Upload
suki-reid
View
27
Download
0
Embed Size (px)
DESCRIPTION
Information Architecture for the World Wide Web. Thunder Lizard’s Web Design World ‘99 Seattle, Washington July 21, 1999 Louis Rosenfeld Argus Associates, Inc. Introduction Who am I?. Brief Bio - PowerPoint PPT Presentation
Citation preview
1
Information Architecturefor the World Wide Web
Thunder Lizard’sWeb Design World ‘99Seattle, WashingtonJuly 21, 1999
Louis RosenfeldArgus Associates, Inc.
2
Introduction
Who am I?
Brief Bio• President of Argus Associates, an
information architecture consulting firm (Ann Arbor, Michigan).
• Bias: Librarianship and Information Science (LIS) background.
• Bias: Work on larger, heterogeneous corporate sites (primarily intranets) for Fortune 500 companies.
• Columnist for Web Review, Internet World magazines.
• Co-author of Information Architecture for the World Wide Web (O’Reilly & Associates, 1998) and others.
3
When do you know you have problems?
When you hear these questions.
Users: “Why can’t I find what I’m looking for?”
Content Owners: “Where should my new content go? And what should I do about all this ROT (Redundant, Outdated, and Trivial content)?”
CIO: “Where’s my ROI? I want my ROI!!”VP: “How come that other VP’s content is
more prominent than mine?”The Web Team: “Who’s in charge here
anyway?”You: “Why can’t I find what I’m looking
for?”
4
“umbrella” site
Common Problem #1
Site development is “organic”, disorganized.
• Content lives in subsites or “silos” that are locally maintained and often reflect the “org chart”.
• The organization and users want an “umbrella”, a common interface to all content.
• Individual subsites are poorly architected and have few or no policies or procedures to deal with maintenance issues.
subsites
5
Common Problem #2
Site structure is an abstract concept.
New medium means that initial focus is on the tangible and sexy; attention has been diverted by the lure of:
• Aesthetically charged visual design.• Hi-octane functionality.• Lucid text (and as much as possible).• Other Cool Stuff.Analogous to building a house without a
blueprint.Subsequent problems are all too apparent
in later generation sites.
6
Common Problem #3
“Information Retrieval” is a foreign phrase.
Information retrieval performance is reduced when addressing the diversity typical of most sites:• Content (formats, types, subject domains).• Users.• Missions/goals/constraints.
Information retrieval is already difficult in narrower contexts.• Rarely one right answer (relevance is
subjective).• Based on language, which is inherently
ambiguous. Example: homographs and synonyms of pitch.
7
What is Information Architecture?
No single definition is perfect...
“Information architects organize content and design navigation systems to help users find the information they need.”
In the context of the Web:• Organize means to group and label content
at the macro (e.g., collections, areas) and micro (e.g., pages, fields) levels.
• Navigation refers to the default organization of the site, the design of page components, and tools such as search engines, indexes, and site maps.
Many definitions; mine clearly biased by librarianship/information science background.
8
Why Is Information Architecture Important?
User’s perspective.
Inability to find information is a major complaint.
Information needs vary (known item, exploratory, comprehensive research).
Preferences vary (searching, browsing; precision, recall).
Expertise varies (query languages, technology literacy).
Scary Fact #1: According to Zona Research, 20% of Internet savvy users have given up at least 3 times while shopping on the Web.
9
Why Is Information Architecture Important?
Producer’s perspective.
Cost of finding information.Cost of not finding information.Maintenance costs.Political costs.Scary Fact #2: According to Jakob
Nielsen and many others, a poor navigation system in a large corporate intranet can cost the company millions in lost employee productivity.
10
Introduction to Information Architecture
Where it fits.
In the context of site development:• Often leads the
discovery/recommendations phase.• Highly collaborative during conceptual
design phase.• Minimal involvement in
production/implementation phase.
11
Introduction to Information Architecture
What the deliverables are.
