Educational uses for Tag Clouds
Well most of them are educational anyway. As promised here’s a summary of some of the ideas you’ve forwarded to me concerning the use of tag clouds in education. Basically this is a list of some of the ‘content; which can be tag clouded to reveal underlying information.
Dissertation and essay summaries: I created a tag cloud of my PhD dissertation in a previous post (Tag Clouded PhD Thesis). I was interested to see if the cloud could reveal any additional information or meaning within the work. The cloud highlighted the following words as being most popular: learning, knowledge, student and method. This work was concerned with computer aided learning and I was amazed at how the cloud revealed the key focal areas of the research. Although I couldn’t detect any hitherto unknown meaning I felt it was an adequate summary of the work – I felt reassured and vindicated.
Strategy & speeches: Robert Wright tag clouded the University of Hertfordshire’s strategic plan, (UH Strategic Plan, Tag-Clouded ). This is useful from a strategic point of view in that senior management can get a sense as to whether the underlying principles associated with the message or brand have been covered. Essentially you can get an overall feel for the key concepts underlying the text. Also you could use the cloud as a basis for measuring a particular concept over others. For instance the strategy mentions ‘research’ more than ‘learning’. Future strategies could seek to re-dress this in the favour of ‘learning’. But more to the point you can check to see if key words, and hence concepts or ideas, are either underrepresented or actually missing!
Student Assignments: Pedagogical Gregory (blog) created a tag cloud In order to locate some connections / patterns in the works of his students. “I randomly selected half of the submissions for an assignment, removed any names, and then pasted them into a giant document. Next, I pasted this into TagCrowd.”
The assignment was a 300 word summary of Kreiser’s “The Enemy Within,” (about the Spanish Flu of 1918). He selected the top 50 words. The resulting cloud can be seen here.
What’s interesting is that he is not analysing the content of one students’ effort but in fact is analysing the aggregated content produced by the class.
Student Feedback: Mark Russell invited his students to describe in their own words what constitutes good teaching, (Tag cloud - good teaching). Interestingly the exercise will be repeated with internal and external faculty. Again another angle on using tag clouds to highlight differences / trends. In this case Mark wishes to examine what both groups value and the language used. In the short term he’s content with the “…quick and dirty messages jumping out with the tag cloud.”. However in the long term a more sophisticated method that captures the context is desired. And, as pointed out by Carl Clare, the exercise is also valuable in terms of what is NOT mentioned by either staff or student.
Audience Feedback / Stimulating Discussion: Mark Russell again asking the same question, this time to faculty “In your own words what do you think constitutes a good (student) learning experience”. In this case the audience were asked this question before he gave a presentation (last two weeks), and the tag cloud showed the top 30 words. Mark points out that this exercise shows how out of class activity can inform in-class activity. Incidentally ‘feedback’ was missing!
Website / Brand summary: Basically tag clouds could act as a summary of the websites content and hence aims / vision. To an extent tag clouds are already used on blogs in which the tagged posts are all collated together. It’s a pretty good way of showing to the reader the main focal point of the blog and the same could be done for institutional websites, (but through processing actual site content as opposed to aggregating existing tags as per blogs). In particular the cloud could change as more material / documents are added to the website – it could be very dynamic.
Creating tags and categories for content: tag clouds can be used to provide keywords / categories / tags summarising a body of work. So applying this to my dissertation I would tag the work with: learning; knowledge and student – words in themselves revealed and emphasised by the tag cloud. This is a good way of classifying the content of a blog post or providing keywords for abstracts and papers, (in conversation with Steve Bennett).
Search results summary: Provided by Carl Clare, Tag Clouds for Summarizing Web Search Results, is a brief report of the using tag clouds to summarise search results from the Pubmed (medical) database. Basically an application, PubCloud, uses tag clouds for “the summarization of results from queries over the PubMed database of biomedical literature. PubCloud responds to queries of this database with tag clouds generated from words extracted from the abstracts returned by the query. The results of a user study comparing the PubCloud tag-cloud summarization of query results with the standard result list provided by PubMed indicated that the tag cloud interface is advantageous in presenting descriptive information and in reducing user frustration but that it is less effective at the task of enabling users to discover relations between concepts.” So again tag clouds are great for initial information provision but less so for deeper connections.
