SiteLines - Ideas About Web Searching

Google Scholar gets better at indexing PubMed content, but it's still several months behind.

With over a year since the launch of Google Scholar, I thought it was time to revisit my test of Google Scholar's indexing of PubMed content. In my Sitelines article, Google Scholar is a Full Year Late Indexing PubMed Content of February 8 2005, I ran a test to see how GS's coverage of PubMed stacked up. Using a randomly selected list of clinical trials on breast cancer (I wanted important articles that no physician would want to miss) spanning approximately 18 months of publication coverage, I discovered that GS was about one full year behind in coverage of PubMed.

So, has anything changed? To re-test the coverage, I chose another 10 clinical trials on breast cancer from the current PubMed database, randomly selected but spanning the period June 2005 - December 2005. The conclusion? The latest PubMed citation to appear in both Google Scholar was from August 2005 -- almost 5 months ago. Yes, it's an improvement over the last test, and a significant one. However, for physicians and those who need to know, PubMed continues to be an important tool for accessing the current medical literature.

Google Scholar Grows - An Update

Google Scholar's chief engineer, Anurag Acharya, contributed a presentation “Searching Scholarly Literature: A Google Scholar Perspective” at the 9th World Congress on Health Information and Libraries, September 23, 2005.

Some key points:

The index has grown significantly in the last six months, although the company does not disclose the actual index size
Coverage by category is focused on medicine and sciences -- medicine 22%; engineering 14%; biology and sociology, 13% each; physics 12%
GS indexes full text of all publishers except for Elsevier and ACS (probably because Elsevier's competitor product, Scirus, is the publisher's preferred source of Elsevier's full-text content)
Google Scholar's sources come not just from publishers (both open-access and commercial), but also from MANY 3rd party hosting services like Highwire, Ingenta, and academic institutional repositories.

Acharya also provided some clues on the question of how relevance is achieved in Google Scholar. He mentions conditions such as "who wrote it, where it was published, how many people cite it, where citations are from" as clues to the relevancy question.

Acharya claims that, contrary to reviews indicating otherwise, the indexing of PubMed content is "fairly complete." (Watch for a test of that assertion in a future issue of SiteLines.)

New! Google Blog Search

Not to be outdone by upstart competitors (Technorati, Blogdigger, Feedster, and more), Google has announced a beta-version of its blog search. This is still a baby-beta version: it covers blog content back only to June 2005 so far, although it's reasonable to expect that the coverage will increase as takeup of the product ramps up.

Unlike most Google search appliances, Google's Blog Search doesn't search the full text of blogs -- rather, it only searches the "feed" -- the part of the blog posting that an author sends out through an RSS feed. Most bloggers only send a short part of their blog posts through feeds, and as a result Google's blog search won't cover the parts that aren't fed. What's more, if a blog lacks a feed (through either RSS or Atom) Google's Blog Search won't index it at all.

Although this seems like a substantial problem for the moment, I expect that most bloggers will catch on to this and adjust their feed content to contain either fuller blog content or more carefully-crafted keywords in the feed content.

Google's Blog Search also provides its own RSS feed. That's not particularly novel, but with a tool of this size, it's a big plus for those of us who are trying to keep up to date in the ever changing web world. Just key in your search, then click the feed button to get feed link that you can plug into your newsreader of choice.

WAY more Google real-estate devoted to ads

Have you noticed that Google has expanded the maximum number of horozontal ads on its results pages? In this example of a search for 360 degree feedback, it's rather startling to see just how much real estate of the initial results page is devoted to ads. In my 1024x768 display, I'd estimate that between 35-40% of the initial screen contains "pure" search results, with the rest devoted to ads or header information. How interesting.

Google Tests "Commercial" Listings in "Organic" Search Results

From August 19's Clickz, an article on Google's testing of commercial listings in the 6th-8th position in "pure" search results. I replicated the test of the keywords on demand and the results are clearly visible. Interestingly, what isn't at all obvious is that the results are commercial/sponsored/paid in nature. The only apparent difference is the line above and below the commercial results, and the absence of either CACHED or SIMILAR PAGES links that usually company Google results.

I hate to say I told you so, but I predicted that Google would cross the line into paid-into-pure integration as early as 2003, when rumours began to surface about a possible public offering of Google. This appears to be the first public indication that the company is seriously testing the waters.

Want a glimpse into the search future? Google Labs and Yahoo Next

Want to see what's new on the (commercial) search horizon? Try the not-quite-ready-for-prime-time search offerings of the major engines, like Google Labs and Yahoo Next.

One tip: If you're like me and you want to try to get a glimpse of trends, try to look for similarities between the offerings of these two hot competitors. Right now I'm seeing pointers to a differentiation of two types of Internet -- one for "serious" research-oriented questions (although I'm curious to see how either of these behemoths defines "research") and another for shopping.

We already see some of that differentiation with Google Scholar, Google Print, Yahoo Mindset and Yahoo Subscriptions. Each of these alleged "testbeds" offer searchers access to more highly differentiated content than would be available in straight-up Google or Yahoo.

Why separate research and shopping, and what will this do for engine revenues? These new spinoffs offer a great untapped market for paid consumer-level content, as well as context-sensitive ads. The big engines are clearly betting that serious searchers are ready to pay, if only to satisfy their need for quality information that they can't find in a sea of search engine spam.

By separating research (i.e. pay-per-view, with a few free tidbits thrown in, like content from .edu sites) from shopping (i.e. everything else), the search engines can return to more traditional relevancy algorithms. Remember relevancy conditions like link analysis, proximity of terms, frequency of term occurance, and currency of information? In an shopping-free engine, those conditions can be re-introduced, leaving the optimizers to focus their beat-the-engine techniques on the shopping side of search.

