DonnaM » Search

Search

Things that make me mad – part 1

Tuesday, May 9th, 2006

I got mad yesterday. So mad that I waited until today to post.

It’s the little things that push me over the edge. And one thing that best pushes my anger buttons is guru statements that are not well-thought out or blatantly wrong.

This post (I’m not going to help its ranking by giving it a good link title) did it yesterday. G McG basically says that search is unnecessary. Based on a sample size of one. It isn’t highly-used on his site, therefore it mustn’t be important at all.

This is so wrong. As I wrote in ‘Four Modes of Seeking Information and How to Design for Them‘, it is all about the types of information-seeking tasks or approaches that people use. Were I in this situation, my conclusion would not be ‘search is irrelevant’ but ‘what are people doing on my site that means search isn’t as important’. I’m guessing most page views come from external search or from his weekly email and may conclude that people don’t otherwise come to the site with a known-item approach. That’s interesting, and you can use this information to design a better experience. But it doesn’t mean search is unnecessary.

And apart from all that, it is so poorly written it made me wince. No wonder I don’t subscribe.

Regular folks searching

Tuesday, September 13th, 2005

Last week I ran a usability test on a very neat system that allows people to search or browse a fairly large set of content. It’s been a while since I ran a usability test with regular folks (part of the hazard of working for too long on intranets, and one reason I’m glad I’m not specialising any more). The interesting thing was seeing whether people’s expectations of search had changed over the last few years (you know, the whole google phenomenon thing).

One of the things that came strongly out of this test is people’s mental model of search, and what they expect to happen. Here’s how I interpreted their expectations of search:

  • It is better to put more than one word in as one word gives too much stuff
  • Adding an extra word gives fewer results (although most search engines give more results with more than one word, people strongly thought it would give fewer – I even probed on this)
  • The first word in the search box is more important than the other words
  • If the words make a sensible phrase (one that humans would recognise as a phrase), the search engine should do so too and return results for the phrase
  • If the words do not make a sensible phrase, the search engine shouldn’t look for the phrase (yes, this contradicts the previous point, but no-one ever said people are logical)

How interesting!

Yay! I’m not stupid

Sunday, April 18th, 2004

After Friday’s rant about library search, I decided to look further and see whether I was just ‘searching wrong’ or if there is something inherently wrong with library journal searches.

So, continuing what I’m looking at for university, I worked on the scenario of ‘finding out the names of any journals with ‘ergonomic’ in the title’. This is real – this is what I want to find out right now.

I searched the National Library of Australia, University of Queensland, University of Canberra and Australian National University websites – each had a ‘journal search’ option on their library site. In each of these cases, putting ‘ergonomic’ into the search returned only titles that started with ergonomic – not with the word anywhere in the title. I was starting to feel like I must be doing something completely illogical.

But finally, Monash University saved the day – putting ‘ergonomic’ into their journal search returned exactly what I had expected.

So what’s the deal here. Am I asking for something illogical? Why can’t I expect to return records with the search string anywhere in the title? Is this some odd library conspiracy that I don’t know about?

Frustrated by library search

Thursday, April 15th, 2004

I don’t think there is a group of systems that I hate working with more than library search systems and journal databases. These are the only systems where I know that my frustration levels will reach the point where I start to go cold, grind my teeth and get an almighty headache. No, I’m not exaggerating – this is truly how they make me feel.

So tonight was no exception. Starting on my university library site, looking for journals that have the word ‘ergonomic’ within the title. Sounds like a sensible search, but no, I can only search for titles starting with ‘ergonomic’. That’s not useful – I don’t yet know what journals are relevant, so I don’t know the beginnings of the titles. So I searched by keyword and subject, to no avail. What frustrated me most was that I knew that there was a journal called ‘applied ergonomics’ available and it wasn’t coming up in any of my searches. I suspect that anyone else would have believed the results and that there were no relevant journals. I spent most of an hour going around in circles trying to figure out how to get a result that I knew existed, and in the end nothing worked.

So I went to search in a journal database that looked promising. The nice system put a cryptic string into the search box. Lucky I’m a clever IA and knew what “jn ‘Ergonomics’ and ft y” meant. But I’m not library-geek enough to remember all of the field codes to continue this search, and had to go hunting – help didn’t, but I persisted and found an advanced search with the list of field codes, in a very nice pop-up window (yes, this was a good thing). The nice system didn’t carry my cryptic string around, so I had to carry it around on the clipboard – but I’m a geek and know these things.

I found a couple of interesting articles. Clicked the little ‘add all’ button next to them & was whisked off to a very nice list of my marked articles, with options to continue searching. Yay! Something went well.

Maybe I was getting a bit overconfident. On the next database, I could easily search within the journal, and found a bunch of useful articles. I ticked the ‘mark’ box next to them and went looking for some way to save my marks (I have previously lost hours of research by not clicking some magic button). No button. So has it marked them or not? – there was nothing in the interface showing that the marked articles were saved anywhere, and no way to go and look at my marked list. It turned out (I think) that they are saved when I left the page – something that I thought was impossible (or someone is telling me stories).

From there it just got worse. Databases with confusing ugly search interfaces, poor communication (can I get full text of this article?), little to no feedback, no suggestions on what to do next. Interfaces that contained almost nothing but jargon.

So 3 hours later and my head is pounding and I’m immensely cross. This just should not happen. Other people in my class spent much more time researching and came up with nothing. It’s not their fault – they aren’t trained in information retrieval, they are university students who want to grab a few articles for an assignment. Someone, somewhere (actually lots of someones) need to start thinking about who is using these systems, what their existing skills are and what they need to achieve. Including ‘basic’ and ‘advanced’ options are not enough.

Search engines again

Thursday, January 15th, 2004

I should clarify – google was not down, nor was vivisimo – something else odd was happening as I can’t get to many websites.

Which search engine?

Thursday, January 15th, 2004

Had an interesting experience today – both Google (australia) and Vivisimo were down at the same time. My favourite search unavailable, and I can’t remember features of all of the other search engines.

Which would you use if these two were offline?

Type of search is relevant here – I’m doing a research type search looking for universities offering information architecture and usability classes. I usually use Vivisimo for this type of search both because it is a meta, and I like the filtering.

Really bad search results

Tuesday, August 5th, 2003

This is one of the worst search results sets I have ever seen. To repeat it, go to Internet World and search for Information Architecture.

Out of the first 20 results, only two showed a summary of the article.

Human OR Boolean

Wednesday, January 15th, 2003

Today I was talking to someone again about searching. He explained that he used AND to join two things. I asked whether he ever used OR. His answer was no, and when I asked why, he said:

“If I use OR, I get both A AND B – that’s too many results”

Another example of how the boolean language doesn’t match plain language…

Human AND Boolean

Monday, January 13th, 2003

I have read a few times that there is a big difference between the way that people interpret boolean AND & OR and the way they are used in information retrieval (most recently an article on searchenginewatch, found via Eleganthack)

I understand boolean logic, so have always had to take this claim at face value. I’ve been waiting to see it really happen (I don’t believe these things until I see them for myself).

Well, it happened today. I was talking to a user about searching a big information store. She showed me how she used one term to search and then checked each page in the result list. I was interested in whether it would be helpful to construct a more detailed query with more terms. I asked a question about searching for two things (being careful not to lead). She was very hesitant about it and I got the impression that she was not comfortable adding an extra term (using AND to join them).

We talked a bit more about this, until I realised why she was hesitant – she thought she would get more results from searching for ‘A’ AND ‘B’ – all of the pages with A and all of the pages with B.

Now I believe!