The official Google blog just announced Google's More Search Options feature that I blogged about back in March. Interestingly, it's no longer working reliably for me.
But it also rolled out Preferred Sites, where you can specify specific URLs that you want ranked more highly. Search results from those sites are flagged so you notice them.
Just learned about Search Cloudlet from Gary Price at Computers in Libraries. It's an extremely cool Firefox-only add-on that displays a tag cloud of the results from a search in Google, Yahoo and Twitter. It delays the search results page a bit while it generates the cloud, but it's a cool way of seeing most frequent terms in the search results.
You can also display a cloud of the sites retrieved - a nice way of quickly finding a few key sites.
Yet another reason to switch from IE to Firefox.....
Google has rolled out several changes to its search results page - some more impressive than others (maybe I'm just getting jaded). Among the notable improvements:
Larger snippet if you use >3 search terms. In the past, the search results would include a short "snippet" of text, usually no more than 2 or 3 lines, showing your search terms. However, when your query has more than 3 words, a 2-line snippet likely won't even contain all your search terms. So now, if your query is longer than that, the snippets are sometimes 4 lines long. I wish it were more consistent, but at least it's an improvement.
Initial stab at semantic search. To grossly oversimplify (my specialization), semantic search looks for occurrences not only of the words you provide but also words that are similar in meaning. In the past, Google has addressed this with a simple list of synonyms for common words. Put a ~ before a common noun, and Google will search for similar words. If you search for ~moon, you'll retrieve pages that mention the words lunar, Apollo (referencing the Apollo lunar space program), and rising. Nice theory but not so good in practice.
Google has gotten more ambitious, and now attempts a limited version of on-the-fly analysis of search results to suss out words that have similar meanings. The results are still a bit squirrely, but I love the fact that they're working on this. While Google doesn't automatically include results for other search terms, it does suggest alternate concepts at the bottom of the search page, and they're not bad. For a search on quantum physics, for example, the suggested terms included quantum theory and equations, as well as time travel, law of attraction, and what the bleep. A search for Barack Obama gave me suggestions such as Hillary Clinton, John Edwards, and phrases such as barack obama muslim, barack obama middle name, and barack obama smoking.
Obviously, one of the biggest challenges with semantic searching is the computing power needed to data-mine the search results and calculate related terms, on the fly, in a matter of milliseconds.
Search refinement tools. A hiding-in-plain-sight feature was introduced recently -- the "search tools" pull-down option.
When Google is able to identify related concepts through its rudimentary semantic search, the search results page will include a pull-down link that easily enables you to filter your results in several different directions.
Most of the limit features are what you'd find from an advanced search, but I like having them all there in front of me. And a few of the features were initially rolled out in Google Labs -- the display of images from each search result, for example.
Three items near the bottom of the Search Tools list are the most interesting. Click the "More Text" link and you get..... you guessed it, more text in each snippet. You can switch the search results page display from the standard format to a timeline (also initially introduced in Google Labs).
And see that link labeled "Wonder wheel"? Click there and all the similar phrases, gleaned from Google's semantic search, are displayed in, well, kind of a wheel. In essence, it lets you explore the "trail" of related terms. While this is nothing new to anyone who uses a more visual, clustering search engine, but it's nice to see Google getting on the bandwagon.
So, for example, with my quantum physics search, I clicked the Wonder wheel, and here's what I got:
When I clicked the link at the bottom for "quantum theory", a new Wonderwheel is displayed with the terms related to quantum theory. Click on any of the links in this wheel, and you'll see related terms for that word, and so on.
Oh, [insert bad word here]. Furl.net has just been acquired by Diigo. And I was JUST THIS MINUTE getting ready to Furl my umpty-billion URLS for my upcoming talks.
From all I can tell, it's got the same or more functionality. That's not the point.
They did not consult with me, personally, before making this switch. I just sent in all my handouts, that touted Furl.net. And right this moment, I am waiting for them to do whatever they do behind the scenes to move my bookmarks over.