Blueprints (from top level to “chunk” level).
Major page mockups/templates.Navigation systems.Labeling systems/controlled
vocabularies/thesauri.Policies and procedures.Production work (e.g., classification and
indexing).Training (e.g., educating an indexing
operation).
12
Introduction to Information Architecture
Top-down vs. bottom-up varieties.
“Top-down” Information Architecture• Tie together disparate pockets of content
for improved searching and browsing.• Highly focused on users and information
needs.“Bottom-up” Information Architecture
• Improve searching and browsing within a single, high-volume pocket of content.
• Highly focused on content, content attributes.
Each approach informs the other (no mutual exclusivity).
13
Introduction to Information Architecture
Top-down vs. bottom-up varieties.
Top-down example: Create a common information interface (“umbrella site”) for a corporate intranet with dozens of separate sub-sites.
Bottom-up example: Re-architect a large collection of technical reports.“umbrella”
site
local subsitesBottom-Up Approach
Top-Down Approach
14
Top-Down Components: Organization Systems
Definitions.
Organization Structures• The shape of the information space.• The types of relationships between
content areas or items.• e.g. hierarchies, databases,
hypertext.Organization Schemes
• Pathways for intellectual access.• e.g. by author, by topic, by audience.
15
Top-Down Components: Organization StructuresHow should a site’s content be structured?
Types of Organization Structures• Hierarchies: useful for the top levels of a site.• Databases: organize large bodies of
homogeneous content• Hypertext: complement other structural types.• Hybrids: often make most sense within a site.
hierarchy
database
hypertext
16
Top-Down Components: Organization SchemesHow should my site’s content be organized?
Exact Organization Schemes.• By name, alphabetically (e.g., white
pages).• By geography (e.g., atlas).• By chronology (e.g., timeline).
Characteristics.• Neat and easy to maintain.• Everything has a place (one right
answer).• Extremely useful for users who know
exactly what they’re looking for.
17
Top-Down Components: Organization SchemesHow should my site’s content be organized?
Ambiguous organization schemes.• By topic (e.g., bookstore, yellow pages).• By task (e.g., buy, find, contact).• By audience (e.g., home, small
business, government).Characteristics.
• Messy and full of overlap.• Hard to implement and maintain.• Extremely useful for users who don’t
know exactly what they’re looking for (subject searching, associative learning).
18
Top-Down Components: Labeling Systems
The basics.
Symbols that represent concepts.Types:
• Labels within navigation systems.• Titles and headings.• Links.• Index terms (keywords).• Icons (visual representations of
information).Strive for systems of labels which are:
• Specific and clear (for intended audiences)
• Predictable• Consistent
19
Top-Down Components: Labeling Systems
Where should I get my labels?
Existing Site/Other Content• Don’t throw out the baby with the bath
water.Other Sites
• Check out the competition.Controlled Vocabularies (CVs) and Thesauri
• Standardized sets of terms which describe a specific domain (thesauri contain CVs, relationships between terms (e.g., broader, narrower, see also), and scope notes).
Users and Subject Experts• Focus groups, query analysis, user testing.
20
Top-Down Components: Navigation Systems
Types of navigation systems.Global (site-wide) navigation systems:
rule-based.Local (sub-site) navigation systems: rule-
based.Contextual navigation systems: hand-
crafted.Supplementary navigation systems.
• Tables of contents/site maps.• Site indexes.• Guides and guided tours.
21
Top-Down Components: Navigation Systems
What makes a navigation system succeed?Navigation systems need to:
• Provide context. (Where am I?)• Provide flexibility (Where can I go?)• Provide guidance (How can I get
there? And get back to here?)• Make sense (Separate global and
local systems)• Avoid competing with content.
Test your site by seeing if users can answer these questions for random pages.
22
Top-Down Components: Navigation Systems
What supplementary navigation type is best?Table of Contents/Site Map
• Reflects site’s organization system (mental model).
