350 likes | 496 Views
Characterising Browsing Strategies in the World Wide Web. Lara D. Catledge & James E. Pitkow Presented by: Mat Mannion, Dean Love, Nick Forrington & Andrew Ingram. Overview. Summary of Paper Further research Relevance to today’s browsing strategies Conclusions. Part 1: Paper Summary.
E N D
Characterising Browsing Strategies in the World Wide Web Lara D. Catledge & James E. Pitkow Presented by: Mat Mannion, Dean Love, Nick Forrington & Andrew Ingram
Overview • Summary of Paper • Further research • Relevance to today’s browsing strategies • Conclusions
Summary of the paper • Context of the paper • Age • Experimental population
More Focused Browsing Intention Drunk Browsing Strategies • Search Browsing: directed search – goal is known • General Purpose Browsing: following links that have high likelihood of leading to interesting content. • Serendipitous: Completely random
Browsing vs. Searching There is a tension between designing for browsers and designing for searchers. A hierarchical searchable database may work well for searchers, but not as useful for those just browsing for unexpected results. First step in solving the problem was to determine what strategies are in use. This information was obtained experimentally
Data Collection • Special version of XMosaic coded to log user interface events at Georgia Institute of Technology's College of Computing • Study conducted for 3 week period in August 1994 • Final log file comprised over 43,000 events with each uniquely identifiable
Results • 80% of requests used http • Following hyperlinks accounted for 52% of all requests • “Back” command closely followed with 41% • Most users didn’t make use of hotlist (bookmarks) and history functions • Only 10% of events activated via keyboard • Popular sites didn’t match the sites put on the hotlist
Results Continued • Pattern Detection Module (PDM) algorithm used to identify repeating sequences of site and document accesses • For example a user going from www.gatech.edu to www.ici.edu to www.ncsa.uiuc.edu a total of 7 times would be identified as a path length of 3 with a frequency of 7. • An analysis of lengths of paths within each site visited for each user was performed.
Frequency and Path Length Analysis • Previous graph gives: Frequency = -0.24 x path length • Using the 3 Characterisations, the following classifications can be made: • Serendipitous Browser (slope < -0.24) – avoid repetition of long sequences • General Purpose Browser (slope = -0.24) – Users perform as expected, 1 in 4 chance of repeating complex navigation • Searcher (slope > -0.24) – performs long sequences often
Within-Site Navigation • Backtracking (via use of “Back” command) heavily used, can be visualised as hub-based browsing • Typical example would be: • http://www.cc.gatech.edu/people/ • http://www.cc.gatech.edu/people/People.Faculty.html • http://www.cc.gatech.edu/gvu/people/Faculty/Neff.Walker.html • http://www.cc.gatech.edu/people/People.Faculty.html • http://www.cc.gatech.edu/gvu/people/Faculty/Piyawadee.Sukaviriya.html • http://www.cc.gatech.edu/people/People.Faculty.html • http://www.cc.gatech.edu/gvu/people/Faculty/Michael.J.Sinclair.html • http://www.cc.gatech.edu/people/People.Faculty.html • http://www.cc.gatech.edu/people/ • As example shows, users rarely reach a depth of more than a few pages before returning to entry point • Paper suggests sites should be organised to exploit this user behaviour
Conclusions of the paper • That there are three different types of browsing strategy • These strategies can be matched to empirical results • That these strategies were valid for pre-web hypertext browsing, and remain so • Website designers should design with these user groups in mind
Linda Tauscher and Saul Greenberg • Two separate papers • How people revisit web pages: empirical findings and implications for the design of history systems, 1997 • Partially based on Catledge and Pitkow’s data • Revisitation Patterns in World Wide Web Navigation, 1997 • Made conclusions on patterns of information seeking
Web page revisits • 58% of web pages in a single session are re-visits • Changing information on pages • Further exploration of a page • Special purpose pages • Search engine • Home page • Information seeking is influenced by browser ergonomics • It is very easy to go “Back” in a web browser
Seven browsing patterns • First-time visits to a cluster of pages • Revisits to pages • Page authoring (refreshing changes) • Web-based applications • Hub-and-spoke (navigating to each new page around a central page) • Guided tour • Depth-first-search (link paths followed without necessarily returning to the source pages)
Huberman, Pirolli, Pitkow and Lukose • Strong Regularities in World Wide Web Surfing, 1998 • Regularities of surfing patterns • Mathematical law of surfing • Massively larger sample than previously studied • In a single day: • 23,692 AOL users • 3,247,054 pages • 1,090,168 unique pages • Also studied statistics of Xerox external site • Can predict number of visitors to a single page from its’ link topology • Implications in e-commerce websites
Choo, Detlor and Turnbull • Information Seeking on the Web: An Integrated Model of Browsing and Searching, 2000 • Many different strategies for information seeking • Sophisticated search techniques • Implications for brand building • Brands are a substitute for information • Information searching strategies make information more freely available
Are the initial results valid today? • Mosaic provided basis for modern browsers • Missing a lot of features compared to newer browsers • but browsing experience is similar
New Features that could affect browsing strategy • Search engines • Not widely used at the time of study • Didn’t crawl • Forms • Not supported by all browsers at the time of the study • Dynamic Pages / Personalisation • Pages can be generated for a particular user • Scripting • Webpages becoming more like applications • Really Simple Syndication (RSS) • Used to keep track of sites with changing content (news, blogs etc).
Evaluation of findings • Heavy use of back button – can now use… • Mouse gestures • Additional mouse & keyboard buttons • Paper envisioned page designers creating different “views” for different types of users (browsers & searchers) • Instead, we have search engines for searchers • Directories for browsing
Evaluation of findings • “alter page design on the fly based on accesses by users” • Most popular products in online stores • Also personalisation of websites (e.g. recommendations) not predicted in paper • Although a “guided tour based on paths most travelled” isn’t commonplace.
Evaluation of findings • Paper suggested image maps • Not widely used • “Document designers need to be cognizant of the classification of expected visitors” • This is generally not the case with most websites • Designers generally more aware of different classifications in terms of browser type, screen size etc • Problem of searching vs. browsing • With more advanced search engines, we don’t really need to worry about this anymore
Conclusions • The web has changed significantly in ten years • The three browsing strategies still exist • An entire industry has formed to cater for search browsing – eg. Google • Designers no longer have to cater for this – your information can always be found
Conclusions continued • While the three strategies exist, users will switch between them so often that offering different views for different users is not commonplace • Though many sites offer user registration to target content more effectively
The End… • … go home. • Oops, there’s another presentation first. • Sorry. • Go home in half an hour. • Or go to the next lecture, we don’t care. • That’s really the end. • Applaud now.