogs, CD-ROM systems, and online search services.
`However, our interviewees found existing systems difficult
`to use for a variety of reasons. Inadequate access to any
`electronic information system is one major problem. Indeed,
`the feature most requested by interviewees for a new
`information retrieval system is access from the workstation
`in their own offices.
`Interviewees also complained about the difficulty of
`structuring queries, the number of diverse user interfaces,
`inadequacy of feedback about unsuccessful searches, and
`the amount of knowledge required before systems are really
`usable. Our interviewees generally disliked any requirement
`or need to consult a human intermediary, or search system
`expert, to access the literature.
`Most interviewees specifically requested or implied the
`need for full text retrieval. Other features commonly re-
`quested include:
`
`• Access to multiple forms of information (abstract, re-
`sume, brief description, full text, bibliographic entry)
`about each document retrieved;
`• Print capability;
`• User annotation facilities; and
`• Ability to establish and work within a personal subset of
`the database.
`
`A usable interface was mentioned often as a needed feature,
`and complaints about the user interfaces of existing elec-
`
`JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE—September 1993 483
`
`004
`
`Facebook Ex. 1006
`
`

`
`tronic information retrieval systems were frequently cited
`reasons for not using those systems.
`Our interviewees want the ability to explore patterns in
`the literature. One spoke at length about the "community
`of discourse," or invisible college of people carrying on
`conversations in print, all reading what the others have
`written. Others spoke of citation indexes, reference tools
`that reveal patterns of citation within the literature, so
`that works evolving from major articles may be identified.
`Ability to locate seminal documents, those which have been
`widely cited, is needed. Interconnections in the literature
`are of widespread interest. People want to use hypermedia
`linking to navigate among documents with common pat-
`terns of citation and to follow chains of reference among
`documents. In essence, they want to be able to follow
`on-going "conversations" in the literature.
`Browsing was another common theme. Users want to
`be able to explore the literature along dimensions of their
`choosing, to home in on particular areas of interest and
`explore those in detail, then move on to broader views, or
`sometimes different views. For some, browsing includes
`the ability to examine the structure of documents, not
`just the citation or the abstract. Users want to identify a
`section of interest in a document and zoom in on it for
`closer examination and more details. Access to tables of
`contents provides part of this capability, but users want
`to move seamlessly between the table of contents and the
`body of a document. They want to see structure at a finer
`granularity than a table of contents allows. Capability to
`search document structures is wanted, so that chapters,
`illustrations, graphs, or sections of code might be located,
`not just whole documents by title, subject, or author.
`For some, browsing is a luxury rarely permitted by
`pressures of time. These users want the ability to locate
`a few critical items of interest and be protected from the
`rest. They are especially interested in powerful filters to
`eliminate "junk" and allow them to easily locate only the
`most highly relevant materials. Offered the possibility of a
`system regularly scanning the literature for them and notify-
`ing them of new publications of probable interest, they were
`fearful of being overwhelmed. Information overload was
`cited as a reason for avoiding Internet discussion groups
`and bulletin boards.
`Interviewees shared reliance on journals and conference
`attendance as major sources of information, with additional
`attention to conference proceedings. Talk with colleagues
`was ranked as equal in importance to journals as a source.
`Colleagues are especially helpful in providing pointers
`into the literature, that is, specific references to works
`likely to be helpful in solving a particular problem or to
`be of particular interest. One interviewee indicated that
`colleagues serve as valuable filters; they point to the few
`best works in an area without providing an exhaustive list
`of less valuable materials. A few interviewees make use of
`network bulletin board services, but most do not.
`Interviewees indicate that they rarely use videos, because
`of the inability to browse or skim video, which is seen in-
`
`stead as an "all-or-nothing" experience. Users are frustrated
`by the difficulty of locating particular segments of video
`that are of special interest. They would like to see a "video
`table of contents" and to be able to create hyperlinks to
`and from specific video frames.
`We asked our interviewees about objects of interest in
`the computer science literature. They spoke of the obvious
`entities: books, journals, articles, videos, bibliographies,
`even figures and tables. People are also objects of in-
`terest, as authors, as researchers, as colleagues. Research
`projects, funding sponsors, conferences and workshops,
`and various types of institutions are also objects of in-
`terest. Additionally, programs, data structures, algorithms,
`animations, programming languages, hardware devices, in-
`teresting problems, and concepts are entities the users wish
`to manipulate. Users want access to source code, ideally
`in a choice of languages. They want to be able to embed
`the code in their own programs for testing and use with
`their own data, without rekeying the code. They would like
`access to analytical data about algorithms, to explanations
`by experts, and to animations that increase comprehension.
`
`Design and Evaluation of the
`Envision User Interface
`
`Responding to interviewees' concern that an information
`retrieval system must be accessible from their offices, our
`design is based on the premise that the Envision user
`interface will run as a client process on a user's desktop
`computer, communicating with the Envision retrieval sys-
`tem via network. Our user interface designs provide flexible
`use of varying configurations of monitors, both in size and
`number of displays. The lowest configuration supported
`uses a single 13-inch gray-scale display. With larger or
`more monitors, tiling of windows becomes feasible, and
`it is easier to work with full-text or page-image retrieval.
`Our interface specification calls for separate windows or
`groups of windows for each of the major phrases or types
`of interaction with the Envision system. These include:
`
`• Query Window (with four query fields and a query
`history);
`• Scarch Results Windows (Graphic View, Item Summary,
`Item Preview); and
`• Browsers.
`
`The next two subjections deal with the Query and Search
`Results Windows (see Fig. 2), respectively. Work on the
`Browsers will be reported in a later publication.
`
`Envision Query Window Design
`
`The Envision Query Window design gives users the
`benefits of natural language query formulation (i.e., no
`complex syntax or use of logical operators is required,
`nor is knowledge of an artificial indexing language), while
`also providing the means to restrict searches. The Query
`Window has two categories of use:
`
`484 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE—September 1993
`
`005
`
`Facebook Ex. 1006
`
`

