Searching The Times online

I've recently been doing a survey of how well search works, or doesn't, across a number of British newspaper web sites. The quality of the results, and the usability of the interfaces, varies widely across the market sector. I thought I'd look at each newspaper I studied in turn, and give a little review of the features, usability and accuracy of their search facilities. I started with The Times online.

The Times default search is a site search, although the paper also offers an MSN branded web-search from the homepage.

The site search currently available is labelled as a 'Beta', and the paper is soliciting feedback on the service.

Times Online has launched a new search engine for content on the site and we'd like your feedback to make it even better.

Once you get to a search results page, the results are actually pushed down very close to the fold of the page. On my 1024x768 monitor with Firefox at full-size, I only get one search result displayed without scrolling.

The results feature the title of an article, an abstract, a date-stamp, specify whether the article was published in The Times, The Sunday Times, or specifically online, and also feature a percentage relevancy score.

As is often the case, I wondered about the usefulness of including a machine derived percentage relevancy score. In this example search, I searched for "BA 'price-fixing' probe", which was the headline of the main story on the homepage at that time, and the top result is the same story "British Airways in 'price-fixing' probe", yet it has only been graded as 49% relevant to my query.

The form to continue searching from a results page is quite intimidating, with several advanced feature selections included, although this becomes much clearer and more useful later on.

When looking at newspaper searches I performed four control searches across all of them, looking to see how easy it was to find the information I wanted. The controlled searches were "Blunkett resigns" which should give a choice of two occasions, "England Sweden result" looking for information on the recent 2-2 draw in Köln, plus "2002 World Cup final" and "September 11th", looking to see how far back archives go. I was really looking to see how well the newspaper searches retrieved original articles about these events, rather than articles that subsequently mentioned them in passing. In all cases The Times search results were heavily weighted to more recent stories, and did not give me the answers I was looking for, until I started using The Times Online search filter system.

On the right-hand site of the search results page on The Times site there are a series of filters. These allow the user to narrow down their search results in three ways: by navigational category, by time, or by the origin of their publication.

Next to each section is a numeral representing the number of results in that category. As a filter is selected, it reduces the number of results retrieved, and also builds a kind of crumb-trail navigation at the top of the page, indicating which filters have been applied.

The filters work on a deeper category structure than the one initially displayed. For example once I have filtered down to News from 2006, I am offered the chance to narrow the results to the UK or International news sections, and/or refine my query down to a specific month. Categories appear to only be displayed if they include results.

I was very impressed with this system. It helped me find the results that I wanted - and also helped me realise the things I wasn't going to find - the archives in the search appears to only go back as far as 2004.

I did have a couple of issues with the user interface of the filtering system. It actually took me two or three searches on the site before I even noticed they were there. I can understand not wishing to clutter the interface, but I felt that as a first time user I needed more of a sign-post to this excellent feature. Secondly, sometimes the text and numerals in the filter option were blue instead of the usual colours, and I didn't understand what that meant - it didn't seem to correspond with any selections I had previously visited. For example, in the results for my search for "2002 world cup final", the Property & Gardens section was highlighted, and there didn't seem to me to be an obvious reason - it wasn't, in this case, the category with the most results.

The Times offers a search tips page, which explains how to use the advanced syntax available from the search, the + and - operators, phrase match with "", and wild-card searching with *.

Although I found the relevancy for some of my queries to not be great, I found that the filter system really aided me in narrowing down the information. I thought the way the filtering worked was very intuitive, and on the whole I was very impressed with The Times online search.

Now, in probably a first for the currybetdotnet site, I'm going to use a <TABLE> tag to actually lay-out some tabular data rather than determine the layout of the site. For each newspaper I've made a list of features which their search services do, and do not, offer. Here is the summary for The Times online.

I'll be looking at the search on another News International owned newspaper site - The Sun - tomorrow.