visual topic exploration – for reals?

Remember back when I was sad about the demise of Ebsco’s visual search?  I got over it, but I never replaced it with the beginning composition students.  They still explore in Wikipedia, and a lot of them have fun with that, and I still talk about news browsing tools like newsmap in the advanced composition classes, but I haven’t had something to show that gets at that general idea of visual browse and topic exploration since the old visual search went away.

Until now?

Well, I don’t actually know.  But I know the answer is “maybe” which is something.  I was pointed to this tool this morning (still in beta, first area of concern is that I can’t tell if its going to stay free) — eyePlorer.com.

It’s a way to visualize Wikipedia information, which is something we’ve seen before.  But there’s something kind of fun and compelling about how it works.  And there are some add-on tools within the interface that could be really, really useful in the topic exploration phase of the research process.  Still, there are a couple of things that are giving me pause – I’ll get to those at the end.

First, the good.  It’s got circles.  No, seriously, I mean it.  It’s a fun interface to browse around in.

When you start the tool, you get an empty circle with a search box.  It does okay at figuring out the topic you want.  My first try was the topic of a student paper from a while ago.  I remembered this one because I had been pleased at the time that Wikipedia had a page for this student, specifically on their topic – orcas san juan.

EyePlorer wasn’t able to figure out what I meant by that search, but when I backtracked broader to just orcas, it did.  And better yet, one of the clusters of additional information was about places – and I was able to click and connect to information on the specific topic.

There’s a tool at the bottom of the screen that lets you zoom in to see more connections:

or out to see fewer:

If you click on the topics, you get a snippet from Wikipedia, and the option to get a little more.  The snippet is a link which will take you to the wikipedia page.  You can drag these snippets over to a notebook space, and move them around.

(Note – you have to have popups enabled for these things to work)

The note book thing in particular seems really potentially useful during topic exploration.

So why am I hesitant?  Two things.  First, I don’t really get the being able to click through to the Wikipedia page thing, because all of these subtopics and broader topics took me to the same page – the killer whale page from which they were all drawn.  It didn’t even take me to the part of the page the snippet was on, which would have put me closer to being able to click to another page — but I kept expecting to do that, to switch topics, within the tool and as far as I could tell in 10 minutes, I couldn’t.

This connects to the notebook as well – unless you do another search on another set of keywords, the notes that you pull over and rearrange are really just rearranging an existing Wikipedia article.  That’s not very useful.  Your notes do stay on the notebook from search to search, so that’s good – but I think you would need to build in specific guidance about research as an iterative, back and forth process, and make it clear that to use this tool to its fullest they should expect to search on multiple keywords.

That’s fine – research is like that and they should be prepared for back and forth and trying different things.  But when the term you want is right there, and you know that it is a hyperlink in the initial article, it is a little frustrating to have to re-search to get it.

The other, and more important hesitation is the clustering.  Much of the informational material on the site is in German, which I don’t read, or in the form of videos, which I don’t use.  So the answers to this might be there and I was too ignorant/lazy to figure them out.  But I don’t really understand how these different clusters (like slices of pie – color coded?  These areas are representing some kind of clustering) work.  If you mouse over the edge of the pie, you get a label – and some of those made sense (like “place”) but others – not so much.

Check this one out -

That refers to the little blue snippet – mean of transportation.  The Hudson Strait, that I can understand (though I’m not sure how it is different than the other bodies of water which go under “infrastructure”) – but Squid?  If you click on the dot, the snippet tells you that squid are a food source, so it seems like they should be below, in purple, with “milk.”

This might be a beta issue, right now it looks like the same categories attach no matter what the topic – at least I saw the same ones for peak oil and orcas.  And it also might be a language issue.  But I think this is worth keeping an eye on as a tool to encourage broad topic exploration.

words that mean pretty

This blog will never die.  It will never die because of this post.  Written in, I think, in about 15 minutes this post was just a quick thing to share a new tool that I was (and still am) really excited about.  And I’m not the only one.

