Exploring patterns in website content structure

1
Exploring patterns in website content structure

• Svetlana Symonenko
• School of Information Studies, Syracuse
  University
• IA Summit 2006
• Vancouver, March 25, 2006

2
Project Background (How Come..)

• looking for patterns
• self-guided tour preferred..

3
Project background

• Problem space
• 16 of the global population went online
  (Internet World Statistics, 2005)
• 21 of users find the information gt 80 of the
  time (Feldman, 2004)
• search most common mode of finding information,
  but most natural?
• Current state of research
• notion of conventionalization of websites
  content organization, driven by industry
  practices
• empirical evidence
• users developed expectations of sites external
  design and content structure
• research focus
• on websites visual design, with insufficient
  attention to the organization of websites
  content

4
Scope of the study

• Signs of conventionalization in the observable
  structure of website content
• Website - an integrated information package
  created by the site sponsor with a particular
  audience in mind
• sponsor audience sites discourse community,
  shaping the site as their communication artifact
  (genre theory, Orlikowski and Yates, 1994
  Swales, 1990)
• Two perspectives / study phases

5
Data Collection Dataset

• 15 websites of 3 types
• universities (EDU) - U.S. colleges
• business (COM) - telecommunications companies
• government (GOV) - state portals
• Perl-based offline browser
• collected elements of sites content structure
• Page Title, Link URL Link Label, Level
• traversed sites in a breadth-first manner
• restrictions applied
• top 3 levels only
• within the original domain
• excluded links to dynamically generated pages

6
Data Collection Dataset

• Resulting dataset

7
Data Analysis

• units of analysis - Page Title, Link URL, Link
  Label, Level
• analytic induction approach
• focus on distribution of content categories
• at different levels within the site
• for particular kinds of links (navigational) and
  pages (Home, About)
• conventionalization degrees (Nielsen, 2004)
  applied to content categories
• standard category present on gt 80 of websites
• conventional category - on 50-79 of websites
• unconventional category on lt 49 of websites
• also looked at naming conventions for particular
  categories

8
Results genre-related patterns in the sites
content structure ?

• Navigational links
• about 1/5 of all links, regardless the site type
• home is the most common link at all site types
• genre-specific importance of some categories

9
Patterns more.. Homepage patterns

• EDU and GOV vs. COM gtgt diverse content
• universal and type-dependent categories

10
Content structure of selected pages About EDU
vs. COM

• pages of the same kind, but on sites of different
  types
• About page, standard categories

11
Visit page EDU vs. GOV, standard categories
12
Type-dependent profile of site content
structure EDU
13
Category-specific lexicons About

• About
• About Us
• About Entity_Name

14
Category-specific lexicons Employment_informatio
n

• Employment (Career, Job) Opportunities
• Careers
• Jobs

15
Type-related features in overall site architecture

• EDU and GOV sites - portals to sites of
  individual departments
• institutional identity at the top 2-3 levels,
  then breaks into kaleidoscope of departments
• COM sites
• much more uniform look throughout the site
• product- or customer-focused architecture
• very few links leading off the site
• relatively shallow (at most, 3-4 levels deep)

16
Type-related features in overall site architecture

• EDU, GOV much more verbose than COM
• average number of links per page, by site type
• observations confirmed on a larger data set (123
  EDU 73 COM)
• lt 0.5 MB - 89 of COM sites, 7 of EDU sites

17
Future research

• Analysis of content structure of a larger sample
  of EDU and COM sites
• Navigational links EDU vs. COM

5 EDU and 5 COM (Pilot)
28 EDU and 27 COM
18
Future research

• Homepage EDU vs. COM

28 EDU and 27 COM
5 EDU and 5 COM (Pilot)
19
Yet More Future ?

• User Study

20
Thank You !