quick summaries of documents and Web pages (01/03/2006)
Summarize is one of a number of products from Corpora aimed at helping you sift through information. If your job or hobby involves wading through a lot of text, its claim to produce 'accurate summaries of long documents at the click of a button' (which you'll see on the Corpora Software home page) is enticing. In practice, though, we didn't find things to be quite that sweet.
The 'click of a button' part is certainly right. Not only do you get a standalone application which can work on documents you locate with a file browser, Web sites you identify by entering their URL and text you paste in, you also get plug-in toolbar icons for Internet Explorer, Microsoft Word and Outlook.
If you want to summarise what you are currently viewing in any of these applications, just click the icon and the summary is produced. Summarize works with plain text files, Word documents, PDFs and HTML files, so it is quite flexible.
You have some control over how summaries are made. You can specify a number of words, or a percentage of the original document length, and you can set up 'profiles' which are collections of terms that are used to slant a summary in a particular direction. It's also possible to get the software to extract what it thinks are the key terms in a document, listing them in a side bar and, if you so choose, highlighting them in the summary.
You can save and e-mail summaries and view them in the context of the original document; you can set the fonts and colours used so that the summaries and keywords stand out well for you visually.
When it comes to the summaries themselves, though, things can be a bit hit and miss. For example, when summarising a Web site the software can't tell the difference between text used in menus and text that matters. If you were glancing at a site you'd see that immediately and ignore the irrelevant stuff, but Summarize may include it in a summary where it looks out of place, can confuse reading and wastes words.
We also tried Summarize on a couple of long reports that were saved as PDF documents. These included tables and graphs. Summarize ignored the graphs, and when it thought a table was relevant extracted information from it, but just listed row and column contents rather than putting one beside the other in their original context. Without the laid out original to hand it was often hard to see what a table was about.
Where Summarize worked best was on long texts that were comprised entirely of words (no graphs or charts). It tended to provide enough information to get the gist of what was going on, and the profiles seemed to do their job of hunting for the key words we wanted to seek out.
In the end, we're really not sure about Summarize. It does what it claims to in that it produces shortened versions of long documents, but the quality of what you get can be variable. Think of it as a tool for use in certain circumstances rather than one to be applied universally, and you'll probably get the best out of it.
Buy Corpora Summarize securely online at a bargain price
£29 inc. VAT
Corpora: 01483 443000