So I never expected that this post would have more legs than any other post I’ve ever made – it doesn’t have the highest hit count, but of all of the posts on this blog it is the tortoise-est one.  Every day it picks up one or two or three views.

Unfortunately, those views come from people who are looking for something else.  Well, I shouldn’t say “unfortunately.”  I suspect a decent number of those people have never seen Wordle, and think it’s pretty cool.  So it’s cool by extension that they see that post here.  But they come here looking for (and this is almost always the exact wording of the search, it’s weird) – “words that mean pretty.” So, they want synonyms for the word pretty.  I’m thinking that they know what pretty means, and that they want some other words that mean that same thing.

So there are a couple of information literacy issues here, right?  The first, and probably most obvious, is that entering keywords into a search engine is not the best way to answer this particular question.  There are better tools out there.

Information Literacy issue #1

Using Google, the Wordle post comes up #8 on the result list for the words that mean pretty search right now.  So I assume that this is where most of the hits are coming from.  It doesn’t appear on Yahoo  (though there is a result about “how do I increase my dog’s understanding of words” which I find really intriguing).  Anyway, sometimes it’s a little higher on the Google list, sometimes a little lower.  Always on the first page.  The reason why people click on it is clear – most of the other results are obviously not relevant.  We have:

#1 – this one seems like it might be relevant, but it is actually a dictionary page for the word perhaps, not the word pretty.

#2 – this is a Yelp San Francisco query looking for non-english words that mean pretty.

#3-5 are random blog posts that use phrases like ‘words fail me” coupled with “pretty boring,” or “pretty words mean nothing.”

#6 is a link to the lyrics to Dirty Pretty Words.

#7 – we finally get a result that might work.  It’s a WikiAnswers question that just says -  “words that have pretty much the same meaning.”  But the description says “other words for pretty, same meaning as pretty” and so forth.  But when you click through to the page, you don’t see those questions. Instead, you find out that the initial question was looking for a definition of the word “synonym.”  Still, asking the question on WikiAnswers would probably work.

And then my post at #8.  Not that I really have to convince anyone who reads this blog that a search engine isn’t the place to find synonyms and antonyms.  So the first information literacy issue is a tool issue – there are things called thesauri and they can be really useful!  Check them out.

Information Literacy issue the second

Which connects to the more subtle information literacy issue here.  Which goes beyond how search engines aren’t a great starting point when you’re trying to find or generate synonyms – to finding and generating synonyms is a pretty fundamental part of effective keyword searching in search engines.  If you understand how keyword searching works, you know that the search  words that mean pretty will bring back anything with the disconnected terms words, mean and pretty. Which as the result list above indicates, is a whole lot of stuff you’re not interested in, including a random blog post about Wordle.   So when you get that result list, if you know how keyword searching works, you can troubleshoot that search and say “hey, I think I need a more specific term to get at the concept words that mean.”  If you’re really savvy at that point you might get the word “synonyms” from the WikiAnswers result and re-search using the terms synonyms and pretty.  That search works – you get result after result listing other words that mean the same thing as “pretty” does.

But here’s the thing – a lot of people don’t know how keyword searching works, in search engines or elsewhere.  Or they maybe kind of know, but they don’t really think about it.  And they don’t know how to troubleshoot that first failed search, or to find synonyms that will work better.  So I went looking – what would work better?  Because, as it happens, I’m working on a new keyword assignment – that I started talking about a few days ago, and that Sara talked about here – for beginning composition that will try to get at some of these issues about keywords and how they connect to critical reading, writing, thinking, as well as searching.

So, if you are wondering where you can find some information about other words that mean pretty – check these out:

Lexipedia: Where Words Have Meaning:  This one is interesting – it is based on the WordNet project at Princeton, and it creates, fairly quickly, cool webs of related words –  synonyms, antonyms,  fuzzynyms and more.   The webs are color coded so that you can glance at them and know that synonyms are olive green and antonyms are dark red.   The site looks a little bit messy, and it is hard to find.  While it has the domain “lexipedia.com” – a search on “lexipedia” brings back a lot of references to another project, about Wikipedia and handhelds.  Still, this one works pretty fast, provides a lot of terms that might be useful, and I like the glanceability of it.

