bulletin of the american society for information science and technology volume 33 issue 1 2006 [doi...

Upload: jijoym

Post on 08-Aug-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/22/2019 Bulletin of the American Society for Information Science and Technology Volume 33 Issue 1 2006 [Doi 10.1002%2F

    1/3

    by Jens-ErikMaihat context, use and actors are important components forT analysis in the design of almost anything in the informa-tion field is a trivial argument. Controlled vocabularies (CV)are no exception, and it has often been argued in both schol-arly and professional literature that design of CVs needs tobe based on solid understandings of context and actors. Thechallenge is to decide which factors need to be consideredand how to outline an approach for including analyses of con-text, use and actors. In this paper I will show how the cogni-tive work analysis (CWA) framework offers one possibleapproach to the design of controlled vocabularies.

    Guidel inesA controlled vocabulary can be defined as a list of termsthat have been enumerated explicitly(ANSVNISO,2005,p. 5)for the purpose of organizing and representing informationto facilitate information retrieval. Controlled vocabulariesvary in complexity from simple alphabetic lists of terms toclassification schemes and taxonomies that show semanticrelationships, and finally, to complex thesauri that also showassociative relationships between terms. The steps that adesigner of controlled vocabularies can take have been welldescribed in the literature. These steps are often representedas some version of the following:1. Analyze literature, needs, actors, tasks, domains, activ-ities, etc.2. Collect, sort and merge terms3. Select descriptors and establish relationships4. Construct the classified schedules5. Prepare the final productThe latter steps - 2 through 5 - are well-prescribed andworked out in great detail in several standards, well-estab-

    Jens-Erik Mai is associa te dean and as sociate professo r in theFaculty of Information Studies, Universi ty of Toronto. He can bereached at jens-erik.maifis.utoronto.ca

    lished textbooks and best practices. These steps deal withtechnicalaspects of the design and construction of controlledvocabularies, including guidelines and rules-of-thumb forhow, for instance, to determine the appropriate form of theterms, clarify the meaning of terms, factor compound termsordetermine the relationship between terms. While these stepsare important aspects and techniques that must be masteredby developers of controlled vocabularies, design decisionsthroughout these steps must be guided by the outcome of thefirst step. However, the first step- analysis of literature, needs,actors, tasks, domains, activities and similar aspects - hasbeen somewhat neglected in the literature.The advice given for the first step is often limited to eithersimply mentioning that the designer needs knowledge aboutthe context of the controlled vocabularyor suggesting that alist of potential terms is to be drawn up by subject experts orto be selectedor extracted from the content objects.Cons tra in ts

    To facilitate access that is as transparent and convenientas possible, the designers selection of terms to be includedin a CV and their determination of the relationships amongthem must be informed by the actors usage of the informa-tion. There are many factors that potentially could influencethe actors use of information and their information seekingstrategies and choices. Each of these factors might influencethe actors perception and understanding of the information.However, it is almost impossible to account for such individ-uality in the design of CVs that are used by many people.How are we to select them?The CWA framework provides an answer by focusing onthe constraints that shape the actors behavior and therebylimiting the number of possible variables that need to be con-sidered. Constraints are factors external to individuals butcommon to all individuals within the same contextor domain.The goal is to identify the constraints that actually shape indi-viduals information-seeking behavior and not just their spe-cific preferences, perceptions and experiences. The constraints

    OctoberlNovember ZOObBulletin of the American Society for Information Science an d Technology 17

  • 8/22/2019 Bulletin of the American Society for Information Science and Technology Volume 33 Issue 1 2006 [Doi 10.1002%2F

    2/3

    I Special Section I

    that shape the behavior of actors in particular situations arethe parts of the context that limit and enab le the actors to per-form their work. It is important to recognize this duality ofconstraints; constraints limit and enable actions at the sametime. For instance, a scholarly domains history, schools ofthought and paradigms both limit and enable actors in the par-ticular domain - he constraints thereby shape possible infor-mation needs. A dom ains history, for instance, enables acto rsto formulate questions and inquiries about the particular phe-nomena that the dom ain studies by providing a narrative ofthe evolution of the knowledge about the phenomena.Simultaneously, the domains history limits the kinds of ques-tions and inquiries actors can pose about the phenomena byproviding the current, consensual understanding of it. Theexact nature and types of constraints vary from domain todomain and can be uncovered using CWA.Understanding the behavior-shaping constraints givesdesigners insight into the context of actors work and providesan understanding that facilitates systems design. The outc omeis not a prescription of wha t actors should do ( a normativeapproach) or a detailed description of w hat they actually do (adescriptive approach), but an analysis of the constraints thatshape the domain and context.

