Thursday, June 24, 2010

Searching Newspapers Using Keywords

I was trolling through the online newspapers at and found a "better way" to search for my people. This may fall into the "you already knew this, but I just found out" category, but that's OK - there may be some users in the vast (yeah, right!) Genea-Musings readership that could use the information to make the searching task easier or better, especially for common names.

In the past, I have always put first name and last name in the search fields when I've searched Newspaper and Book collections on I received an email today that said "have you looked at keyword searches?" Ummm, no, not really.

I started out on the Newspaper Collection page on - and there was a helpful link to a Juliana Smith article from 2009 titled "Searching for Newspaper Stories." This article suggested that:

"Typically when searching for a person, it’s best to leave the “Match all terms exactly” box unchecked. This allows us a little wiggle room, but when searching for common terms, it’s easy to become overrun with results. For keyword searches it’s helpful to keep that “Match all terms” box checked. (Just bear in mind that when you go back to searching for people, you’ll want to uncheck this box again.)"


"Remember that if you want to search for several words together, you can put them in quotes which will tell to look for that exact phrase (e.g., “cholera epidemic”)."

Read the whole article - it helps! Here's a screen shot of the article page:

To illustrate the differences between a Name Search, using names in the first and last name search fields, and using a Keyword Search, I'm going to use my grandfather's name, Frederick Seaver. For the Name Search, I entered First Name = "Fred*" (because he went by Fred and Frederick) and Last Name = "Seaver." I clicked on "Restrict to Exact" for both names. The Search fields are shown below:

I received 341 matches from my Name Search, in a number of newspapers:

The Fitchburg (MA) Sentinel was one of the local newspapers for Leominster MA, so I clicked on that link, and saw 53 matches:

The results page above has snippets showing one or more matches on a given page. When you look at the matches produced by the Name Search, you see that not all of the matches are for Fred* Seaver. Some matches are for a page with Seaver in one place and Fred* in a nearby place on the page. Some of the matches found using the Name Search are for Frederick W. Seaver, which I definitely want to find.

For the Keyword Search, I entered "fred* seaver" in the Keyword field of the Search box:

I received only 138 matches in fewer newspapers from this Keyword Search:

I clicked on the Fitchburg Sentinel link again and saw:

There were only 26 matches for the "fred* seaver" Keyword Search, and all of them were for my "Fred* Seaver" or his wife listed as "Mrs. Fred* Seaver." However, none of the matches for Frederick W. Seaver are listed. Drat! I did another search for "fred* w seaver" in the Keyword Search and there were 16 matches.

The lessons learned here include:

* a Keyword Search, when used with quotes around a person's name, will return exact matches for the search terms.
* a Name Search will return more spurious or extraneous matches because it finds the first and last names in proximity to each other on the newspaper page.
* If a person had a middle name or initial, then a Keyword Search should be performed for the name with and without the middle name or initial. By the way, a wildcard for the middle name or initial doesn't work - it violates the Ancestry wild card rules).

When should the user use a Keyword Search? I would advise that it should be used in databases indexed by Optical Character Recognition - Newspapers, Periodicals, Books, Stories, etc.


Michelle Goodrum said...

Thanks Randy! I didn't know this. Already I can think of several ancestors where this method will make newspaper searches much easier.

Jennifer said...

Very helpful, indeed. I would add a few tips: use the names of relatives with uncommon surnames when searching for information on relatives with common ones; and do manual searches of local papers around important dates like births of children, marriages, deaths, and wedding anniversaries. OCR is great and helps a lot, but it is unfortunately still flawed enough that it doesn't replace a manual search in all contexts.

Jo said...

Thank you, Randy - I didn't know about using names in the Keywords box, and get disheartened trawling through the results which show the two names on different parts of the page. Very useful info! Jo