8 thoughts on “Read to Learn: Updated”

1. Hi Owen
   I’ve been trying this with our public library catalogue, using the url method, rather than isbns. I tried the following –

   http://www.meanboyfriend.com/readtolearn/studysuggest?match=0.5&url=http://www.surreylibraries.org/02_Catalogue/02_004_TitleResults.aspx?page=1&searchTerm=java&searchType=99&searchTerm2=&media=&branch=&authority=&language=&junior=&referrer=02_001_Search.aspx

   I didn’t get any recommendations. I suppose this could just be because the stock in our libraries is understandably less academic than Huddersfield and Lincoln and therefore can’t match based on the ISBNs in the opriginal dataset. I’d really like to see how this works with our catalogue, so I’ll have to think of another plan of attack of identifying what ISBNs in the dataset you’re using match ISBNs in our system. Think I’ll have to do some comparisons.

   I couldn’t try it with any items I’ve got on order, or with a list of titles I borrowed, as the url just presents a generic aspx url, without details.

   It is an interesting idea and I would like to see it working with some of our real world data.

2. OK – realised that this is because the URL needs to be encoded – otherwise the ampersands (and other characters) in the submitted URL are mistaken for part of the applications URL. I’ve now updated the script – you have to submit an encoded URL. If you try:

   http://www.meanboyfriend.com/readtolearn/studysuggest?match=all&url=http://www.surreylibraries.org/02_Catalogue/02_004_TitleResults.aspx%3Fpage%3D1%26searchTerm%3Djava%26searchType%3D99%26searchTerm2%3D%26media%3D%26branch%3D%26authority%3D%26language%3D%26junior%3D%26referrer%3D02_001_Search.aspx

   You should find this works (note I had to use the ‘match=all’ as a 0.5% overlap doesn’t get you anything.

   I’ll look at adding some better error checking/responses so you can more easily see whether it was that no ISBNs were found in the page, or if they were found, but there were no matches.

3. Thanks Owen.
   I’ll have a look at which other parts of our catalogue this will work on.

4. I’ve added in a bit more feedback in the xml file so you can (hopefully) see what is happening, even when no courses have been recommended. At the top of the xml file is a new element, which contains a list of all the items (with isbns) that were used by the script. This list includes any errors encountered when trying to retrieve information on the ISBN from the MOSAIC usage data.

   I’ve also included any errors from attempts to search the UCAS course catalogue for course codes in an element (within the top level element)

5. Pingback: Tinkering in March 2010 « Information Twist

6. Hi Owen

   Following on from our recent chat here’s some thoughts on the results Read To Learn is returning. If I was someone considering a course, I’d be interested in where the course was and a brief description of it. I wouldn’t be worried about course/provider codes at this stage. It’s handy having the link to the course.

   I tried it with this url. It worked on the 2nd attempt, but was very slow. First attempt timed out. Not sure where the speed problem is.

   http://www.meanboyfriend.com/readtolearn/studysuggest?match=all&url=http://www.surreylibraries.org/05_ReadingLists/05_004_ReadingListTitleResults.aspx%3Ftype%3DLIST%26code%3DEATING%20DISORDERS

   I like the idea of this. It could be very useful.

7. Thanks Gary,

   The speed problem you are seeing is because one of the courses matched by the ISBNs is course code L500. In the MOSAIC dataset there are over 40,000 rows related to this. My script gets the information from the MOSAIC dataset using the URL:

   http://library.hud.ac.uk/mosaic/api.pl?isbn=&ucas=L500&show=minimum&prog=&years=&rows=

   If you follow this, you’ll find it is slow to respond. However, what I’ve found testing this evening is that the speed I’m seeing from the server on which I’m hosting the script is much much slower than I get when running it off my laptop at home! At home it takes about 15 seconds to respond, whereas on the server it takes around a minute! I’ve no idea why – I’ll see if Dave Pattern at Huddersfield has any ideas.

   In terms of what information about each course is returned by my script at the moment I grab this information from the course search results page on the UCAS website – and to be honest there is little point in me not returning any of this, as it doesn’t cost anything to do. I absolutely agree that a short description of the course would be useful – but the UCAS website doesn’t actually include any useful information here (if there is a link somewhere to more course information it is back on the institutions website)!

   With some extra effort I could grab the ‘mode of study’ (that is full-time or part-time), the level of qualification (BA Hon etc.) and the length (in years) from the UCAS results page if that would be of any use? I think that’s about it though.

   I know there is some work going around using a format called XCRI-CAP to enable institutions to publish their course catalogues in a ‘harvestable’ format, and there is a demonstration ‘aggregator’ which harvests XCRI-CAP data and offers a unified course catalogue – which would make it easier to do all of this – but at the moment only a handful of institutions are doing this 🙁

8. Thanks Owen.
   It seems bizarre that the server runs slower than the lap top, but that’s the fun of computers, isn’t it? Not entirely predictable.

   If L500 has so many rows I can see why it’s so slow.

   If it doesn’t add any overheads, then I agree it makes sense to just keep on returning what you are already returning in the XML. I was just thinking from the point of view as a possible student. Shame about the course description – some of the links to the courses do provided a bit of extra info, but it’s not consistent. If XCRI-CAP took off that would definitely help.

   Thanks.

This site uses Akismet to reduce spam. Learn how your comment data is processed.