Trip Database Blog

An early review of TRIP (click here) was interesting to read, at the time. Probably being biased, I felt it was a lazy review, although the concluding comments were good:

“TRIP offers a friendly interface and quick access to the evidence (particularly if the search is limited to titles) with user friendly organisation of search results. We recommend this resource for those seeking pre-appraised evidence, reviews, and guidelines.”

One issue they highlighted was the lack of a synonyms function. So, we set about creating one! To do this we took a list of the top 8,000 search terms used in TRIP (covering around 70% of all search terms) and manually produced a list of synonyms. This was time consuming and hard work. However, when produced it worked pretty well.

Probably 3 years ago we started to use the metathesaurus function of the UMLS system produced by the American National Library for Health. This was an automated project facilitated by the MMTx program (a java-based, public, approximation to the full-blown MetaMap system). This complimented the manual synonym function. Unfortunately, this was too slow, so we had to abandoned it. However, we’ve recently had some interesting news from those behind MMTx. The full program is likely to be released and will be considerably quicker than MMTx. So we’re hoping to re-introduce this as soon as it’s available. This should add an extra layer of synonym functionality

Anyway, why am I blogging this? Well I’ve just finished the latest edit of the manual synonym look-up table. It’s taken an awful lot of man-hours spread out over two months. We’ve weeded out ambiguity, added new synonyms – all to make recall better!

This will be rolled out with the current round of updates.

It’s been exceptionally busy for us, although I doubt we’ve worked as hard as our servers. Due to this workload I’ve been a bit slack in updating on our various projects, so here goes.

TRIP Update
For a number of reasons, mainly me changing my mind half way through, the update is delayed and will hopefully be released mid/late April. It seems to be dragging on but we’ve had to make some changes. So, the final list of changes are:

• RSS feed. Each search can now be outputted as an RSS feed
• Mark PDF records. Some people really don’t like PDFs – we’ll now mark them for you on the results page.
• Conclusions. A number of documents have conclusions and our system has been programmed to grab them. Users, if they wish, will be able to view the conclusion by placing their mouse over the ‘Conclusions’ button
• Snippets. Most search engines generate snippets of text to help improve the context of the results. These have shown to improve search recall significantly, hence us adopting them. From a personal perspective I’m not convinced, so I’m trying to make these ‘switch offable‘. It seems the pragmatic decision to me.
• Advanced search. This has been significantly improved in terms of both functionality and accuracy (we’ve been experiencing a few problems on that front).
• Redesign. As I’ve mentioned previously our revenue from Google Ads has been woeful. We’ve been communicating with Google to improve thing. The first stage of which is to move the adverts to a more prominent position. It’s a shame, but we need money to further improve! Coupled with that is the inclusion of snippets. The net effect of this is to take up more space on the screen (apparently it’s called screen ‘real estate’). So we’re going to redesign the screens to free up more space.

In addition to these immediate changes we’ll be producing a e-newsletter for the 2,000+ registered users on TRIP sometime in the near future. There are some great examples which help keep communication going between organisations and users. The blog is one way, and is read by a small number of people (around 250 per week), but an e-newsletter gives us other options.

Gwagle

The alpha is going very slowly. We’re not 100% sure why although we believe that the alpha-testers are too busy to go to a site, add stuff and get relatively little in return. For Gwagle to be useful we need to have an active community. We’ve decided this route isn’t the most useful. Part of our re-think is to integrate the site, much earlier than initially anticipated, with TRIP. TRIP gets large amounts of traffic (nearly 150,000 searches last week) and so if we can make using Gwagle meaningful to TRIP users we’re sure things will develop more meaningfully. We’ll keep you posted on those developments. But, if you fancy giving Gwagle a go, let us know via ‘Contact us‘.

Research on Q&A

Nothing major to report except that we’ve started some question analysis. We’ve been wanting to for ages. With around 5,000 genuine clinical questions from NLH Q&A and 2,000+ from ATTRACT that’s a large pool of data. The big downside is that writing is not hugely motivational for me, that goes for writing articles, writing up my PhD (never finished, so in good company there) and writing research grants. As my supervisor said – the only way to write a PhD is to write a PhD. His obvious ‘flair’ for motivation missed me 😉

“As articulated by Chris Sherman, social search is information retrieval, way finding tools informed by human judgment. Social search is people helping people find stuff, including allowing users to ask plain English questions and get plain English answers. It’s collaborative content harvesting, directory building, voting and ranking, sharing, tagging, commenting on bookmarks, web pages, news, images, videos and podcasts. Social search reflects the wisdom of crowds.”

Read more here.

This is not an April Fool, but TRIP has broken the half a million searches per month. In fact, for March, we were searched 515,191 times (that’s 3,873,391 page views). We were helped in breaking this mark by a significant upswing in searches towards the end of the month, most over 20,000 searches per day, and two over 25,000.

The 515, 191 gives an average daily search of 16,619. So, if we maintain seach figures for 20,000+ searches, we’ll be having another, predictable, post on 1st May!

ATTRACT has been running since 1997. For years the questions were added onto a webpage I ‘designed’ and created using MS Word. However, around 2001-2 (can’t remember) we created a new website (click here to view) which was lovely – at the time. However, I’ve been unhappy with the look for a fair old while. I’d prefer not to pay for a new service – so I thought I’d try creating one using blogger.

After literally ten minutes I had a site that looks significantly better, had broadly similar functionality and has a better search. Not bad for free. We’re not sure if we’ll adopt this, as it’ll need to be ‘user tested’. If that happens we’ll move it to a better domain name.

To view the new site click here.

When we were a subscription service we were searched between 20-25,000 times per month. Yesterday, we went about the 25,000 searches in a day. I suppose that equates to a 30 fold increase in traffic in 7 months.

Citizendium, an alternative to wikipedia has launched as a beta-service today.

The Citizendium (sit-ih-ZEN-dee-um), a “citizens’ compendium of everything,” is an open wiki project aimed at creating an enormous, free, and reliable encyclopedia. The project, started by a founder of Wikipedia, aims to improve on the Wikipedia model by adding “gentle expert oversight” and requiring contributors to use their real names. We have over 1,000 articles and hundreds of contributors. But we will avoid calling the Citizendium an “encyclopedia” until the project’s editors feel comfortable putting their reputations behind that description.

An interesting blog post that reports that TRIP and Gwagle are helping to usher in the semantic web by having a narrow focus.

My interpretation of this is that, as people using clinical terms there is a significant loss of ambiguity. In Google, if you type weight, it might be about the weight of a whale, yacht, car, person, building – pretty much anything. However, in TRIP or Gwagle, a search for weight is pretty much likely to refer to the weight of a person (or other biological entity) or something like low-molecular weight heparin.

I don’t think I’ve ever discussed traffic on this blog, always a first time! This story discusses a peer-to-peer mobile network to pass on details of road conditions between users.

This is another example of a network enhancing ‘experience’. The whole being greater than the sum of the parts. Due to the work on Gwagle I’m very interesting in ‘enabled’ networks enhancing whatever outcomes the ‘actors’ are involved in. With Gwagle the actors can enhance knowledge acquisition, information support etc. In the case of the traffic network it could enhance the driving experience.