Reblog…

My recent posts on male-as-norm language, To gender or not to gender and Default: male, did not go down well with everyone. I heard from one or two readers who thought I’d been a bit sweeping. Aren’t there, they asked, some cases where the unmarked term refers to a woman? And isn’t that just as […]

via Sauce for the goose — language: a feminist guide

Search history – your identity & your narrative

doyoubelieve

Screenshot of the interview on the Rhizome Blog.

This is a fascinating post on a project using one users search history over a three-month period and how that search history can not only identify you (e.g. via ego searches, searches of others close to you, etc.), but can also reveal your story. If you have an interest in privacy, security, and big data read this. Even if you think you have ‘nothing to hide’ you should think about the things you might keep private for whatever reason and how those might be used by businesses or others for good (e.g. ‘you might be interested in…’ based on your purchases/browsing/watching history) or evil.

Do you believe in user 711391? a search engine drama

ProQuest and systematic reviews – part 1

The plan is to write a series of posts about using ProQuest for systematic searching for systematic- and other reviews, this first post will be a brief-ish background to some of the problems.

ProQuest presents a number of challenges for undertaking systematic searching. Recently a number of pleas for help have been posted in mailing lists and on social media regarding time outs, discrepancies in search results, and general frustrations with using the ProQuest platform. I have also provided a lot of support in my institute over recent months to postgraduates working on systematic reviews as part of their thesis.

Proquest-page 3 display

An all too familiar screen in ProQuest. I have a folder with a couple dozen of these screenshots. I finally gave up taking them.

A bit of background

In the past, many information specialists and health researchers rarely encountered ProQuest because the offerings tended to focus on non-health/medical databases. This is still mostly true, but those who work in the NHS in England are now having to use ProQuest instead of OVID if they want to bypass the all-in-one search of HDAS as provided by NHS Evidence. (And avoidance of ProQuest probably shouldn’t be true of anyone doing public health research.)

Additionally, many questions are now cross-disciplinary and sticking with the usual suspects in the health- and medical-fields (e.g. MEDLINE, Embase, CINAHL, PsycInfo to name a few) is limiting. Much of the work that I have done almost always involves going beyond traditional health- and medical databases and using those that focus on education, social services, and even marketing. This has meant I’ve had to get very comfortable with a lot of different interfaces and the idiosyncracies of those interfaces.

I want to keep these posts short so I plan on doing three posts to start

  1. general notes about ProQuest (this post)
  2. the discrepancies many find in their search results
  3. running complex searches (this pretty much covers any search designed for systematic review)

If other topics come to mind, or I’m reminded of them from conversations, I’ll add to the series.

So onwards with some general notes about ProQuest.

500 error

Another familiar screen in ProQuest.

My experience

When I started to heavily use ProQuest in 2011*, and encountering serious problems using any of their databases, I got in touch with their help desk on more occasions than I can recall. (They probably all deeply inhale when they see my name pop up.) Problems encountered included:

  • time outs mid-search
  • time outs mid-selection of results
  • time outs mid-export
  • discrepancies with results day-to-day (in a way that didn’t make sense when you account for updates)
  • discrepancies in number of results during selection of results
  • discrepancies in number of results during export
  • a pop-up telling me I can’t do something I wasn’t trying to do
  • errors when trying to change the page display (e.g. from detailed to brief, or 10 results per page to 50)
  • errors when trying to go to the next page of results
  • selected results not showing up in the folder even when the folder shows there should be results in it (literally no results would be in the folder but it would display there were X results in the folder)
  • resetting which database you are searching when you move between screens

There were probably other issues, but these are what I recall off the top of my head. These problems often resulted in an additional 3-7 full days of work depending on how many of the databases I needed to search. This was of course infuriating because I could do 6-10 other databases in a single day (by doing translations in advance), but one Proquest database took at least one full day if the number of results didn’t exceed 1000.

Information from ProQuest

When I spoke to their help desk they were surprised to learn that anyone would use the databases for such complex searches, and that one would actually want to export hundreds, if not thousands, of results versus using simple searches and cherry-picking of results. They conceded that many of the problems I was encountering was because their systems were not set up for being hit like this which is why I was having so many issues with time outs. In the end it meant I had to do a lot of work arounds which only slightly improved the reliability and speed with which I could perform the searches. I’ll touch on these in a future post as I rarely have to use these nowadays since they have obviously improved both their hardware and algorithms to cope with complex searches and exporting.

They apparently looked further into these problems and discovered that something like 1-2% of the searches run in ProQuest are complex. This was in 2012/13 so this may have changed ever so slightly given the move to ProQuest by the NHS.

Working with ProQuest to improve things

I have actually spent a lot of time communicating with ProQuest about these problems over the years and on a few occasions have provided expert feedback and testing of new features/functionalities. While my interactions with various people at ProQuest have generally been pleasant and professional, I don’t think they entirely get it when I tell them that something intuitively doesn’t make sense or is so different from other platforms that it is confusing.I think they come at things from a developer perspective which really isn’t uncommon, but it is unfortunate for the end users.

searchhistory

I understand they are trying to make their product unique, but I think this is often to the detriment of usability both in terms of how people actually seek- and understand information, and in the functions they provide (e.g. hiding the search history under a weird icon – this is now remedied with search history being both under the weird icon, and in a text-based link under the search box (see image at left)). I think many of the new ‘improvements’ are anything but when it comes to using ProQuest for systematic searching, but I try to remember that, I’m guessing, at least 90% of the users are not using ProQuest for complex searches.

I realise this post doesn’t really offer you any practical advice about using ProQuest, but I wanted to provide some background information from where things were several years ago to where things are now, and that ProQuest is willing to engage on some level with improving their platform for the expert/systematic searcher. The next post will address some of the common problems, but I’d be interested to know if you’ve experienced the problems I mentioned above, and what kinds of questions you have about using ProQuest.

* If you’re curious, the databases I use most often in ProQuest are Applied Social Sciences Index and Abstracts (ASSIA), ERIC, International Bibliography of the Social Sciences (IBSS), Social Services Abstracts, and Sociological Abstracts. There are others I search too, but these feature in many of my search strategies.