Similar to this is Visuwords – an online graphical dictionary.  This one is prettier, but the resulting display isn’t as complete, and I’m not sure as a tool for finding additional terms and synonyms it would be more useful

And for the more textually oriented, there’s Definr, that also uses the WordNet project data.  Interestingly, it’s main selling point seems to be speed.  And it does define words really, really fast.  Not surprisingly, given the source data, it also provides some synonyms and related terms.

Both definr and lexipedia are user interfaces on top of the data generated by WordNet at Princeton.  This project, which groups words into “sets of cognitive synonyms” has about a million related projects listed on its website.  And the idea of cognitive synonyms is interesting, right?  For thinking about connecting terms to concepts and troubleshooting searches?

And now, as a bonus librarian answer – according to the OED, the first definition of the word pretty (adj.) is “cunning,” “crafty” (originally), and “clever,” “skillful” or “able” (later).  It was first used in this way in 1450.  “Aesthetically pleasing” is the second meaning, and it was first used this way about 10 years earlier.

“Sitting pretty” dates back to 1915, in Lincoln, Nebraska and “pretty please” dates back to 1891.

happy birthday to me from delicious, kind of

I don’t have an iPhone, so there are many web sites I’d like to use on my phone that just don’t look very good. Delicious has always been one of those middle-ground sites that looks okay because there just isn’t much to it, but that isn’t really optimal. There’s always been a lot of stuff to navigate through before you get to your bookmarks – like your list of tags. And if you have a crazy long tag list like I do, that’s a real hardship.

So I was pretty happy to see that there is a mobile delicious option now.

http://m.delicious.com

I’m not sure it’s 100% awesome. It’s basically, login, list of bookmarks, and the ability to filter by tag (if you know the tag you want). I think for most of what I want to do, this will do.

horrible mac photo booth photo of my non iPhone

horrible mac photo booth photo of my non iPhone

But I don’t see an option to search my bookmarks and as someone who frequently makes the case in presentations that the searchability delicious offers is one of its best features, that might be a problem. Still, as a friend of mine just said to me on another topic – “progress – yay!”

on tag clouds and teaching

Inspired in part by a conversation from the Information Literacy Summit last week, and in part by this post from the awesome David Silver, we made tag clouds in my class this morning.  U-Engage is a first-year experience class, providing an introduction to what a research university is in general and to OSU in particular.  There are 100 students in the section I teach (actually, there are somewhere around 90) and the room is a classic, old-school lecture hall.

The size and the room make the course difficult in terms of planning engaging activities for the lecture time (the assignments and recitations are a different story – I think they’ve had lots of interactive activities there).  This has been compounded by the fact that most of the course content in the middle of the term has been delivered by guest lecturers.  I’m co-teaching the class and both of us have been feeling that we really wanted the last few weeks to be interactive and engaging because we have been kind of on the sidelines for so long.

My co-teacher had a brainstorm for last week’s activity, but we still didn’t have anything ready for this week – the final week before their group project presentations.  The textbook wasn’t inspiring.  There is some great information about the value of reflection (which has been a theme throughout the course), and on goal-setting but that stuff didn’t lend itself well to the kind of activity we had in mind.  Reflection and goal-setting kind of needs to be personal to be meaningful the way these things are presented in the text, and we wanted something social and collaborative.

So at the IL Summit my husband Shaun talked about how he has an assignment requiring students to pull keywords out of the readings they do.  It was fascinating to hear about that because his students’ difficulties finding keywords in many ways mirrors the difficulties we see students having in the library when it comes to choosing keywords.  But because he is working from a known thing “what are the keywords that the author uses that capture the key ideas in THIS TEXT” he has an advantage we don’t – it seems like that would work way better as a first step than what we frequently have to do – “try and predict what words authors will use in this discourse that is entirely or almost entirely unfamiliar to you.”

