Ovid's New Interface...Not Sure I Like It
*Please note this is an older post*
After talking with Ovid and looking at the product, I have changed my opinion. Please look at this more recent post about OvidSP.
I have been playing around with Ovid's new interface, OvidSP. I know my users will like, but I can't say that I will. My dislike has nothing to do with the general look and feel of the database, that is fine. I do not like how it searches. As a librarian, I want to know why and how things are searched for and retrieved. So it bothers me greatly, when I see odd and completely different results for searches.
To illustrate my point, here is a sample search on asthma and ragweed in children.
Here is what you get in old Ovid.
For search statement #1 I typed in asthma and ragweed and children, similar to how many of my users incorrectly search Ovid. I got 32 articles.
Search statement 2-7 are how I normally would conduct an Ovid search. Ragweed maps to Ambrosia which became a MeSH term in 2003. I did not explode Asthma as I was not interested in Exercised Induced nor Status Asthmaticus. I limited the search to English and All Child. I got a nice set of 59 articles in Medline 1950-Present.
This is the search I did in OvidSP.
First of all you don't see your search statements right away. You are defaulted to Basic Search. I don't have a problem with that because it is very easy to click on the link to see your search statements.
Search statements 1-3 are all examples of how library users usually search for literature. As you can see "And" and "In" make no difference in the results. Since people now search Google like this, it makes sense to replicate it.
Search statement #4 is identical to #1 EXCEPT the Include Related Terms box was checked. As you get different results. But, according Ovid the Include Related Terms box is supposed in BROADEN your search not NARROW it like it did here.
This was confusing problem number one for me. I thought perhaps something was wrong with the related terms and it was maybe "anding" Ragweed and Ambrosia together instead of "oring" it like I thought it would/should.
Search numbers 5-9 are how I would think a librarian would normally conduct an a search while checking the Include Related Terms box. As you can see in searches 8 and 9 when I started to combine Ambrosia, Ragweed, and Asthma together, I only got TWO ARTICLES. I hadn't even had a chance to limit it to children yet! These results are different from the 59 in old Ovid and a totally far cry from the 887 or 702 in OvidSP. I used OvidSP's check boxes and hit "Or" and "And" accordingly
This was confusing problem number two for me. I thought perhaps something was wrong with OvidSP's combining. So in search #10 I literally typed out 5 and (6 or 7). Ok that got me an odd results (more than my largest search term asthma) so I thought perhaps having the Include Related Terms box checked meant that it was also searching the numbers as numbers instead of just as search statements.
Search number 11 is my attempt the search 5 and (6 or 7) typed out without the Include Related Terms boxe checked. Still odd results. This became confusing problem number three for me.
I thought perhaps I needed to add the # sign in front of the numbers, similar to PubMed. Nope. That produced a nice red error message (not pictured) which told me the wildcard character is not supported. Hmmm. This is confusing problem number 4.
Finally for search number 12 I decided to just type in Asthma and (Ragweed or Ambrosia) which I limited English and Children. I got 103 results. Which is totally different from the number of results I got in old Ovid. Which brings me to confusing problem number five. Why are their completely different numbers for every single search result between old Ovid and OvidSP? Does this bother anybody or is it just me?
I would like to know why we have different results between old Ovid and OvidSP results. I want to know why the combine feature seems completely broken. I want to know what we are getting or what it is looking for when we click Include Related Terms, because it DOES NOT ALWAYS broaden a search as the imply.
I would love to learn more about OvidSP and how we can get RSS feeds and TOC for online journals (all journals, not just Ovid online journals) but until I have answers as to why I get drastically different search results I can't really recommend using OvidSP. I will cling to old Ovid until February then be forced to used PubMed. I don't know. But I would sure like some explanations.
Anybody have any ideas or thoughts as to why this works like this? I would love to hear them because frankly I am stumped.
posted by The Krafty Librarian @ 2:18 PM
8 comments
8 Comments:
I have actually found SP to be quite an improvement over the previous version. I do have some beefs which we will hopefully resolve when we change our default settings at the admin level. For now, I've just being clicking on the OVID Syntax tab which allows you to search MEDLINE just like you could in the old interface.