• Good for subject searching.Site Index
• Flattens organization system (greater granularity).
• Supports known-item searching.• May provide multiple browsable indexes.
Guide• Highlights a few of the site’s resources for a
specific audience, topic, or task.• Good for introducing users to an aspect of
the site’s content.
23
Top-Down Components: Searching Systems
Searching really sucks...
“Using an on-site search engine actually reduced the chances of success.” (1998 Usability Study by User Interface Engineering)
http://world.std.com/~uieweb/searchart.htm
Percent of Successful Tasks
53%
30%
0%
10%
20%
30%
40%
50%
60%
Without Search With Search
24
Top-Down Components: Searching Systems
…but users demand it.
“Search is one of the most important user interface elements in any large web site...Our usability studies show that more than half of all users are search-dominant.” (Jakob Nielsen, Alertbox, 1997)
http://www.useit.com/alertbox/9707b.html
User Behavior (rough figures)
0%
10%
20%
30%
40%
50%
60%
Search Dominant Link Dominant Mixed Behavior
25
QueryQuery
Top-Down Components: Searching Systems
Finding involves more than searching...
BrowseBrowse
SearchSearch
AskAsk
SearchSearch
AskAsk
BrowseBrowse
BrowseBrowse
SearchSearch
AskAsk
BrowseBrowse
BrowseBrowse
BrowseBrowse
BrowseBrowse
SearchSearch
SearchSearch
BrowseBrowse
SearchSearch
BrowseBrowse
BrowseBrowse SearchSearch
AskAsk
BrowseBrowse
SearchSearch
BrowseBrowse
SearchSearch
SearchSearch
BrowseBrowse
AskAskAskAsk
BrowseBrowse
BrowseBrowse
SearchSearch
26
QueryQuery SearchInterface
SearchInterface
Top-Down Components: Searching Systems
…and searching involves more than an engine.
SearchEngine
SearchEngine ContentContent ResultsResults
QueryLanguage
QueryLanguage
QueryBuilders
QueryBuilders
27
Top-Down Components: Searching Systems
How can searching be improved?
Utilize multiple search interfaces.• Expert vs. simple. Distinguish by user
background/discipline/expertise.• Support common information needs
(known item vs. exploratory vs. research).Utilize search zones and leverage document
structure.Support iterative, integrated searching and
browsing (including a “no-dead ends” policy).
Explain what is being searched and how it can be searched.
Avoid default engine configurations, especially default relevance rankings.
28
Top-Down Illustrated
Typical scenario.
“umbrella” site
local subsites
accessby taskaccess
by topicaccess
by audience S
S
SToC
ToC
Ind
29
Top-Down Case Study
Background and challenges.
Began with unclear goals for site.Tension between central administration
and subsidiaries (political divisions, “rogue” Web sites, look and feel issues).
No in-house process for developing a large, distributed, cross-departmental information system.
Many different purposes for site.Multiple audiences for site.
30
Top-Down Case Study
Solutions.
Created umbrella architecture that draws appropriate borders between centrally-maintained site and autonomous departmental sub-sites.
Multiple ways of navigating content.• Search.• Browse by topic, task, audience,
political unit.• Table of contents, site index.
Clear divisions between central and autonomous content embodied in the architecture and related procedures and policies for maintaining content.
31
Bottom-Up Components: Content Analysis
Look for order in the chaos of your content.
Analogous to user surveys.Requires review (generally iterative) of content
samples.Looking for:
• Logical patterns within messy content (e.g., press releases vs. product descriptions).
• Ways to group content (leads to “bottom-up” organization systems design).
• Meta-information opportunities (e.g., existing meta-information, approaches and sources for adding new meta-information).
• Relationships between document types (e.g., parent-child, related, sequential).
32
Bottom-Up Components: Content Modeling and MappingGetting control of content.
