In the development of our second interactive prototype we focused on four general areas of change:
- Addressing the most severe heuristic violations raised by Manis;
- Improving the visual layout and design;
- Adding enough content suitable for a usability test;
- Adding in some functionality we did not build into first interactive prototype but is necessary for a usability test.
Additionally, we decided as a group to ignore some of the general terminology issues unless they were misleading or inconsistent in our usage of them in our site. There is a language that is developing around the concept of blogging. Some of this language (such as the word "blog" itself) are becoming more and more mainstream. It would probably be counterproductive to create an interface that makes these terms "intuitive."
Major changes made:
New format for individual search results
Based on some of the heuristic
violations raised by Manis we redesigned the format for the search results
so that the
"context" information would be offset from the search results "content."
We actually had attempted to do this in our paper prototype, but found that
users would only look down a list of results and would not look to a side area
that had data about the post. However, by placing the search results title
(the blog post title, the author, and so forth) above the content and context
information we could address this issue.
We also limited the number of characters displayed in the search result content highlights (the KWIC sections). Ideally, the right content would be pulled out in the same manner as Google or other search sites. We incorporated some rules but could not implement the ideal.
Addition of results summaries
Manis found the section headings confusing, and after conversation with them,
we believe that this was primarily the case when there were no results found
for a particular section. We discussed various ways of dealing with this
issue. In the end, we decided to summarize the results at the top of the
page and then not include that section if there were no results for a particular
section.
Removed the post topics filter from the topics page
The original idea was that
if someone were one a topic page for a topic that had lots of subtopics, there
would be a lot of posts on that topic and it would be beneficial to allow them
to filter by those subtopics. However, the Manis evaluators were confused by
how the subtopics filter related to the topics browsing hierarchy in general
and it was especially confusing on this page (as we allow users to navigate
down the hierarchy by presenting the subtopics first in the page). Therefore,
we traded power for simplicity and removed the topics filter in the posts section
on this page.
Modifications to the context page
Partly in response to some feedback from Manis, Professor Hearst, and our own
intuition, we made some changes to the context pages. We had a lengthy debate
of somehow showing chaining history and could not reach a solution we liked
as a group. The only thing we were sure on is somehow visually indicating
which post a user had just come from in the chain. That post would appear
first in either the "responses to this post" list or the "links
from this post" list. Our current solution is another icon. For the
moment, the text should be considered as placeholder until we find a better
solution.
Note: The solution is still not quite working as intended. Rather than sticking to "the chain," clicking the graphic only takes the user back to the post he/she just came from, not to the one that resulted from a click on "show more context."
We presented some thoughts to the class on 4/12/04. They had some interesting ideas. We're still trying to figure out how to incorporate them.
We did decide to incorporate news articles in the context page. We realize that this potentially opens up many questions such as "should there be a profile page for each news source?" We decided to ignore news articles, for the moment, in all other areas outside of this page. As so many blogs link to news or non-blogs, it seemed impossible to even discuss "context" without including them here.
Finally, we switched the placement of the context sections so that chaining down took precedence over chaining up.
Icon design
One of Manis's evaluators found the icons unreadable in general and another
found the profile icon confusing because of its proximity to the blog link
and some inconsistencies. In general, they liked the context icon, though.
We still felt we needed the profile icon rather than only a text link to
the profile page. So, we made the profile icon more of a graphic that included
the text "profile." While, we're not settled on a final design, we feel that
this will eliminate the confusion while providing visual interest on such
a text-heavy page.
Links to "home" and "help"
While they liked having the search box on every page, the Manis evaluators
wanted a way to start over from the front page of topic browsing. We considered
having an "All" topics page (like any other topics page) but decided to start
with having an explicit link home.
We also added a link to a general help page, though did not implement the help.
General design changes
The first interactive prototype did not emphasize visual look and feel, though
it did work on layout. We were happy to see that the layout passed the heuristic
evaluation. We made some changes to incorporate a softer look.