`
`Query
`Wjndow
`
`Queiy Fields
`
`Qucry Options
`
`Query History
`
`Interface Design
`
`Search
`Windows
`
`Graphic View
`Window
`
`Item Summary
`Window
`
`Preview Item
`Window
`
`2-D Piot
`
`Tabular List
`
`Text Field
`
`Layout Options
`
`Layout Options
`
`Hypertext Buttons
`
`Layout Options
`
`FIG. 2. Main components of Envision's user interface design.
`
`New queries are created and searches performed from
`this window.
`Access to previously completed (old) queries and the
`results of the related searches are provided. OId queries
`may simply be viewed or they may be revised and used
`for another search. Results of searches from old queries
`may also be redisplayed via a query history feature.
`
`The Query Window offers a user three ways to create new
`queries:
`
`By entering document descriptors in four new query
`fields for authors, title words, words related to content,
`and words found in other parts of the document as
`specified by a pop-up menu labeled "Special Query."
`By editing earlier queries.
`By combining results of previously completed searches,
`using set operations.
`
`Query Fields. The Query Window, shown in Figure 3,
`features four query fields for Authors, Words in Title,
`Content Words, and Specia! Query, plus a Query History
`field. The Special Query field has a pop-up menu control
`that aliows users to specify searches of other document
`parts—abstracts, chapter titles, figure headings, and tabies
`of contents, as well as to enter a complete bibliographic
`citation as a single biock.
`When creating a new query or editing an old one, the
`user may make changes in addition to or instead of simply
`editing the text in the four fields. Other options include
`changing the matching types (exp!ained further below) used
`for each field, changing the relationship among fields, and
`changing filters that restrict search results. The filter con-
`trols—for publication year, publication language, number
`of items to be found, and type(s) of items desired—are at
`the bottom of the window.
`As shown in Figure 3, matching type options are given
`to the right of the query fields, with text and related radio
`
`buttons. Users have control over whether terms within
`a given field are "ANDed" ("Match all .....) or "ORed"
`("Match any .....). In some cases, as for Word(s) in Title,
`and user may specify that the order of terms in the query
`must be matched as well. The relationship among the four
`query fields is also user-controlled via radio buttons below
`the group of fields.
`
`Query History. As queries are stored or related searches
`are performed, the user establishes a history that is acces-
`sible through the Query History field across the top of the
`window, shown in Figure 3. In the Query History, a one-
`line summary form of each query is displayed in order by
`query number, along with the number of items retrieved by
`the related search. The Query History provides access (0 the
`results of previous searches, means to redisplay the full con-
`tent of previous queries for possible revision, and a mech-
`anism for combining the results of completed searches.
`
`Formative Usability Evaluation of the Query Window De-
`sign. Prior to building an interactive rapid prototype,
`the Envision Query Window design was modified several
`times as a result of critique sessions with prospective
`users, using paper versions of the design. Sessions with the
`Human—Computer Interaction Research Group at Virginia
`Tech were particularly productive. An Aldus SuperCardTM
`prototype was then created on a Macintosh and used for
`the formative usability evaluations described below.
`The foremost goal in our usability evaluation of the
`Query Window was proof of concept: We needed to verify
`that users could understand how to formulate a complicated
`query using this window and a!so how to formulate revised
`queries based on previous queries. Usability evaluation
`was conducted with four participants (a reference librarian
`and a Computer Science Department undergraduate student,
`graduate student, and faculty member).
`
`JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE—September 1993 485
`
`006
`
`Facebook Ex. 1006
`
`

`
`Query History:
`1
`20 A:Wickens,Christopher D.
`50 C: signal detection theory
`2
`3
`12 land2
`
`Usa Help to Iearn to combine results of completad searches.
`
`aaui
`
`[New Query]
`__________
`[ Do Searchl
`
`Author(s) :
`Enterfamilynan,e, comma, then
`m:nderr;?fnam:Entr one
`[
`authors with semi-colons. Enter 1
`coiporateor agencyauthor
`namesas usually pnnted.
`
`Words in Title
`Enter con)plete tjtle or
`words from tit'e.
`
`[
`
`Wfl
`
`Query #4
`xampIe: Smith, John J. Jr.; Jones, A. L. ABC Computer: ................................................................................
`Matchfull name(s)
`Match any author(s)
`0 Match tamily name(s)
`1TOFinci closest match(es) O Match all authors
`..................................................................................
`IA1 ôj Match exactly as entered
`O Match all wo