Advice to Physicians on Google Scholar

Jim Henderson, director of the McGill University Medical Library, contributes to the assessment log on Google Scholar in the latest issue of the Canadian Medical Association Journal. Henderson weighs in with a brief but pointed assessment of the relative value of Google Scholar for practising physicians. Using simple but effective examples Henderson shows that the ability of GS to provide fast, easy access to important recent medical information is highly variable, and he concludes that "it is not a useful choice for physicians."

Google Scholar Now Enables Direct Linking from Any Institution

Google Scholar has just enabled a new preference that enables any library to provide direct linking capabilities to users who search Google Scholar but want to find only those links to resources held by their institution. In addition, GS has improved its link resolver capabilities. For more information, see Gary Price's posting in Resourceshelf.com, which makes reference to several helpful resources on the topic.

Why is So Much New Google Stuff in "Beta"?

In his February 11 posting in Searchenginewatch's blog, Gary Price considered why so many of Google's innovations can be labelled "beta" for up to 5 years.

Rather than concluding that "beta" means nothing to Google, Price suggests that keeping new initiatives in beta has positive marketing value for the company. "Having the term beta attached to many of their services for extended periods of time has marketing and pr value. Huh? By using just a single word Google can tell users, the press, and anyone else who is interested, that of course problems with a service might and will likely exist. Why? Because it's just a beta release. It's a simple, clean (at least up to this point), and easy way to deflect or soften negative attention that a service might receive. More brilliant marketing from Google."

How Google's Blogspot Spreads Spyware

Interesting article today by Harvard economics student and spyware sleuth Ben Edelman. "How Google's Blogspot Helps Spread Unwanted Software" reflects on how many blogs hosted by Google-owned Blogspot have javascript that tries to trick the user into installing (un)necessary software to power the blog -- only to install spyware on the user's machine.

Google Uncle Sam -- What It Hits, and Misses

Interested in best practices for using search engines to find US federal government information on the web? Then make haste to Peggy Garvin's excellent assessment of government web search tools in this month's LLRX.com. This is a great guide to help researchers wade through the forest of publicly accessible government information.

Although Garvin starts off with a discussion of the merits and limitations of Google Uncle Sam, the article also covers the Firstgov search appliance, and suggests Vivisimo's FirstGov cluster search enhancement and the Department of Defense's own internal search tool.

Garvin concludes with the expert web searcher's maxim -- use more than one search tool!

How Google Scholar (and others like it) help libraries

I always appreciate when readers take the time to write to me with their comments -- most of us have so little time, so the effort is doubly appreciated. Ben Toth, Director of the National Electronic Library for Health (in the UK), took me to task (very gently) on my generally positive views about libraries and rather negative views about many commercial search engines. Toth felt that my views didn't fully take into consideration the many good things that popular commercial search tools have done for the library community:

"Many of us who welcome Google do so because of the challenge it is making to traditional libraries, not because it is a superior search tool. My feeling is that we don't fully appreciated the value of the disruption that Google/Amazon etc are causing to libraries - forcing us to think outside our professional boundaries to a degree that wouldn't have happened had Amazoogle not been invented. For my money Google is working towards the dream of Paul Otlet for universal access and for that reason alone it is vital to work with Google rather than against it."

Google Scholar is a Full Year Late Indexing PubMed Content

In case you don't want to read any more, tests conducted by me on Feburary 8 2005 suggest that Google Scholar is currently missing almost a full year of PubMed records. The conclusion is pretty simple: No serious researcher interested in current medical information or practice excellence should rely on Google Scholar for up to date information.

Read on if you're interested...

I've been following the buzz lately in weblogs and around the water cooler as medical students and faculty discover that Google Scholar can search against Medline records from PubMed. One medical school's student web site has this caption beside a link to Google Scholar: "Give Google Scholar a try. It searches Medline and it's fast."

Academic librarians find themselves in a quandry -- responding to students who just want it fast, thank you -- when they know about the deep limitations of these quick one-box tools like Google Scholar and other search engines. But they haven't done a very good job so far at showing just how seriously flawed these third-party tools are when it comes to serious medical or scientific research.

I've written about this before in a previous Sitelines posting but then decided that perhaps some real research into the differences might help clear up the problem, or at least give librarians more ammunition to convert users.

MYTH: Google Scholar searches Medline and it's fast.
FACT: Google Scholar is fast.
FACT: Google Scholar does not search Medline. It searches whatever Medline records NLM happened to give Google. We have no idea when NLM gave Google the records. We can't anticipate when the next batch will be delivered and the Google Scholar database updated. Remember, Google Scholar is just BETA. PubMed is...well, decidedly NOT beta, and full of the important checks and balances that make it so special.

It is possible to test the existing PubMed content in Google and draw some conclusions about currency, in order to advise students and researchers on best practices (and dangerous practices).

Here are the details of my test.

I conducted the following searches in order to compare the total number of Google Scholar records for 2004 and 2005 with those in PubMed for the same period.

Google Scholar - I searched for any link from site ncbi.nlm.nih.gov (which is the closest I could get to limiting to stuff from PubMed) AND dates anywhere in 2004 or 2005:
http://scholar.google.com/scholar?hl=en&lr=&q=site%3Ancbi.nlm.nih.gov&as_ylo=2004&as_yhi=2005&btnG=Search

PubMed - any article published between Jan 1 2004 and April 1 2005:
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=PureSearch&db=pubmed&details_term=%222004/01/01%22%5BPDAT%5D%20%3A%20%222005/04/01%22%5BPDAT%5D

Results: For the 2004-2005 period, Google delivered 29,500 records; PubMed delivered over 658,000 for the same period.

While some of these PubMed records in the larger set are PreMedline records that will ultimately be dumped from the database, there is simply no accounting for the enormous difference in the numbers, except to hypothesize that Google Scholar is missing very significant quantities of relatively recent records from PubMed.

To check my hypothesis, I searched PubMed for randomized controlled trials about breast cancer, published from mid-2003 to the end of 2004. I selected about 20 from this group of important articles (that any practitioner should be interested in) , which spanned the entire time period of the set.

