ECCO on JISC and Contextual Word Searches

After Dave’s most recent post on keyword searches, a short discussion followed noting the (currently) UK-only ECCO, via JISC, supports conceptual word searches while the US ECCO only supports keyword searches. To be honest, I was unaware the US had not upgraded to the JISC version and was asked (since I’m a student at the University of London) to post some screenshots of the differences and perhaps write a post on it.

I am fortunate enough to have both access to the JISC Beta and the older version of ECCO–my individual school (Royal Holloway) provides JISC while the Uni. of London Library, Senate House, has the version with which most of you are familiar. There are several key differences between the two that result in completely different search results, as you will see throughout this post. NB: I’ll be providing a lot of screenshots here, so I’ve placed them after the jump.

If you’d like to enlarge the images, simply click on them and you’ll be redirected to the larger version.

First, the ‘home screen’:
Image

I already included my search parameters, but this will give you a sense of the home screen. I’m using ‘Drapier’ in reference to Swift’s 1724 pamphlets, The Drapier’s Letters. I set the minimum year to 1724 (as I’m not interested in any ‘drapier’ before Swift) and the maximum to 1730 so I may track popular responses to the halfpence controversy both during it and shortly thereafter. I also set my results to 25, the maximum, so I don’t have to keep flipping through pages. The option to search simultaneously on EEBO and the British Library’s 19th Century Collection may be helpful to some of you, but for me it is not necessary, so I unchecked those options (which are checked by default).

 Image

Here are my search results. Those parameters, by themselves, retrieved 360 results. Let’s assume I want to make that number a bit smaller. To the left is a word cloud which provides suggestions that could narrow down my results. The larger the word, the more prevalent it is within those 360 results. I chose the middle-sized phrase, ‘Woods halfpence’, at the bottom of the cloud, as my next step.

Image

I now have 6 results. Because the sample pool is so small, there is no longer a contextual word cloud. However, if you look along the top, you can see my results are filtered by both my original parameter (1724-30) and the new one provided by the cloud (as evidenced by the cloud icon next to the phrase). If you want to remove a parameter, it’s as easy as clicking it.

Image

Hovering your mouse over the phrase will strike through the text and clicking will eliminate it. After moving the date constraints, my results look like this:

Image

I now have 120 results, and the cloud returns with different suggestions (e.g. Dr. Jonathan Swift was not in the Drapier-cloud), as well as different results from my initial search. It is worth noting, however, that if you want to alter the years of your search without having to go back to the original search or removing them completely, you can through the left sidebar.

When clicking on a document, this will open:

Image

The tools along the top right allow you to zoom, view one or two pages, and download the document. Unfortunately downloading is more of a hassle with JISC, as you have to wait in a queue before you can download. It’s usually not an issue, but you’ll get the rare moment in which you wait 5 minutes for your article (as petty as it sounds, it’s irritating!). However, the search-within-this-document feature is great and, much like ctrl+f on a Word document, it lists the pages where your search term appears with an excerpt of its sentence to provide context.

That said, since Dave’s post suggested concerns over searching, let’s hop over to the old ECCO. My initial search will have the same search terms as my first: ‘Drapier’ with years ranging from 1724-30.

Image

The results are completely different. First, there are only 62 provided rather than the 360 given by JISC. Second, the results are mostly irrelevant to the halfpence controversy and Swift (I have it sorted by author here, but a substantial portion of these texts are unrelated). Assuming I’m familiar with the circumstances of the Drapier’s Letters, I have the option to narrow down the search pool by adding another parameter. I add ‘Woods halfpence’, as the JISC cloud had previously suggested. As a side note, ECCO in general doesn’t like apostrophes, so–as JISC did–I opted not to use the apostrophe for ‘Wood’s’.

Here are the results:

Image

Or shall I say, lack of results?

Bill Blake’s ASECS talk, as Stephen Gregg generously summarised, on keyword searches becomes relevant here. Gregg suggests utilising broad search terms on ECCO is frequently unsuccessful. What JISC does, then, is take those broad search terms and gives them opportunities to become successful and fruitful searches. One can see the various directions a simple term such as ‘Drapier’ may take on JISC. This exercise proposes JISC’s ECCO is a tool for discovery in various forms, whilst, unfortunately, Cengage’s requires you to know your search terms precisely. That is, admittedly, sometimes a boon; I have found things on the Cengage ECCO that I could not find on JISC, but those moments are rare and are usually the result of having an exact title for which I am searching.

–Kelly Centrelli

About these ads

3 responses to “ECCO on JISC and Contextual Word Searches

  1. This is fascinating —- we had to fight to get basic ECCO at UPEI a few years ago —- I wonder if we’ll have to fight for the upgrade, too. Thanks do much for this really clear and detailed post —- much appreciated.

  2. Eleanor Shevlin

    Many thanks, Kelly, for this extremely useful post. The JISC platform seems to be a relatively new development, but I had been seeking a user since my EMOB post JISC’s Historic Books: Searching EEBO, ECCO for meaning a month ago, so I was delighted to see that Dave had spurred you to post. I have posted some comments to larger issues raised by semantic searching in the comment threads to my EMOB post. While the JISC platform can certainly be seen as an upgrade, it is not something the publishers offer and does not seem to be available outside the UK (I think the previous comment is from someone in Canada–University of Prince Edward Island?). Instead, the UK through JISC funded the meaning-based search capabilities through its JISC Historic Books initiative (combines EEBO, ECCO, and the new 19th-Century Collection) and has offered it to its institutions of higher education..

    The “search-within-this-document feature is great and, much like ctrl+f on a Word document, it lists the pages where your search term appears with an excerpt of its sentence to provide context” that Kelly mentions is also available through Cengage searching.

  3. Anna Battigelli

    Ever since Eleanor’s post on JISC a while back, I have been wanting to know more about how JISC’s word cloud gets assembled. Knowing that will help us understand more fully just how “relevant” JISC’s “relevant” terms are. It would be great if someone from JISC could explain this more fully. Thanks Kelly and Eleanor for very interesting comments!