Trip Database Blog

It’s that time of year, so my last post (not sure that this counts as a post) till around 9-10th August.

I’m leaving in 30 mins, but this AM I had the final specification meeting with the web-company regarding the latest upgrades to TRIP. These documents will be finalised for me to read on return. Once I’ve signed them off there’s around 2-3 weeks of development before I get a test version. So we’re looking at an early September deployment. To re-cap two main changes:

• Improvement to the search algorithm, much better results as well as the ability to arrange results by date (the most frequent request we get).
• Baby TRIPs. We should launch with 25 specialist search engines. Each specialist search will search the core TRIP content PLUS content specific to that speciality. This is not trivial content, we’re talking about 5-20,000 area specific content in each baby TRIP.

Bye for now…

An important question that is as relevant to Q&A as to systematic reviews. This Annals of Internal Medicine article (click here) attempts to answer this question. The important bit, for me was this:

“This survival analysis of 100 meta-analyses indexed in ACP Journal Club from 1995 to 2005 found new evidence that substantively changed conclusions about the effectiveness or harms of therapies occurred frequently within relatively short time periods. The median survival time without substantive new evidence for the meta-analyses was 5.5 years. Significant new evidence was already available for 7% of the reviews at the time of publication and became available for 23% within 2 years.”

A mixed bag to end the week:

• As someone who studied social networks I was fascianted by this NEJM article “The Spread of Obesity in a Large Social Network over 32 Years“. This was blogged on the Clinical Cases and Images site (click here).
• I’ve just broken my record for number of questions answered in a day – 18 – and still a few hours to boost that. I think I’ve overdone it.
• Talking of Q&A we recently extended our contract with the NLH for a further year, supplying up to 50 answers per week.
• Great news on the redevelopment front, we should be have the test version by the third week in August, with probable (?possible) deployment in early September. I’ve already mentioned the new and vastly improved search algorithm (click here for more details). The other development is the creation of a significant number of specialist TRIP search engines. More details when I get a moment…

Another glance at our webstats brings pleasing news. I’ve analysed how people get to TRIP, the top five being:

1. No referral – 72.02%
2. www.google.com/search 10.77%
3. google.co.uk/search 6.72%
4. search.yahoo.com/search 1.37%
5. ww.google.ca/search 1.01%

The top ‘referrer’ being ‘No referral’, this means people who have bookmarked the site or type in the TRIP URL directly. So last week, of the 110,000 searches, over 79,000 came from regular users of the site.

I suppose it’s difficult to know if that’s god or not. I suppose I’ve just been ignorant of the ‘loyalty’ of TRIP users. Over 72% seems pretty loyal. Of the remaining 28% how did they get to TRIP? Perhaps via searching for TRIP Database, perhaps searching other terms!

Given our server strain it might be wise to stop Google spidering our content, that will buy us some breathing space, one to ponder!

TRIP is always looking for good content to link to and this week three new publications have been added and are now searchable:

1. Clinical Evidence – Not new and have been on TRIP for a while! However, we have worked with CE to allow users of TRIP to get access to the first page of the relvent chapter, which is great news.
2. The Green Book – I’m not really sure why we haven’t added this before. We use the Green Book extensively in our question answering work. Better late than never!
3. Australia and New Zealand Horizon Scanning Network – I like this site….

With great content adding to great content and a significantly improved search algrithm just around the corner this is a great time for TRIP and it’s users!

The TRIP algorithm is the source of much pleasure and some embarrassment.
Why the latter emotion? Simply put, every now and then the system returns awful results? Not convinced, try this search menopause libido. The top result in a PRODIGY guideline on Parkinson’s.

I’ve posted before why this happens, but a refresher. In the case of the above guideline it is picked up by the search as it contains both terms. Even though it scores really low on ‘text relevancy’ it scores highly for being a recent document, being published in PRODIGY and being in the guidelines category.

For a fair few months now I’ve been trying to get a better system in place and yesterday I finally took ‘delivery’ of a test-bed for a new algorithm. This is a no thrills version on the site which allows me to see what the new algorithm can do. It does a number of things differently.

So how does that affect the above search – see the results below. Still needs a bit of tweaking, but it’s a vast improvement. The new system won’t be released until mid-September (with a number of other new features), we can’t wait.