In his presentation, Shaun mentioned that he is still working on ways to make the larger class discussion about the keywords work better, and I thought about David’s post above.  But it wasn’t until I was working at the reference desk on Saturday that it all came together in my head – collaboration, social, keywords and tagclouds.

So here’s what we did – I demonstrated Wordle because, like David, I expected that most of the class wouldn’t know what a tag was and by extension, wouldn’t know what a tag cloud was.  I wanted to get two ideas across:  the idea that the tags needed to be single words or at most two-word phrases, and the idea that the tags would display larger if they were used more often.

(I also thought that Wordle would be of interest to some of the class, that it might come in useful for its own sake in their final projects, and that some of them might find it a useful study tool).

Then we asked them to come up with 5 keywords that described their first term/ transition to college and write a reflective paragraph about that.  Once that was done, we go to do the social part.  We had them get in groups of 4-5 and create a tag cloud out of all of their keywords.   This is why it was important to me to make sure the concept of “more use = bigger text” was clear.  To do that right, they would need to talk about their keywords, what they had in common, and if they maybe chose slightly different words to get at the same concept.  These small groups also had to come to consensus on five “group” keywords.

Looking at the group keywords after the fact, I was surprised that a lot of the groups seemed to put together their five not by choosing the five most “popular” keywords (the ones used by most people) but instead that they tried to choose one of each group members’ keywords.  I think this makes sense when the assignment is as personal as this – if they were trying to choose keywords that would reflect the meaning of another person’s text, I would expect this might be different.  In addition, some groups used keywords for their “group” keywords that didn’t appear on any of the individual lists.

ETA – we took pictures -

uengage1

Then, each group had to send one member to the chalkboard to come up with a group tagcloud, using the chalk that was available.  On one level, this would ideally take a while as the groups compared their five words and decided how best to represent them.  In practice, it didn’t work out that way, but I think that the way it did work out was better.  As it turns out, there was very little overlap in terms – in fact, there was less overlap in the group terms than there was in the individual terms in one sense.  So while there were a few terms that should probably have been bigger because lots of groups had them, there weren’t many like that.

ETA – more pictures -

uengage31

Secondly, even though there were 14 or 15 people up at the board making the tagcloud (remember – there were about 75 people in class, and the small groups were only 4-5) that meant there were still WAY more people than that still sitting in the lecture hall seats.  This means that every minute the representatives spent figuring out the “right” way to make a tagcloud was a minute the vast majority of the class had to be dis-engaged.

So instead, people talked some and made some words bigger because they were repeated.  But they also made some words bigger because they were the words that the smaller group had felt were the most important.  And there’s a validity to that, and a meaning to that, that I don’t think our original plan had captured.

ETA – last photo –

uengage21

All in all, I was pretty happy with how the exercise turned out.   I think this kind of exercise could have been especially effective earlier in the term, before the students knew each other as well as they do now – to humanize the 100-student classroom environment.  And on that note, I think this kind of exercise would work really, really well in library instruction sessions as well.  Combined with Shaun’s idea above about pulling keywords from a text, or perhaps using keywords generated another way – it’s a safe, collaborative way to talk about the connections between ideas and the terms we use to describe those ideas.

And isn’t that the big picture philosophy behind keyword searching – I mean, isn’t that the fun part?

Finally, I have to say that we could have just had each group representative put their terms into wordle – but I don’t think that would have worked as well.  I think the physicality of the chalkboard and the actual social/cognitive act of having to do themselves what the computer does for us was important in making this engaging.  In this case, the sheer number of people involved and the limited time we had meant that some got more out of it than others – and the chalkboard session had more of a free-for-all feel to it than a Deep Thoughts feel.