My goal was to see which of these articles showed up in Google Scholar. By tracking the dates of the articles captured in Google Scholar against those that were missing from Google Scholar, I could venture a guess as to the last update of the PubMed content in Google.

I tested by copying and pasting a distinctive phrase from the title of each article into Google Scholar, then clicked SEARCH to see if Google Scholar would retrieve the article. My results showed that Google Scholar failed to retrieve any PubMed content after February-March 2004, making Google Scholar almost one full year in arrears of PubMed.

A First Look at Google's Video Search Tool

Richard Wiggins reflects on Google's newly launched video search tool, in this newsbreak from Information Today, and finds it a little too lite. Speculating that Google launched the tool mainly as defense against other competitor products by Yahoo and Microsoft, Wiggins tried several searches and discovered that, for most search results, there is no actual video to watch. To which Google says, stay tuned...

Librarian Blogs on Google Scholar

TJ Sondermann, a reference librarian in Massachussetts, started a blog about Google Scholar.

Librarians and Publishers weigh in on the Google Library Project

Barbara Quint, in her December 27 NewsBreak, posted at the Information Today web site, asks lots of questions about the Google Library Project -- and publishers and librarians have lots of answers. But where is Google? Company reps continue to evade the questions and provide frustratingly low (i.e. no) information that can help us understand their goals and plans for the project. (For an exercise in evasiveness, just see Google's non-responses to a few straightforward questions posed by Gary Price in his December 15 Resourceshelf.com posting.)

A way for Google to monetize article searching?

In "Google Magazine Search" Susan Kuchinskas explains how Google may be able to monetize its popular Google News services (and no doubt other Google services that deliver magazine content, like Google Scholar) without alienating publishers. According to Kuchinskas, "Google's problem with its News service is that there's no way to monetize it. News publishers would cry foul if it displayed contextual ads against their content, even if it is just headlines and openers. "

U.S. Patent Application No. 20040122811, filed under the name "Method for searching media, " gives some insight as to how Google could "offer pay-per-view documents, scanned documents with clickable ads and even the ability for print publishers to swap out ads in digital copies of their printed pages.

"There are two key elements of the patent: a method for executing a permission protocol so that the publisher could authorize Google to display more text from the relevant publication; and storing scanned versions of printed documents along with data sets representing the ads that went with them."

The creative aspect of the patent is that allows Google to take a flexible approach to delivering publisher content to users -- through ad-revenue sharing arrangements with publishers, who could use Google's permission protocol to deliver more content via referral from Google than it might through other sites (and to share the ad revenue generated from the accompanying ads), as well as straight pay-per-view options with shared revenue between Google and the publishers.

Kuchinskas speculates that the patent application is a likely indicator that Google has already developed the technology to do all this.

Understanding the Tension between Libraries and Google

There's an excellent article by May Wong of the Canadian Press and Associated Press, which appeared in the December 15 issue of the Globe and Mail. "Google Move Could Commercialize Libraries" discusses the differences between libraries goals of digitization (better access, free access) with Google's (unclear at the moment, but likely better access, pay-to-view).

Many library leaders wonder about who will win the war for eyeballs. "There is anxiety about whether the student researcher, scholar or citizen will be guided into the free public access rather than being lured into a purchasing relationship with the publisher," said Duane Webster, the executive director of the Association of Research Libraries.

Cool Tool; Google Suggest

Nifty, simple, auto-complete tool at Google, called Google Suggest produces an automatic drop down menu of possible choices drawn from Google's most popular searches. OK, good for spelling help, or novice searchers looking for basic information. By the way, there seems to be some basic safe-search restrictions here -- I received no auto-complete options for a few typical roots.

What's behind Google's massive library digitization project?

Gary Price has provided the essential details of Google's just-announced project to begin rapid digitization of millions of out-of-copyright library materials. Although a handful of libraries (albeit major ones) are involved at this point, there is no doubt that many more will follow. (Google is footing the cost of much of this, so the offer is hard to pass up.)

What's behind this venture, and what type of business value can Google expect to see from this? Database size alone doesn't make for a sound business model, as revenue must follow, particularly in a newly-public company with shareholders looking for ever-increasing quarterly results.

One can see that all this free non-copyright full text content will enable Google to successfully market its now-way-bigger database as the source of a major portion of "serious" full text knowledge. That's an important first step that can allow Google to fully dominate the information market as the "first place to look" for ANY type of information. And following that, it will be much easier to convince trade publishers of commercial books to FULLY digitize their content AND make it indexed and available via Google Scholar with revenue generated from the shared sales of that for-fee material. (Remember that the trade content that's available through Google/Google Scholar right now is either just abstracts OR a few pages of the book, not the entire book.)

But book and article purchases alone probably aren't enough to produce the necessary revenue for Google, so I'm guessing that the company also factors in a large increase in advertising revenue into this business model.

Why will so much new advertising accrue? Simple -- because all this new digitized content will be dumped into an already-too-huge Google database. Giant databases make information harder to find, not easier, particularly when the only finding methodology is keyword guessing and algorithmic ranking.

Huge databases often frustrate users who are looking for something beyond basic information and general opinion. That, plus the unpredictable ranking of search results, ends up convincing advertisers that the only reliable way to reach their customers is to buy keyword placement. Even subdividing the Google database into sections (which currently exists with Google Scholar, Google US Government, and other Google special searches) is not necessarily good for normal human searchers (too many places to look), But for Google, it's all good -- a bigger package of targeted, Google-branded real estate to hold ads; and more advertisers competing for a limited number of keywords.

Here's my holiday wish -- for those academics and librarians that are now busily sharing this hot "news" about this latest digitization project: Please, don't go so gaga over this new Google venture that you see it for more than it actually is -- simply a way for Google to further increase its database size, thereby maximizing its revenue potential over time.