Published material: Damian Bariexca (Apace of Change), comments:
“I’d like to feed this thing an act or two of Macbeth and analyze the tag cloud results for imagery, then draw inferences about the abundance or dearth) of references to specific imagery.”.
A good example is provided by Pedagogical Gregory (http://gz7comp.blogspot.com/) who uses a tag cloud to visualise the Homeric epic, (although ideally it should have been filtered for ‘footnote’), (view cloud).
Perhaps there’s scope for quantitative analysis of qualitative data? For example student feedback, student interviews, staff /student views and interpretation of institutional strategy. All could be converted to text and subjected to the cloud treatment.
So can tag clouds actually provide anything other than brief summaries of content? I’m still digging but the following uses have been suggested:
Animated timelines, tracking changes, highlighting differences: Robert Wright suggested obtaining an earlier strategy document from the previous year and “use only a selection key words from both reports to generate two word-clouds, and then show an animated evolution from one cloud to the next, the result being the change in focus in certain areas visually displayed over time, (some words shrinking and others growing in size) - based only upon their frequency of use.”
Unfortunately we couldn’t locate an earlier strategic document so that will have to wait. However I did see tag clouds based on two successive the State of the Union address by George Bush (Tagging Speech ). The clouds showed clear differences in the underlying tone of the speeches, reflecting the current political climate.
Mark Russell also suggested doing something similar but in poster format. This year will see the third International Blended Learning conference (details here) hosted at the university. Mark intends to tag cloud the abstracts submitted for each year of the conference to produce a set of three posters – one for each year which not only details the educational concerns for each year but also provides a timeline in which rising and falling trends (or concerns?) can be detected. We have no idea what the clouds will produce and I guess this is part of the excitement of using them.
Related tracking-change ideas spring to mind such as comparing student evaluations over time or comparative analysis between sets of users (students vs, staff), website content and group blogs at particular points.
Checking what’s missing
Equally as important as what is present is what is missing from the cloud. Tag clouds could be used to check to see if anything is missing from assignments, feedback, strategy and speeches e.g. are students mentioned?
Finding Hidden Meanings?
After tag clouding the University of Hertfordshire’s strategic plan Robert Wright teased out some key-phases which came together from the alphabetical ordering, For example “business-accelerated, future-global-graduate”, which is not unlike the kind of student this university aims to produce.
Current Limits & Future Dreams
Of course there are limitations as pointed out by Carl “Without context many tags are not of much use - there are ways of ‘clouding’ common phrases but the real golden egg will be the semantic cloud - tag clouding by meaning as opposed to pure word repetition..”. So tag clouds are useful for initial exploration but at present there are limitations to the themes that are generated.
Carl suggested another golden egg which would be to plug multiple RSS feeds into a tag cloud generator to measure the Zeitgeist in an area. This would particularly apply to news or sites whose content are updated regularly. As Carl points out this is doable manually by creating a ‘mashup’ and manually plugging it into a cloud generator but the preference would be for a ‘live’ cloud generator that then fed into a time line. I could see further applications here such as capturing online group activity, online forum discussions and module website activity. Another one sprung to mind and that would be to tag cloud the exchanges / content created within a social network or friends or blogging community, (yes I know standard tag clouds exist but they are mainly based on author applied tags to posted content and not necessary the content itself. I wonder if a tag cloud based on all my posted content would be identical to the standard tag cloud, I wonder if what I feel the posts / blog represent as a whole would be reflected?).
Some useful links: (thanks to Carl Clare).
* Joe Lamantia provides a good overview of tag clouds here and here
* Also this post includes links to text/tag cloud generators.
Cloud Generators:
I used IBM’s Many Eyes. Just cut and paste your text into the box and it generates a tag cloud for you.