This also seems like a great way to connect class work/ reading with library session work.  I think we all feel like the “teach how to drive the databases” one shot feels disconnected from the learning that goes on in class in a way we don’t like.  We talked after the Summit about building in some keyword exercises in our beginning composition classes, at the start of the term in the non-researched papers, and using those keyword exercises to get at the critical reading piece that Shaun talked about with his assignment.

I think that’s a great idea on its own, but I also think that then building from keywords there to keywords while exploring ideas and finding your own sources might draw a connection between the idea of critical reading, writing and research — all of the pieces of the course.   With the tagcloud exercise, it’s easy to think of ways to do that in the disciplines as well – a pre-library session assignment identifying keywords from a class text, a library session tag cloud of that text and another of the student keywords.  And so on and so on and so on, leading up again to student-generated tag clouds representing ideas about research and suggesting pathways they can use to research about ideas.

browsing the public discourse

Last spring, I talked about showing a news visualizer from MSNBC called Spectra to a group of advanced composition students. And I talked about how none of the students chose to use that tool when they got to the hands-on portion of the class.  I thought about that again this morning because three sections of that same course came in for library instruction.

An aside – these classes aren’t the typical “how to find a scholarly article” sessions that I do.  The students are being asked to engage with the public discourse in these papers.  Instead of the “find three peer-reviewed journal articles” requirement, these students have to find three editorials or letters to the editor, as well as public conversation on websites like blogs or discussion boards.  So what I show is very different than the things I show in most of my sessionstechnorati authority ratings, advanced searches on Lexis-Nexis and the like.

So we want to encourage some broad exploration of the conversations going on online and in the news media in this class.  Especially this term, because in all 3 classes the students were turning in one paper and just beginning to think about the research paper assignment.  Especially with this kind of find-the-public-conversation topic, it is so much easier when they browse the public conversations and find something that catches their interest and sparks their curiosity than it is when they decide on a specific topic and then have to knock themselves out to find discussion about that.

So this year, I pointed them at Wikipedia, which is an obvious place to browse.  But assuming they all knew that and how to use it, I also showed them newsmap.  This visualization tool has been around for several years now (long enough that they describe themselves as being in need of an upgrade).  The information here is just Google News, but displayed (using flash) as a treemap.  It’s a slice of what’s being highlighted right now (or ten minutes ago, or an hour ago) by Google News.  You can’t search it, you can only browse it.  You can browse by some very broad subjects (health, sports, world news, etc.)

And you can also drill down a bit by geographic location.  Newsmap lets you choose to look at stories from Australia, Austria, Canada, France, Germany, India, Italy, New Zealand, Spain, the U.K., and the U.S.  You can also display all countries at once and see them next to each other.

newsmap

So here’s the thing – lots and lots of the students chose to use this tool for browsing.  Even though it can’t be searched.  Part of that, I think, is because of where they were in their process.  Most of them hadn’t even taken the time to think of a general topic area – they JUST finished the previous paper last night or this morning.  So they were more amenable to the idea of browsing broadly.  Part of it, though, is that they obviously found the interface intuitive.  They weren’t just clicking on stories, they were using the tool to browse by country and by subject – everyone I saw was very active in how they used the site.

I don’t know if they were getting the larger ideas about the patterns of the data that the treemap re-presentation of information is designed to provide.  I don’t know if they were seeing how the format “ironically accentuates the bias” of the news, as the site creator claims.  They asked me things like “how did you get to that colorful site,” and “where did you find that really visual thing.”  And in each class, at least a third to half of them were using it for at least part of the time.  So compared to last year, I call that a win for visual browsing.

Gray Lady not so gray, actually

Somewhere, I thought that I had listed some of my favorite news visualizations from the New York Times.  The NYT has really set itself apart among major newspapers with its creative and useful and glanceable visualizations.  It’s the only newspaper I see regularly featured on my favorite infoviz blog – Information Aesthetics.