Peter Jacso reviews Google Scholar (with feeling)

Peter Jacso of the University of Hawaii has written a long review of Google Scholar in the December issue of Péter's Digital Reference Shelf. Although most observant reviewers have noticed that Google Scholar has missed a lot of stuff that one would normally expect to find, Jacso conducts many more tests and finds Google Scholar lacking in pretty well every aspect -- content and relevance being the major gripes.

Side-by-Side Measurement of Google Scholar vs. Publisher's own "native" search engines

Peter Jacso of the University of Hawaii has developed a simple but very effective tool to help web search evaluators measure the relative capabilities of Google Scholar against the publisher's own "native" search tools. Side-by-Side Native Search Engines vs. Google Scholar easily demonstrates that there are some limitations to Google Scholar's capacity to deliver results as good as the publisher's own native search tool.

From Jacso's commentary:

"Preliminary tests have shown that Google Scholar often retrieves far fewer unique items than the native search engines of the publishers. On the positive side, Google Scholar links to citing references if the document was cited by journals indexed in Google Scholar, and provides the immensely useful citedness score of the documents.
When Google Scholar has more "hits" for a query, they often turn out to be duplicates and triplicates (not always displayed adjacently) with a separate hit for the TOC entry, the abstract, the PDF file and (if available) the HTML file. Although their URLs are slightly different, they take you to the same spot in the archive. "

This conclusion is consistent with similar tests that I ran against Google Scholar's capacity to retrieve PubMed records, versus the same records available from a search of the PubMed site. Not only does PubMed retrieve far more results than Google Scholar, but it has the capacity of highly sophisticated native search logic that would never appear in Google. My search of the keywords asthma children in Google Scholar (with an added limit to retrieve only PubMed results) came up with approximately 14,400 records, as compared to the same search in PubMed, which retrieved over 23,000. I reported similarly disparate results earlier this year in a side-by-side comparison between Google and PubMed, titled "Just because it's indexed doesn't mean you'll find it."

Expected (but missing) content on Google Scholar

Jay Bhatt, engineering librarian at Drexel University, conducted a fairly detailed assessment of Google Scholar's indexing of engineering free scholarly content, and came up with a lot of missing stuff. See his comments, reposted from ELDNET-L, at http://stlq.info/archives/001683.html#001683

It's important to remember that Google Scholar is still very much in beta form, and it's hard to know how much effort Google will put in to two critically important tasks: first, undertaking the laborious process of identifying and indexing more scholarly content; and second, tweaking of the algorithms that Google Scholar uses to drive results ranking.

Like Bhatt, I have been playing with Google Scholar now for several days and testing more "scholarly" searches in both Google Scholar and straight-up Google. On the plus side, the ranked results in Google Scholar are typically VERY different from those in straight-up Google. On the minus side, the keywords that I use to deliver known content from Google Scholar tend to work better in straight-up Google.

For example, I know that Dalhousie University has a superb web site devoted to recommendations of handheld content in medicine. Using the keywords Dalhousie handheld in Google, my results pull the web site up in the third result. But conducting the same keyword search in Google Scholar, my results don't pull up the site at all in the first 20 results.

My conclusion from these early-on examples suggest the following recommendation: continue to use straight-up Google in cases when you know the site you want, but you just couldn't remember the URL. Consider adding a second search in Google Scholar when you want to bring up different information on a topic when the precise source of information is unclear.

A first look at Google Scholar Beta

Today's launch of the new Google Scholar beta site led me to spend an hour testing the site for quality and quantity of information.

Google Scholar uses a special algorithm (unknown to us, and probably a state secret) to calculate the "scholarliness" or seriousness of a particular hit in the Google database. The idea is to create a subset of the enormous Google database in order to satisfy less consumer-focused searches.

First some background. It's important to remember that over the last year or so, Google has been working feverishly to add a variety of invisible web content into its databases, including library catalogue records, invisible web databases like PubMed, and full-text books as part of its strange-bedfellow partnership with Amazon.com through A9.com. This previously unindexed content is likely the source of many of the 3 billion pages that Google added to its index last week. Several of us in the search assessment world have been griping lately about the fact that most of this content would never be found in a typical Google search. (See my article Just because it's indexed doesn't mean you'll find it for a little background.) This beta rollout of Google Scholar directly addresses that criticism.

In my first glance at Google Scholar, I see that although much of this newly indexed content is delivered through this site, much of it leads only to citations or licensed databases of full text content that can't be accessed for free.

For example, a search of "human resources" benchmarking turned up approximately 4700 results today (versus 383,000 in the same search in straight-up Google). However, many of those results are delivered as links to full text articles or books from popular commercial publishers and aggregators like Wiley, Ingenta, and Blackwell, with relatively few of those available for free.

I also tested a medical search, for the keywords sumatriptan migraine in Google Scholar. As I had hoped for and expected, most of the initial results came from the PubMed database content that Google has been indexing for the last year or so, and the PubMed result comes up on click. However, this should not be taken as a recommendation to search PubMed through Google Scholar rather than directly through PubMed, which offers an updated database and better interface options than Google or Google Scholar.

At this stage of Google Scholar, there are MANY results that users simply won't be able to access without paying themselves for the article, or finding an enterprise partner (like their school or college library) that owns the journal. That will likely create a level of frustration among users. One easy way for Google Scholar to solve the problem would be to model Findarticles.com, which enables users to check a box and omit all but free articles. (This would certainly lead to lower revenue and some pouty publishers, so don't bet it will happen.)

The one truly fabulous feature of Google Scholar (which really WILL help serious searchers and academics, rather than just confuse them) is the "cited by X" feature of the database. If Google has a link to another document in its database which has cited the article that comes up in the hit list, it will produce a link labelled "Cited by [#]" where # is the number of cached documents which cite the document. You can click on this link to bring up a page of the cited references. This isn't just cool, it may actually be useful to researchers interested in expanding their range of citation searching options, and also could be used in determining relevancy.