Concentrate on meaning and value rather than physical formats (e.g., MS Word vs. Lotus Notes).
Develop document types (logical).• Use grouping exercises.• Enlist content specialist and expert user
input.Develop templates (physical).
• Delineate required vs. optional content components.
• Determine prominence, grouping, and sequence of components.
• Strive for consistency in presentation across as well as within templates (usability gains).
33
Bottom-Up Components: Content Modeling and MappingMatching the model with reality.
ROT (Redundant, Outdated, and Trivial content) removal:• Policy based.• Impacts all aspects of content creation,
maintenance, deletion.Identify missing content.Determine granularity for content
“chunking”:• Break up longer mixed-concept content.• Join together fragments.• Consider contexts: usability, display,
retrieval, reuse, authoring.
34
Bottom-Up Components: Meta-Information
Information about information.
If stored as a record, meta-information constitutes a document surrogate; if stored within document, acts as a representation or labeling of document.
Provides context (e.g., date, publisher).Facilitates retrieval (e.g., author, title, subject
index).• Serves as alternative to major organization
scheme.• An effective alternative to full-text searching.• Can be leveraged for creation of browsable
indices/menus.Consider value of controlled vocabularies for
descriptive indices.
35
Bottom-Up Components: Meta-Information
Indexing: to automate or not?
Automated Approach• Software identifies keywords, eliminates stop
words, assembles inverted index.• Relatively inexpensive.• Poor performance, especially with
heterogeneous content domains and document “richness”.
Manual Approach• Human reviews content, identifies key
concepts, and selects keywords from controlled vocabulary.
• Expensive: Forrester Report, v2, n8, October 1997 says $960,000 for average corporate intranet.
• Better performance (context sensitive).
36
Bottom-Up Components: Relationships
How should content types be linked?
Ad-Hoc Links• Handcrafted; expensive to create and
maintain.• Capable of making powerful associations,
but often subject to interpretation; therefore, “hit or miss” in cognitive value.
Rule-Based Links• Simple: “If a process document and a rate
table document both deal with Product X, then they should always be linked to each other.”
• Rules allow for easy human link creation or automated linking (useful with huge collections of content).
37
Bottom-Up Design
Content modeling
“Messy” Content
relationships (links)
Product Process Reference content model
ContactHandling
“How To” document types
Sequence(step 1, 2, …)
Conditional(If…, Then...)
template typespresentational meta-
information Format/Style
ColorPlatform
logical meta-information
NameContent OwnerProcess Owner
Geographic Eligibility
38
Bottom-Up Case Study
Background and challenges.
Distributed inbound call centers (8,000 customer care associates).
Intranet-based work support application (6,000 unstructured documents).
Users were memorizing, not navigating.Erroneous information was being
provided to customers.Negative impact on training and churn.Single semi-topical hierarchy… hand-
maintained.Content management nightmare.
39
Bottom-Up Case Study
Solutions.
Structured content model.• Suite of document templates.• Linking relationships.• Indexing system with multiple
controlled vocabularies.Functional specifications (auto-generated
browsable indexes).Development and production indexing.Training and documentation.
40
Conclusion
A convergence of perspectives, communities.
Traditional Top-Down Perspective• User-centric work is what “we” (the Web
community) have been doing.• Goals: presentation and usability.
Traditional Bottom-Up Perspective• What mark-up and data modeling
communities traditionally have been up to.• Goals: content reuse and maintenance.
Another Kind of Convergence• Web community increasingly content-
centric.• Mark-up/database communities
increasingly user-centric.
41
Information Architecture Design Process
A phased approach.
42
Argus Associates
Contact information.
Louis Rosenfeld ([email protected])Argus Associates, Inc.221 North Main Street, Suite 200Ann Arbor, Michigan 48104734.913.0010 voice734.213.8082 [email protected]://argus-inc.com
This presentation available from:http://argus-inc.com/conferences/tl-seattle/