    D imens ionsRather than enumerating the factors that influence actors,CWA defines a set of dimensions (Figure 1) that holds con-straints, as explained in the introduction to this Bulletin section.The issues to be addressed at each dimension vary from domainto domain, dependin g on the type of domain, goals of the CV

    and the activities that the system supports. Below, I define anddiscuss each dimension and demonstrate how it contributes tothe analysis of information behavior to support design of CVs.

    FIGURE 1. Dimens ions of Cogni t i ve Work Analys is

    1) Environment.Actors in a given scholarly domain, for instance,are constrained by such aspects as he domains discourse, his-tory, schools of thought, paradigms, research fronts and activi-ties. These constraints limit and enable the types of informationneeds ac tors have in the particular scholarly domain. Likewise,a comm ercialR&D division that is engaged in the developmentof Web search engines is constrained by the previous research andcurrent state of know ledge in information retrieval. This con-text shapes the kinds of qu estions that are asked and addressedby the R&D team and the creation of a CV for their intranet isinfluenced by the tradition of research in information retrieval.2 ) Work domain a nalysis. The work domain provides theframework in which the actors op erate and actors generatetheir information needs in this context. For instance, actorsin inform ation retrieval research are constra ined by the goals,priorities, functions, processes and reso urces of their particu-lar work dom ain. Wh ile researchers in the domain share someof the constraints in the environment, their particular workdomain prese nts other constraints that are unique to the workdoma in. Actors in a comm ercial R&D division work underconstraints that are significantly different from a ctors in a uni-versity setting. While actors in these two work dom ains maywork on the same problem, the fact that they operate in differ-ent work domains -with differe nt goals, priorities, functions,processes and resources - will cause them to approach theproblem d ifferently. This differen ce affects their informationneeds and how they search for information, which shoulddetermine how the information is to be indexed.3 ) Organizational an alysis.Workplaces are analyzed in termsof their orga nizational structures , management style s, organiza-tional culture, nature of the organ ization and allocation of roles.The organ izational analysis gives the designer an understand-ing of how the domain is structured both e xplicitly and implic-itly. While actors in research wo rkplaces might have a highdegree of autonom y in their work and their information needsmight therefo re develop relatively independe ntly of the orga-nizational structure, actors in more structured organizations,like an insu rance company, develop their information needs inaccorda nce with their particula r tasks. Actors in such organi-zations are often assigned particular tasks, and they developinforma tion needs to react to these a ssigned tasks. We there-fore need an understandingof the organization to gain an insightinto how work is delegated, assigned or otherwise acquired.4) Activity an alysis.Activity analysis examines what users doto achieve their tasks. The CWA framew ork divides the analy-sis of activities into three sepa rate parts:

    4a ) Activity analysis in work domain terms: Actors are con-strained not only by the environment, work domain andorganizational structure, but also by their activities. Theactivity analysis in work domain terms teases out the natureof the actors tasks to understand how, where and when theyneed information. Actors in an insurance company, forinstance, might want information such as claims, policereports or photos of damaged material for their work.

    18 Bul let in of th e American Society for in for matio n Science and Technology-OOaoberlNovernber 2006

  • 8/22/2019 Bulletin of the American Society for Information Science and Technology Volume 33 Issue 1 2006 [Doi 10.1002%2F