There's no advertising on Google Scholar -- at least for now. It's kind of a moot point anyhow. Although Danny Sullivan reported in today's Search Engine Watch that Google claims that is not earning any revenue from new subscriptions between searchers and publishers, my guess is that Google may have enticed publishers into making their content available on Google with the promise of some sort of revenue sharing or micropayments of some sort, and as a result Google will likely see enough revenue to make this return on investment a viable business option.

The presence of Google Scholar presents a really interesting problem to libraries around the globe. Libraries have been working furiously to make their licensed content accessible through federated searching INSIDE their library web sites, and it's been pretty easy to make users understand that the only way to access licensed content would be by entering the library's portal site.

Google Scholar has the potential to turn that model on its head. Students will begin to access links to journal and book content directly through Google Scholar, and depending which IP they are coming from, may or may not be able to access the full text content directly on click. Talk about confusion. In a typical scenario, when accessing Google Scholar from inside a campus IP which is connected to, say, Ingenta.com, students might have access to a full text journal link. But from OUTSIDE the IP, the same link will produce a "please buy me" result. Yikes.

John Dvorak on the Microsoft-Google War

Spotted on the TVC Alert, a very interesting article, "A Google-Microsoft War" by columnist John Dvorak in PC Magazine. Dvorak speculates on what drives the current dust-up between the two web/pc superpowers; the unlikely prospects for better search functionality or natural-language developments; why rumours of browser takeover of your PC's O/S is unlikely; and Microsoft's uphill battle for search supremacy over Google. According to Dvorak, "the only light at the end of the search engine tunnel for Microsoft is that Google's once-vaunted page-rank methodology, which thrust it into leadership, has been rendered useless by the ever-changing nature of the Web. But Google still seems to have the most crawlers, the biggest server farms, and the most depth. If Microsoft can't get close to that, then these other features, even if they work, won't make much difference."

Google Cheatsheet

Google has produced a printable one-page cheatsheet of selected commands here

Another author spots overcommercialization of Google search results

Journalist Cheryl Woodard, of media-watch site Askquestions.org, spotted a raft of commercial results while searching Google for information on troubled teens. Her findings indicate just how much more commercial search engine results are becoming:

"None of the magazine articles, parent-to-parent advice groups, expert opinions, or public policy information I know about turned up on Google. .. And although plenty of community-based programs are available, they didn't show up in my Google search... From a user's viewpoint, this search experience is frustratingly commercial. All of the most effective and affordable treatment options are missing. Journalists, academics and other impartial information sources are scarce and hard to find in the overwhelming numbers of sites served up. Welcome advice from other parents cannot be heard. And important government research information does not appear. The results are wildly out of balance in favor of commercial rather than consumer needs. It's time for Google to upgrade their product for consumers. "

You can read all of Woodard's excellent study at http://www.google-watch.org/woodard.html

A Google Engineer Stuggles Against Spammers

Spotted on ResourceShelf, this slide presentation from Amit Singhal, Senior Research Scientist at Google. The slides (available here in PDF format) contain some great examples of how spammers fool Google and what Google tries to do to combat their efforts. Singhal does a great job at making Google look like the great white knight of search. Fascinating.

Yahoo's Stake in Google

Until I read about it in an Associated Press story, I didn't know that Yahoo! had a huge stake in Google. It's not just this week's news that Google settled a Yahoo-initiated lawsuit by giving Yahoo 2.7 million shares (over $300M) of Google stock, but that even prior to the lawsuit settlement, Yahoo! already owned 5.5 million shares of Google, acquired through a $10M investment made in June 2000. This is a fascinating twist, as all this Google cash gives Yahoo a huge pile of cash to compete more effectively with.... Google!

How to Search Google from your IE address bar

I've reviewed several articles lately on how to change various Internet Explorer's MSN search defaults so that Google becomes the default search engine in IE rather than MSN.

Almost all of the articles suggest registry changes, which is fine if you're comfortable changing the registry, but most of us would never touch the thing.

Happily, there is a fix: If all you want to do is change your default address bar search engine to Google, you do not need to edit the registry. You can make Google your default address bar search engine by clicking the Search button on the top toolbar of IE, then click on Customize, and choose Autosearch Settings at the bottom of the box.

For those comfortable with registry change settings, here are some reference articles, but readers should be aware that I haven't tested the instructions provided and can't vouch for their effectiveness.

http://pubs.logicalexpressions.com/Pub0009/LPMArticle.asp?ID=87
http://www.google.com/options/defaults.html

The Nature of Meaning in the Age of Google

In the April 2004 issue of Information Research, Terrence Brooks of the University of Washington's Information School wrote a very interesting article on how Google's algorithmic method of indexing has created a new culture of "lay indexing" with a variety of consequences for serious information seekers.

Brooks notes that in the age of Google, the nature of meaning has changed, moving from trust in the expertise of a few to reliance on the aggregation of many opinions, including many uninformed ones. The article is full of interesting insights, including thoughts on the competition between effective algorithmic measures of meaning and optimization attempts to distort it; Google's bias against obscure web sites; and much more.

Use of Google by Judges to Substantiate Information

In the May 13 TVC Alert, Genie Tyburski came across a fascinating article on how US judges have used Google to substantiate information used in trials. Some of the examples betray a rather startling lack of understanding about Google's limitations, and leaves the reader wondering just how Google got to be such a trusted source among those (like judges) who ought to know better.

Google Has It's Own Blog

Google has finally started up its own weblog on all things Google. Although only a few days old, the initial postings (come work in our Zurich office, and more pseudo-apologies on the JewWatch.com controversy) make me think that this will be more of a public relations feed than anything else.