What I really like is the ability to view the abstract without leaving your search results. What I'm not liking is the fact that you have to expand the limits and the search history in order to see them at all.
I like the layout, like seeing the abstract within the search results.
What I don't like is having to click on the Ovid Syntax tab to do a regular "librarian" like search.
I especially do not like the discrepancy between search results while searching OvidSP. It doesn't make sense to me that checking on the Related Term box yields fewer results.
I'm hoping we can set it to default to the Ovid Syntax tab on opening, just as we've always had old Ovid defaulting to the Advanced search. Ovid Syntax seems to give control of the search as we need and are used to. Who knows what the basic search is doing?
I don't like the Results Manager down the left side. You have to move your way down it making your selections, and then scroll back up to the top to do the action. Not good workflow, but I guess OK if you just want whatever the default is.
Also don't like having to click to expand the limits or see the search history. Perhaps they are things that can be set as defaults too? Being able to adjust the number of items seen per page with only one click is great though.
I'm off to an info session on SP on Monday. Perhaps they'll answer some of my questions? We will certainly be sticking with the Gateway interface for as long as possible at this stage.
What's even more confusing about including related terms is that you seem to get not only fewer articles, but different results. Here's a thoroughly hypothetical example:
1) dogs and cats [no related terms] 350
2) dogs and cats [including related terms] 325
3) 1 and 2 174
4) 1 or 2 520.
The numbers are made up, but that's the gist of it.
When Ovid first came out with their now "old" interface, the librarians' comments at the Cleveland Clinic mirrored the Krafty Librarian's comments on the new OvidSP interface. The reason we subscribe to Ovid is for its advanced search capabilties and ease of manipulating search statements.
Although we haven't investigated it yet with OvidSP, we set the "default" interface of the "old" system to the Advanced setting. This would be equivalent to making Ovid Syntax the default for OvidSP.
My educated guess (without discussing with my colleagues) is that we will continue to do this with OvidSP. Users always have the option of changing back to the Basic interface.
From a librarian's viewpoint, I applaud Ovid & PubMed for creating user-friendly front end interfaces to navigate the literature. Since many users don't bother to learn how to search, their chances of getting better search results has just increased.
Now we have 2 products, Ovid SP Basic and Pubmed, that "think" for our users.
At my library, we’ve found the “Related Terms” feature very confusing. So far, it seems to narrow the search at least as often than it expands it. (See, for example, "end stage renal disease," 976 with related terms, 1381 articles with no related terms; "diagnostic imaging," 504 with related terms, 646 with no related terms).
I’m glad you can see what related terms OVID SP is mapping to, so at least you can see what’s missing: it’s disturbing, for example, that ambrosia and ragweed are not considered related terms, and I don’t see how a normal user would know they should be without using the Search Tools tab (and I don’t see my users doing that). Nor does "related terms" explode the term: the only related terms for "diagnostic imaging" are "imaging diagnostic" and "diagnostic images not elsewhere classified," so no PET scans, CAT scans, tomography, or even "diagnostic images," which is what I think my users would expect to get.
However, it does seem to be a work in progress: we noticed that enoxaparin started mapping to lovenox some time in the middle of the day yesterday, for example.
I emailed Ovid asking them how the "Include Related Terms" works. Here is their response:
"In OvidSP, the 'Basic Search' uses natural language processing (not MeSH terms), so the searching would be quite different.
NLP will return up to 500 citations, often more, and they will be ranked with the most relevant at the top, and also those with full text links at the top. The results will highlight the search term in red.
When you click the checkbox for 'Include related Terms', OvidSP will expand the search to include all synonyms it can identity.
Once your search results display, if you look to the left pane, in the 'Search Aid' section, you will see the search term you entered with an arrow to the left. If you click the arrow, the search term will expand to show all the related terms included."
Has anyone noticed that there are a lot of "stop" words? For example, if you type in evidence based nursing, the system only searches on nursing. The same with evidence based practice - you only get practice. I tried using quotation marks and parentheses to no avail. However, hypenating seems to get you phrase searching. evidence-based-nursing. Yikes!
Post a Comment
<< Home