    3/3

    However, these needs develop in response to specific activ-ities the actors perform. We could ask whether the docu-ments are needed to address a specific issue in a class actionsuit or in response to a retention schedule. Designers of CVsneed to understand actors work activities to understand thedifference between these two types of information needsand to make decisions about the organization and represen-tation of the material. Likewise, when faculty members atuniversities search for information in relation to their schol-arly activities, they may need information for their classes,their research or their service activities. We could askwhether a scholar is interested in a document in prepara-tion for a class presentation or to confirm specific ideaswhen reviewing a colleagues manuscript. These activitiesconstrain the type of information they are interested in andthe type of information system they will use. Without anunderstanding of these activities and constraints, designerswould not know how to design useful indexing systems.4b) Activity analysis in decision-making terms: The pur-pose of this analysis is to clarify what information usersneed to make decisions, what information is actually avail-able and what information is desirable but not available.Researchers in a commercial R&D division might needinformation about a specific functionality in a search engine,and they might be able to fmd this information, for instance,in their personal files, on their intranet or in public digitallibraries. Their search for this information is constrainedby the decision they have to make. Thus, depending onwhether they are exploring issues related to functionalityor are searching for design requirements of a search engine,they will need merent typesof information.This differencein types of information needs could influence the designof the indexing systems for this work place.4c)Activity analysis in t e r n of strategies that can be used:The search strategies employed by actors in current sys-tems can be good indicators of preferences in their search sit-uations and might be valuable to understand as backgroundfor the formulation of search strategies in future systems.However, actors current search behavior might not be rel-evant in future information systems, and the analysis ofactors activities in terms of strategies should focus onpos-sibilities for searching and not be limited to descriptions ofcurrent practice. The analysis of strategies should thereforeask questions about possible strategies that actors can take,independently of whether actors actually use those strate-gies today. To idenhfy possible strategies, he analysis wouldexamine which strategies an actor could use to find specificinformation in an effective way. For instance could the actorsearch by using index terms, browsing the system or goingdirectly to sources that are known to himher?

    5 )Analysis of actors esources and values. The purpose of thisanalysis is to gain insight into the actors cognitive resourcesand values, such as their knowledge of the subject matter dealtwith in the domain, their preferences for information sources

    For Further ReadingAitchison, J ., Gilchrist, A., & Bawden,D. (2000). Thesaurus con-struction and use: A practical manual (4th ed.) Chicago: FitzroyDearborn.American National Standards Cornmittee/National InformationStandards Organization (ANSI/NISO). (2005.) Guidelines for theconstruction, format, and management of monolingual con-trolled vocabularies:239.19-2005. Bethesda, MD: NlSO P ress.Rasmussen, J ., Pejtersen, A.M., & Goodstein, L.P (1994).Cognitivesystems engineering. New York: Wiley.Rosenfeld, L., & Morville, I? (2002.) Information architecture forthe World Wide Web ( 2n d ed.) Sebastopol, CA: OReilly.Vicente, K. (1999). Cognitive Work Analysis: Toward safe, produc-tive, and healthy computer-based work. Mahwah, NJ : LawrenceErlbaurn Associates.

    and format of information and values in terms of objectivityvs. subjectivity in representation of information. For instance,while designers of systems for actors in scholarly domain mightexpect a certain level of subject knowledge, the informationsources used in scholarly domains might vary among differ-ent user groups. An analysis might find that senior researchersin the domain prefer short conference papers while studentsprefer review articles and monographs. Such a finding shouldhave an impact on the design of the indexing systems. Likewise,such an analysis might reveal that researchers in a commer-cial R&D division prefer more recent information in digitalformats that contains lots of graphics representations.By moving the focus from descriptions of what actors doto an analysis of the constraints under which actors operate,studies of human-information interaction can become usefulfor design. It is more useful because design of CVs cannot bebased on knowledge about the behavior of individuals; designof CVs is better served with analyses of the constraints underwhich actors operate. These constraints remain relatively sta-ble over time and among different actors and therefore serveas better guides for potential information needs.Co n c lu s io n

    CWA provides a powerful framework for analyzing infor-mation behavior for the purpose of designing controlled vocab-ularies. While factors that can affect human-information inter-action are almost unlimited, the CWA framework offers anumber of dimensions along which one can identify variousconstraints that influence actors information needs.Each dimension contributes to the designers understand-ing of the domain, the work and activities in the domain andthe actors resources and values. The analyses ensure thatdesigners bring the relevant attributes, factors and variablesto design work. While analysis of each dimension does notdirectly result in design recommendations, these analyses ruleout many design alternatives and offer a basis from whichdesigners can create systems for particular domains. To com-plete the design, designers need expertise in the advantagesand disadvantages of different types of indexing languages,the construction and evaluation of indexing languages andapproaches to and methods of subject indexing.

    OctoberlNovernber ZOOb Bu l le t i n o f th e Amer i can Soc iety fo r In fo rmat ion Sc ience and Techno logy 19