Google - Loyalty or Laziness

Chris Sullivan comments in this week's Search Engine Watch on a recent study conducted on behalf of search engine marketing firm iProspect about apparent loyalty to search engines. According to iProspect's press release, "56.7 percent of Internet users use the same search engine or directory when they are looking for information, and another 30.5 percent of Web users have a few specific search engines they use regularly. A small amount of Web users, 12.8 percent, said they use a different search engine each time, depending on what they are looking for at that moment.

To iProspect, the study proves loyalty to search engines. Although the research methods aren't explicitly stated, the sample size of almost 1700 respondents is significant. And results come as no surprise to any teacher of web search skills, who already knows that users use what they know and can remember (and it's usually a popular commercial search engine).

I'm somewhat more inclined to side with writer John Battelle, who says, "I'm not sure I buy the whole search engine loyalty thing. I think folks aren't loyal, they're lazy." But surely in addition to the ease and simplicity of anywhere/anytime searching, I think that many searchers are simply unaware of alternative quality resources.

Evidence suggests that even serious academic researchers seem to go for Google over databases in a big way. In "Is Google the Competition" which appears in the April 1 2004 issue of LibraryJournal, Carol Tenopir describes an (unreferenced) Elsevier study, in which Elsevier asked librarians and scientists to name the top three most reliable online services. "Librarians named ScienceDirect, ISI's Web of Science, and Medline. Scientists, on the other hand, named Google, Yahoo!, and PubMed."

As search engines continue to devolve into big-box shopping and entertainment channels, it will be interesting to see if brand "loyalty" to these tools continues. Watch your favorite science prof for details.

Google's Ranking and the Jewwatch.com Controversy

The New York Times of April 13 carried a story on the high ranking of the anti-semitic site Jewwatch.com in Google's search results of the keyword jew.

Google was asked to remove the offensive "content" but declined, asserting that it would be inappropriate to attempt to manually regulate its algorithmic rank ordering except in cases where inclusion of the link would be illegal. However, on a search of the word jew in Google, searchers will see the unusual positioning of an official Google statement in the space typically reserved for sponsored links.

Although Google may be caught in a no-win PR flap, the real story may be the stunning and continued reliance on a one-box search engine for all the world's information. Writer John Battelle, co-founder of Wired Magazine, founder of the Standard, and currently a visiting professor at UC Berkeley, suggested in his weblog that the real issue "is the fact that Google is taken as the first and last word on what our culture believes to be important w/r/t any given term."

Google FAQ

There is an excellent set of frequently-asked questions about Google at http://www.geocities.com/googlepubsupgenfaq/. The FAQ contains extracts of popular questions asked and answered at the newsgroup google.public.support.general. In addition to typical "How do I ..." search-related questions, there are answers to questions about Google backlinks, an explanation of "Supplemental Result" and "Cache"; and how Google redirects users to specialty sites (like Google.ca for Canadians) using geotargeting. (Spotted on Researchbuzz.com).

Subtle Look & Feel Changes to Google Favor Advertisers

Google launched some subtle changes to its desktop today. Although they are minor, most of them provide for increased visibility of ads (though Google calls this "enhanced usability" and "easier access to frequently used tools and services"). Ads on right are now separated by a vertical line, not boxes; a larger ad font is used; the ads that are placed above the results have a uniform background color rather than different colors for each ad; removal of the tabs that led to the Google Directory, Newsgroups and Google News, replaced by above-the-search-box links to all of those plus Froogle, Google's shopping site.

Details on Google's SEC Disclosure

Spotted on Resourceshelf.com, "Google facing public filing as it grows" by Steve Gelsi from CBS Marketwatch provides more details on the possible/likely requirement that Google will have to disclose certain financial details to the Securities and Exchange Commission even if it decides not to file an initial public offering. Gelsi notes that they hype surrounding the Google IPO is tremendous: "Nevertheless, expectations continue to be high for Google's IPO, which could fetch as much as $4 billion in proceeds as the biggest Internet deal of all time."

More than 55M Europeans Search with Google

In a press release from Nielsen Net Ratings, European searchers rely on Google most. Over 47% of Europeans polled use Google as their search engine, well above the 23.1% that rely on MSN Search, the second choice.

Because Europe lagged somewhat behind North America during the 90's in embracing the Internet, there are fewer high-quality made-in-Europe directories and other alternative choices for information searchers who want to go beyond Google. Europeans who want a Europe-focused search tool must rely more heavily than North Americans on commercial search tools, Yahoo-lookalike commercial portals, and existing made-in-the-USA directory search tools.

Google Guide

This is a handy new web guide to using Google. Produced by Nancy Blachman, a co-author of the book How to Do Everything With Google. The site is designed for searchers who want to go "behind the box" to understand how Google actually does its job. Although the guide provides better search help than Google's own meagre help files, the coverage focuses on functionality and features.

Google's Goofs - including some that may actually matter

In his February 1 issue of ResourceShelf.com, Gary Price recently pulled together a series of short entries about issues and concerns regarding Google. There are several inconsistencies in search syntax that need fixing; using the backward link search syntax (used commonly to see who is linking to a particular page) only pulls results with a minimum PageRank; announcing command-line search services that either don't work properly (like airline flight numbers) or are removed shortly after announcement (like the WHOIS lookup). Price concludes (and I agree) that these gaffs, as well as the introduction of many potential revenue-generating services that are unrelated to search (like Google email, which was rumoured by several sources last week), could indicate that Google is moving away from its primary focus on search relevance in favor of expanded revenue generation models.

More on Google command-line searches

Google has added several new "command line" search options to it's search capability. You can now search by UPS or Fedex tracking numbers, US patent numbers, and FCC equipment tracking number, in addition to some now-well-known options for definitions, US phone numbers, stock quotes and US street maps. Google provides brief information and examples at http://www.google.com/help/features.html#number and more options will likely be added.

Let's deconstruct Google's definitions feature using the syntax define: your_word_here in the command line. For the search define: optimization several definitions from multiple sources are returned, but we have no ability to find a list of the sources or how they were selected.

