Trip Database Blog

My post of two days ago was slightly premature! After using the new algorithm I started noticing some strange results. So much digging around with the maths helped me discover a variable I had overlooked – inverse document frequency. After understanding this concept, a bit, I have now used it to our advantage.

The net result being the algorithm is now, without doubt, a significant improvement on algorithm we launched with – just 7 weeks ago. If the first algorithm was version 1, the version earlier this week must have been version 2, then I guess we’re now at version 2.1.

Happy searching and as ever please let us know if you conduct a search and receive strange results!

Internet explorer is now available for download.

It looks very good and seems intuitive and so far no complaints. An added bonus is the embedded search box – which I’ve defaulted to search TRIP. It works in the same way as the recently released firefox extension. (see top right corner of screen shot below).

Following on from my earlier post the new algorithm has now been implemented. Our testing regime returned results quicker than anticipated. While still not perfect it is a definate improvement on the previous version. We are not assuming that we cannot tweak the algorithm further. We will continue to monitor performance and anticiapte further testing soon.

The new system is gradually being rolled out. No massive changes and whether users will notice much difference is open to debate. I’m sure some will but the majority probably not.

The system works exceptionally well (my view) when a category is selected (e.g. systematic reviews). However, the first set of results, where all the categories are mixed, is more problematic. Generally, it is very good. However, in a small set of results <10% the results trouble me. They're not overly bad but they could certainly be better.

My concern is that users of TRIP do not, necessarily, filter the results by category – so they get the mixed results. So two possible directions – and I’d welcome input from the readers of this blog:

1) Remove the first set of results, so users are forced to select a category to see results.

2) Continue to work with the algorithm and be happy that for the vast majority of searches the results are pretty good.

Over to you……..

An interesting paper I found while looking through the rapidly expanding UBC HealthLib-Wiki. This paper was published in the American Family Practice Management journal – click here to view the paper.

Our first month of being free has seen a huge surge in use – as we had hoped. The total number of searches of TRIP, for September, was 132,538 – amazing!

The full September zeigeist with the Steve Irwin influence at number 1!

1. Stingray Envenomations (eMedicine) – viewed 474 times
2. Failure to thrive (ganfyd) – 368
3. Exercise therapy for multiple sclerosis (Cochrane) – 289
4. Growth reference charts (RCPCH) – 276
5. Increasing the cultural competency for healthier living (NHMRC) – 273
6. Failure to thrive (GP Notebook) – 251
7. Children and healthy eating (EPPI) – 210
8. Buprenorphine transdermal patches (Norspan) for chronic severe pain (NPS) – 198
9. Management of patients with dementia (SIGN) – 182
10. Hypertension (PRODIGY) – 179

An idea in early planning stages – TRIP Answers.

Our view, at TRIP, is that if someone searches TRIP and can’t find what they’re after, then TRIP has let them down. So what can we do?

Our idea is to allow users to record their clinical questions on the TRIP site; the ones TRIP has failed to help answer. What’s the point?

Well we see this as having two main purposes:

1) It will allow us to identify themes in where TRIP is letting users down. Once we know where these are we can seek suitable, evidence based, resources.

2) We’ll try and get as many answered!

The process we’re planning is to have a banner on the results page – inviting users to go to TRIP answers if the site has let them down. They will then be presented with the following options:

1) UK health professionals get directed to a relevant Q&A Service
2) A non-UK health professionals get sent to TRIP Answers
3) Non health professionals get sent to a relevant ‘answers’ site (e.g. Microsoft QnA, Yahoo answers)

Those sent to TRIP Answers will record their questions and we will attempt to ‘answer’ as many as possible. We do not have the resources to answer many questions and therefore have decided to return ‘Personalised results pages’. Our view is that a user of TRIP expects a page of pertinent results. So TRIP Answers will return a set of highly pertinent, human derived, results. Those unanswered will be recorded and used for analysis.

An additional feature is that we will prioritise questions from the ‘developing’ world. To that end we have started discussions (still at very early stages) with the Global Healthcare Information Network. Our thinking around TRIP Answers has coincided with the launch of their new campaign ‘Healthcare Information for All by 2015’.

If there are any willing volunteers out there who want to get involved, we’d be delighted to hear from you.

Watch this space for further developments.

http://searchenginewatch.com/showPage.html?page=3623646