But I can’t find it.  I still think it’s there, but I can’t remember what I was talking about when I wrote it.   So this isn’t my favorite example – just the most recent one I remember:

New York Times Endorsements Throughout the Ages

So this morning I read (in Information Aesthetics, of course) that the NYT is partnering with Many Eyes to open the visualization lab up to the rest of us.  There are only a few data sets there right now to play with but the topics range from baseball to religion to Sarah Palin. You have to work with them as-is, it’s true.  So like many other projects the ultimate value of this will be determined largely by the quality of the datasets the NYT makes available.

From the About page:

With Visualization Lab, NYTimes.com users will be able to visualize and comment on information and data sets presented by Times editors, share those visualizations with others and create topic hubs where people can discuss specific subjects.

The visualizations that have been done will look familiar if you’ve looked at Many Eyes before – charts, maps, network graphs and more.  There are also tag clouds, and Wordles, though I’m not sure what Wordle’s connection to the project actually is.

The awesomeness of the NYT visualization project isn’t an accident, it’s intentional.  At last year’s InfoVis conference, Matthew Ericson’s keynote on bringing visualizations to the masses underscores this face (this account at the Visuale blog is thorough and interesting, though more focused on maps and mapping than the keynote was.  It also includes a link to the slides).

Bertini, linked above, says at the end of his account that the one thing that remained obscure after Ericson’s keynote was the tools the NYT was using to make these visualizations.  Certainly, the tools from Many Eyes and Wordle have been available to all of us for a while – this doesn’t answer that question.  But it does highlight how powerful some of the tools available to us on the emerging web are.


How I’ll transfer my Olympics obsession to politics

via TechCrunch – C-SPAN gets a lot of things right…

Several months ago, Karen pointed out how libraries could learn something from information portals created by major media outlets and news organizations.  The example she used at the time was a site about the U.S. Elections produced by the Globe and Mail.  There was a lot of stuff going on on the site, but the overarching theme was a rich body of information presented in interesting-looking and attractive graphic forms.  By presenting and re-presenting the information, the Globe and Mail provided some context and analysis to the data, and also supported the user as they made meaning out of that information themselves.

Karen asked then -

is this a kind of instruction that we’d like to work towards developing?  Should we be training ourselves to create materials like this–if not a guide to the electoral process, then maybe an interactive history of African-American migration after emancipation, for that American History class we teach every year?

I think it is.  Another major media outlet that has generated a bunch of notice lately for its informative and engaging visualizations has been the New York Times. Examples here, here, here and my favorite, here.

Karen pointed out at the time some of the barriers to this kind of thing, one of them being money and another expertise.  After all,  So today, when I saw that C-SPAN is augmenting its coverage of the upcoming major party conventions with a variety of familiar social software tools, I thought back to this post.

C-SPAN’s got two connected convention “hubs” augmenting their main politics site.  They’re going to launch for real in the next few days, but you can get there now from the politics site or the TechCrunch review.  They’re video-heavy, but much more stripped down and flexible than the videos normally provided by C-SPAN.  These will be embeddable.  There are also connections to YouTube, and to Twitter, and content aggregated from several blogs – some state-focused, some national.  Bloggers can ask to be included in the aggregation, and can also ask for help getting their hands on video not readily available from the site.  It’ll be interesting to hear how well that works.

There’s also a twitter feed.  By using official tags (#DNC08 and #RNC08), people twittering from the conventions can get their comments included on the C-SPAN feed.

What I like about the site isn’t just that it’s exciting and more dynamic because of this content, but the social content seems purposeful.  One of the TechCrunch commenters grumbled about C-SPAN jumping on the twitter fad, but I don’t know – I think this is the kind of focused thing twitter can do well.  I use twitter now to follow live sports events in other countries, or film festivals I can’t go to.  And recently I picked up a few people to follow through the  TV Critics Press Tour and Comic-Con.  This seems similar – and like it could provide two kinds of information that twitter provides very well – that vicarious sense of being there and in on exciting things as they happen and the sense of being connected to other people participating in or watching the same thing you are.

So I think this might be a connected but not exactly the same example of the ways we could build information portals that would help our students and our users make meaning out of the information and events around them — both with us and with each other.