These simple one-box auto-search features are good for simple lookups like Fedex numbers -- but for information where some intellectual analysis may be needed, is it appropriate to leave it up to Google to choose the recommended database?

Danny Sullivan of Search Engine Watch wrote an interesting piece titled Searching with Invisible Tabs in December 2003, on Google's propensity to try to search just about anything from the command line (he dersctibes it the "swiss army knife for search"). Google's back end would recognize certain qualities of the search string, and would automatically query one or more prescribed database for the result. As with most easy-search solutions, there are benefits and risks, and Sullivan summarizes these well.

20 Great Google Secrets

Web-maven Tara Calashain, author of the book Google Hacks has written a neat little piece to remind us of how many Google search tools exist under the Google banner. 20 Great Google Secrets provides helpful reminders about search syntax tricks, directory lookup features, spelling features, and more.

Big Changes Coming for Google

OK, things are heating up in the search wars. The big news last week was about Google's plans to issue an initial public offering in February 2004, plus a report today in the New York Times that Microsoft is in discussion with Google to explore a more connected relationship, with murmurs of partnership or merger. Either way, expect big changes, which may not spell good news for searchers.

Dogged by Overture, Google tests local search

The big Google news of the last few days is Google's test of a local search function. In development by Google Labs, this feature would allow a user to type in a search term, along with an address or ZIP code (US only for now), and find web results along with a mini-map of the area.

More interesting than the function is the timing of the announcement, which comes just weeks after Yahoo!-owned Overture, Google's principal search-ad competitor, started testing a similar local-search feature on AltaVista, an Overture search brand. (When I checked today, the demo had been taken offline.) For both Overture and Google, the processes of selecting what constitutes "local" are algorithmic, with crawlers programmed to identify page conditions that would seem to qualify as local information.

Why is local search so hot? Google company spokespersons were quoted by CNET News that local search is part of "Google's goal ... to connect users to the information they need, whether it's half way around the world or just around the corner." Yeah, sure, whatever. Local search is driven by the quest for ad revenue. The advertising stakes for local search are huge (think of all the pizza parlors lining up to buy Google Adwords) and at the present time, growth in the local search ads marketplace is largely unexploited.

Google is testing "Related Searches"

Aaron Schwartz reported last week that Google is testing a "related search" presentation near the top of search results.

You may be familiar with this feature in other search engines, but it would be new on Google. I'll watch for additional information on this topic but in the meantime you can see a screenshot from the haymeadows.com site, although I'm not sure precisely what search it is displaying.

Google's Synonym Operator

Google recently announced that it has introduced a synonym operator using the tilde ~ character. Google provides some very brief information on the feature . Greg Notess has a useful description and some examples.

Of course, Google is silent on how the synonym function actually works. Clearly it is algorithymically based: no human has coded the contents of Roget's Thesaurus! However, Marydee Ojala ran some tests and reported them in a recent news alert, "Google Introduces Synonym Searching, News Alerting Features", part of the Information Today web site. Ojala concluded that the influence of Googles PageRank is obviously present, which of course skews results toward popularity, particularly technology, business and entertainment.

Andrew Baio of Waxy.org came up with a neat way of identifying which synonyms Google is using. Search for your synonym keyword and then exclude the same keyword from results. For example, searching ~help -help will result in search results that lack the keyword help but include the synonyms (in bold type).

Tips and Tricks for Google Geeks

In the April 21 2003 issue of the International Herald Tribune (but just noticed this week!) Lee Dembart reviews a few great Google "hacks" covered by Tara Calashain in her book Google Hacks: 100 Industrial-Strength Tips & Tools. Tara also has a spot on the web where you can find all the hacks that she has devised. Many of the hacks are either odd (Do I want Google to generate poetry?) or useless (Who needs another random URL generator?) but there are a couple of notable ones for serious searchers.

• Google Alert allows you to enter your search terms, and the site, which is not affiliated with Google, automatically runs a Google search every day and e-mails the new results to you.
• Goofresh lets you search for pages that were indexed today, yesterday, in the last seven days or the last 30 days. A good one for web collection developers to try out.
  Posted by ritavine at 04:47 PM

The Experts Declare "Nothing New" on Google Proximity Searching

In the June 20 issue of SiteLines, I wondered in print on whether Google could conduct a proximity search. I had tested a search using wildcards without the "double quotations" required for phrase searching and noticed that the search seemed to return results that suggested that proximity might actually be at work here.

Well, I was wrong. To my rescue came colleagues and web search experts Gary Price, Genie Tyburski, and Greg Notess, along with a couple of able readers who contributed comments to the posting. All confirmed that while Google is capable of “sort of” doing a proximity search, it’s very limited and there is certainly nothing new on the topic since it was originally reported in 2002.

Genie Tyburski of the Virtual Chase pointed me to Greg Notess’s succinct discussion of Google’s limited proximity possibilities in his excellent search engine roundup site called Search Engine Showdown. In this feature article, Greg notes that although Google does not directly support proximity searching, there is a trick which can be used for a wildcard word within a phrase placed within “double quotes”. I won’t repeat the entire explanation here: Greg’s description above has excellent examples.

Kevin Shay has produced a very unofficial Google API Proximity Search tool that can reproduce proximity searching up to a distance of 3 words in just one step. Serious searchers may want to bookmark this site for reference should proximity searching ever be called for in a Google search. Shay’s proximity search tool is one of a number of applications built with Google’s API Developers Kit, which provides information on Google’s search syntax and functions so that developers can design programs to query Google’s database in precise ways without having to go to Google directly.

Greg Notess also pondered what might have happened to make me think that the asterisks might be producing true proximity searching without encasing the search in “double quotes”. He tried my original search on bush * * * iran without the quotes and noticed that although you get different results for the two searches, you don’t get a different number than for the search "bush * * * iran", which suggests that the same results are retrieved for both searches but just ranked differently in each.