Other changes
- implemented the topics filter check-box interaction (we had a plan before the first prototype but didn't put it in our scope)
- made the search term appear in bold everywhere on the search results page
- changed ordering and layout for blog matches in the search results
- fixed a font issue we were having that did impact readability of results of first prototype
A note on visualization
First, though, we spent some time thinking about the use of some type of information visualization in the context of addressing some the heuristic violations and addressing some concerns of Professor Hearst. After some thinking and debate, we decided that while visualization would probably be helpful in many areas (perhaps even more effective than text) even a simple visualization such as a line graph to represent post activity over time would be difficult to implement in a way that would be useful for our usability tests. For example, we could see creating some non-dynamic visualizations to put in front of users, but a dynamically driven visualization would not be easy given our group resources. And, if the visualization was not dynamically driven like the rest of our content, users just wouldn't get it. However, we have been in contact with some groups in the Info. Viz class and hopefully, there may be a way to link the work in the future.
Response to Manis heuristic evaluation
We noted above that many of the changes made were in response to violations found by Manis. Below is a summary of how we addressed those issues. The numbers at the left were ones that we provided for Manis's violations. This table only includes the level 3 violations but supports adding in other changes to less severe violations.
# | Description | Violation | Severity | Our changes |
1 | Initially, it's unclear how to tell which post has been linked to by the most bloggers (task 1b). Eventually I found it under "show context of this blog" (while I was browsing through the interface). | H1 Simple and Natural Dialog | 3 | Changes to search results format as described above addresses these issues. We have taken away the word "activity" and reformatted the results so that the high level context information stands out more.
|
2 | The blog statistics are hard to find. The tool tip for the icon says "Profile," which doesn't necessarily suggest statistics. | H1 Simple and Natural Dialog | 3 | This is a task terminology issue... not sure what the right solution is right now. |
8 | The visual similarity of the text makes it difficult for users to distinguish between different elements of content. | H1 Simple and Natural Dialogue | 3 | Resolving (1) addresses this issues. |
9 | Users may find it difficult to navigate the site. Is the topics hierarchy meant to aid in navigation of the site? The breadcrumb structure seems to indicate that it does, but I'm not certain that was the intention. | H1 Simple and Natural Dialogue | 3 | After discussion with Manis, this had mainly to do with confusion about topic browsing, not site navigation. The two main changes we incorporated to address this issue were:
|
13 | The 3 section headings on the search results page are too terse: Blogs, Topics, and Posts. It would be nice if there was more information about how these sections are related to the search term (e.g. Posts containing the term "Clarke"). | H2 Speak the User's Language | 3 | We are not having a search results summary at the top of the page. If no results are found for a given section, that section won't appear below. We don't think that the problem is that the labels are too "terse." We think that the content they found when they did the search did not provide a good understanding of what they were seeing. Additionally, we are making the search terms bold. |
19 | Currently there is no indication of what the user entered as a search term. It would be helpful for users to know what they entered or clicked on when looking at a result set. | H3 Minimize User Memory Load | 3 | We have made the search term bold wherever it appears (following convention of Google). |
20 | Reduce the screen real-estate taken up by the post summary so more results fit on a page without scrolling, allows the user to scan more results and assess current search. | H3 Minimize User Memory Load | 3 | Solved by (1) |
22 | Inconsistent use of terms. The Context Page says "linked to..." while Results Page says "responses to..." Unclear if these phrases are supposed to mean the same thing. | H4 Consistency | 3 | Solved by (1) |
28 | Tool tips for hyperlinked icons would help users figure out what they are. Without some sort of help documentation, they may confuse some users. In general, a good help section will be essential for Paparazzi. A tagline under the site name may also help users figure out what the purpose of the site is. | H10 Help and Documentation | 3 | We have included a link to a main help page, though the help has not been implemented. We are also considering appropriate places to add more context-sensitive question mark icons. Manis understood those. |