Facing and bridging gaps in Web searching, is the title of an article by Reijo Savolainen and Jarkko Kari published in the Information Processing and Management Journal vol. 42 (2006) 519–537. The results in this paper are derived from videoed web searching of only seven subjects. Comment is also made on previously obtained data collected during an earlier usability studies by the author, one of which on search behaviour has previously been reviewed by Sci7. To a large extent this paper appears to be an attempt to simply get out another publication based on incredibly similar experimental work and set-up to the previously published work.
While dressed up in psudo-academic language, what the authors essentially investigated was the type of problems individuals run into when searching for information on the web and how they overcome them. While the authors should be commended for trying to find relevant parallels and prior academic work which pre-date internet search - in this case the concept of “sense making”, they do not appear to have gained anything by doing this, other than giving an academic aura to their writing, and hiding their findings under an unnecessary veil of metaphor.
The author’s previous study is referred to which classifies page views as arising from a number of different actions by the user: search 8%, back button 25%, and link 45%, with other methods such as typing a destination comprising the remainder of page views. Web design which inactivates or assigns non-intuitive behaviour to the back-button is often criticised by usability experts, and is commonly an area where AJAX developers fail (Recently an early beta of Gmail suffered from this flaw, which has now been fixed|), This research confirms that comments on the importance of the back button may well have been insightful and should be taken seriously by those designing and optimising websites.
No consideration of software used by the participants is made, tabbed browser? Search application such as Apple’s Sherlock or web browser? Which browser? Such information is crucial for those who might wish to interpret or even repeat the research - though with the constant evolution of the web repetition in the same environment is impossible, and each study such as this can only hope to give a snapshot. The paper states that participants were free to choose their own browser from those available. Without knowing about the software used and the participant’s familiarity with it is is impossible to make an assessment for example on the importance of the difficulties encountered with navigation of histories.
One of the most interesting new information to come out of the work (to be viewed with the small sample size in mind), is the below table showing the importance of various navigation methods. These were used while searching for information on a broad spectrum of topics at the user’s instigation.
| Navigation method | Number of Instances | % |
| Activating hyperlinks available on the screen | 315 | 48.2 |
| Activating Back button | 158 | 24.2 |
| Inputting a search term into a search engine | 71 | 10.9 |
| Finding search terms for a query | 39 | 6.0 |
| Recalling and typing a URL | 21 | 3.2 |
| Drawing on an information resource known from earlier use | 20 | 3.1 |
| Specifying search by trying narrower search terms | 9 | 1.4 |
| Finding a useful directory or a list of links | 6 | 0.9 |
| Finding individual hyperlinks | 5 | 0.8 |
| Browsing hyperlinks in a random way | 4 | 0.6 |
| Finding advice provided by the search system | 2 | 0.3 |
| Switching language in formulating search terms | 2 | 0.3 |
| Finding useful information resources by chance | 1 | 0.1 |
The most common problem encountered by the subjects was following links to a page only to find no content of interest, oddly though the researchers say that the most common way out of such a page was to follow a link on it. Surely if a page contained a link of high enough interest to warrant clicking then the content of the page was of interest and value, certainly not suitable for classing as “no relevant content available” as the authors have done, this may account for their assigning this class of problems as the most commonly encountered.
Technical problems - sites presenting information in a format which could not be interpreted by the client machine was one further problem encountered. Users sometimes found themselves unable to navigate sites. Browser problems were also encountered, most explicitly with difficulties in navigating the “history”.
Interestingly the fact that users were separated from lists of saved sites which they would have on their own machines was thought to be important, users were noted to use their experience and memories both to take them to specific sites and to direct them to types of sites that they perceive of being particularly valuable, including university department and magazine sites.
While participants were observed to be taking note of the number of results returned for a particular search term, the authors, and to some extent the participants saw a large number of search results as “overload” regardless of the number or relevance of the links actually displayed. Clearly this information is of use to the user but perhaps the way it is interpreted negatively rather than simply as an indicator of specificity indicates that getting his information across to users is an area where search engines need to improve, clustering of results is one promising development which might help in this area.
DOI Link to original article: 10.1016/j.ipm.2005.01.009