Greg continues: "Because one of Google’s standard ranking algorithms is to rank phrase matches higher, a search on bush * * * iran finds the [bush AND iran] results and then ranks any matches for “bush * * * iran” at the top or closer to the top. Each search is still finding [bush AND iran] but the ranking changes due to the location of the asterisks." Bingo. I think he's absolutely right with this explanation.

And my friend Gary Price of Resource Shelf emailed me with much the same information, reminding me that he had penned a note about this very topic a year and half ago when Kevin Shay's proximity search tool first became available.

Thank you to everyone who contributed to solving the mystery of proximity searching.

Does Google Enable Proximity Searching?

I spotted a reference by Elwyn Jenkins about possible proximity operators available in Google. The reference, part of the article "10 Ways to Search Google" mentions the availability of the asterik wildcard (Jenkins calls it WildQuote) to enable proximity searching.

Jenkins tested the search bush * * * iran and it seemed to turn up pages that had the word bush up to three words away from the word iran. I tested it and confirmed Jenkins' results. However, I couldn't find any information on Google's help site about this interesting feature.

Searchers use proximity searches to improve relevancy or narrow down very large search results. Even though Google rank-orders results with preference for the keywords in close proximity, use of the proximity wildcard seems to improve results further. For very common searches like the Bush-Iran one above, which will deliver hundreds of thousands of search results, using the proximity wildcard may help improve search results and display better, more relevant sites. The sites delivered in the first screens of Google's search of bush * * * iran look quite different from those for a basic search of bush iran, although in this case it is debateable that quality was improved by using proximity searching.

If you have more information to add that may further explain this apparent functionality, I would welcome your comments on this post in the Comments section that follows this posting.

Search Engine Position Analyzer - A Tool for Web Trainers

A search engine position analyzer enables you to check to see if a particular web address would have been displayed in the first few pages of search results (the only ones most people look at) if you searched using selected keywords.

This type of tool can be of great benefit to those who teach web searching. A position analyzer provides an easy way to discover if an allegedly "invisible" web search result would have come up in a typical keyword search of a major search engine. I often use them to test my class examples.

Search engine position analyzers are employed by search engine optimizers to check relative positions of a client's site. There are several professional-level SE position analyzers but I came across a free one that worked reasonably well in several tests that I conducted. The Search Engine Position Analyzer Script enables users to enter a URL, then enter the keyword(s) that would be searched. Check up to 7 search tools to test, and the analyzer returns information indicating if the URL would turn up in the first 50 results for that keyword search. The analyzer also links directly to the source search tool's results, so you can check the hit list to make sure.

Please note that during my tests, some of the listed search tools did not deliver results properly, indicating that the script may be somewhat out of date. (I attempted to contact the script's author without success.) The script's links to Google, Hotbot, and Altavista worked properly, with Hotbot defaulting to AlltheWeb's results. The script linked incorrectly to or returned no sites from Excite, Infoseek (now Disney's Go.com) and Lycos. Although it also linked to Magellan, that site no longer exists.

Does Google prefer recent pages?

In the article "What Google Leaves Out" Elwyn Jenkins of MicroDoc News describes a test he performed to try to identify what types of information pages Google didn't spider. Jenkins set an automatic crawler to look for occurences of the benchmark word "googlology" (he knew something about the history of the term and the pages that contained it) and found that Google picked up about 29% of available pages that had the term. The pages that are left out of the Google database tended to be older uses of the term.

Does Google prefer to index recent pages rather than older ones? In the long term, what impact will this have on the availability -- and retrieval -- of legacy information on the web? I hope that this early-stage research will be expanded and we'll see more search engine legacy studies in the future.

What happens when Google updates its index?

Denise Bisson, librarian and indefatigable selector of quality web sources for the Search Portfolio, found information on the "Google Dance" -- the process that Google goes through to recalculate its ranking algorithm, PageRank. The process can take several days to complete. During this period, search results may fluctuate; sometimes minute-by minute, hence the use of the term "dance" to describe it.

Phil Craven's article about the Google dance provides a good explanation in plain language. I also discovered a helpful although unattributed visualization tool of the "dance" in action, which allows you to check the index at all 3 Google servers simultaneously. Possibly not useful for the average searcher, but the story feeds our inner geek.

Google World - Everything Google

Since its inception, Google has quietly taken on many small experimental projects -- some quite useful to serious searchers. Tools like Google Sets (algorithmically finds synonyms to keywords in the Google cache), Google Viewer (delivers snapshots of web pages in the hit list, slowly) and others are interesting and fun to try, and some offsite searches (like the links to Google's patent applications) offer clues to development strategies taking place at Google.

It's always a struggle to find these pieces on the web or within the Google sitemap but now there's a way. Google World, an easy to use mini-directory from Indicateur.com, links to all those little Google pieces that are hard to find unless you can remember the exact names or URLs.

Thanks to Genie Tyburski of the Virtual Chase for finding this little gem from Chris Sherman's longer Search Day article.

Google launches Canadian and UK news services

Google announced today that it has launched both Canadian and UK news services. Readers should be advised that there are only the most minor differences between these regional Google news versions and the baseline version of Google News (http://news.google.com).

I compared http://news.google.com with http://news.google.ca and there are only minor differences. Although both baseline Google News and Google News Canada cover the same 4,500 news sources, an algorithm selects Canadian-focused news stories in the .ca version and places them more prominently on the front headlines page.

The differences pretty well stop there. I searched the keyword "sars" in both the .com and .ca versions of Google News to discover precisely the same results in the results page, in precisely the same order, so no visible differences there.

The news announcement received loads of press today -- all this for a barely altered algorithm?

From Google: Why Pages are Missing from the Google Database

Although it's certainly not a complete explanation of why so much content is missing from Google's database, Google provides some answers. The article is directed at search engine optimizers but is helpful to advanced searchers as well, and reminds us that Google is but one of many search starting points.