This transition better go smoothly, or someone's going to get hurt.
We've heard for years about the invisible, hidden, deep or dark web -- that portion of the web that resides in databases and other formats not accessible to search engine spiders. One of the latest attempts to make this content searchable is DeepDyve, formerly known as Infovell.
DeepDyve offers a federated search of a wide variety of content, some of which is free and some requires payment to the publisher. Sources include MEDLINE, Clinical Trials, PubMed Central, BioOne and SAGE Publications - some heavy-hitters. Searchers can limit their search to various slices of this content, such as legal, business & finance, life sciences, or clean tech and energy.
What sets DeepDyve apart is how it handles text and queries. The founders of the company worked on the Human Genome Project prior to starting Infovell, and they approach search differently than most. For DNA research, scientists must take small snippets of genetic material and match them together. To the information processing algorithm, the actual meaning of the terms didn't matter; it's looking for patterns. DeepDyve takes that same language agnosticism to offer a very different approach to textual analysis than the usual syntactic analysis.
Like Yahoo's Y!Q, DeepDyve is designed to work with a block of text rather than just a few search terms to find useful information. Given its founders' background, it's not surprising that DeepDyve can even handle searching for chemical formulas; in fact, you can point it to a collection of articles you have selected from a prior search and say, in essense, "These are all helpful. Find me more like them."
DeepDyve requires (free) registration for its basic version; the Pro version, which costs $45/month, includes additional search and display features but no additional content.
Ever wonder how papers -- around the country or around the world -- are covering a topic? The Newseum has a web page with the front pages of over 650 newspapers from 63 countries. For those of you who haven't heard, the Newseum is an interactive museum focused on the news industry, located in Washington DC.
Click on any of the images, and you get the image of the front page, along with a link to that newspaper's web site, if you want to read more.
They also offer a bit of an archive -- you can see a collection of front pages for selected historically significant events, but it's fairly limited. I'd use this site more for getting a quick look at how a news event was covered in the next day's news than for in-depth research.
If you want to search the best of the legal web (OK, I mean that portion of the web that has law-related content), a nice tool I use is the Legal Research Engine, created by the Cornell Law Library.
You can limit your search to legal research guides, the best of the legal web, and/or academic blawgs from law professors.
What I really like about this site is that it's clear, easy to use, and a great example of what any information center or independent info pro could do for a client group. These are "just" Google custom search engines -- very well-designed, well-tuned search engines, but the capacity is there for anyone to build a similar resource on a topic they are equally familiar with.
One of the challenges of using Wikipedia is that your topic may be mentioned in a number of articles, and Wikipedia's interface doesn't make it easy to jump from one article to another. Wikipedia Roll is one more tool for enabling readers to explore the full range of a concept in Wikipedia. It's intuitive and easy to use, once you get used to the elephant that periodically runs across the screen.
First, note that I'm not primarily a sci-tech researcher; I've been a business researcher almost all my adult life. That said, WorldWideScience.org sure is cool. It offers a one-stop metasearch in 50 national science portals, ranging from the Indian Academy of Sciences to Science.gov, Canada Institute for Scientific & Technical Info, CERN, and a host of others. I find this an easy first stop when I'm looking into something I suspect would be in the public literature, searching for something that I think a government agency is likely to have studied, or trying to identify the leading voices in a field.
Yeah, it's not perfect. Its advanced search option isn't advanced; since it's a metasearch engine, it dumbs down each query to the lowest common denominator. There's no way to limit your search to only free sources and yes, a good amount of the content in WorldWideScience isn't free. When you click through on a title, you're dropped immediately into that database. If that file is hosted in a non-English speaking country, you may have trouble using the interface.
That said, their search results screen gladdens my heart. There's clustering on the left, sorting and limiting features, and the ability to export the citations. It also shows how many results you got from each of the portals, which might prompt you to go straight to that portal and search it directly.
