Experiments in writing, by Marjolein Hoekstra @OneNoteC

Archive for the ‘Search Engine News’ Category

Supercharging Summize Searches in Firefox and Flock

with one comment

There are several ways you can integrate the superior Twitter search engine Summize into Firefox or Flock, turning your browser into a very efficient Twitter research tool. In this post I discuss these three:

  • adding the Summize search plug-in
  • creating a Summize quick search command
  • using the SmartSearch add-on to perform in-context queries

Summize Search Plug-in

Add Summize to your Firefox search bar by clicking on the “Install Search Plugin” link on the Summize home page:

Summize Search Plugin

The search plugin is available for use immediately after installing. To enter a Summize query, put cursor focus on the search bar with your mouse or press the Ctrl-K keyboard shortcut. Then type your query as usual and press Enter.

Summize Quick Searches
Besides performing searches from the search bar, Firefox also offers the browser address bar to execute search commands. It comes in handy if for some reason you’d rather not change search engines from your search bar. The functionality to search from the browser address bar is generally called Quick Search. In our case you would type in something like “s Obama”, press Enter and then, because of a keyword shortcut command that points to Summize, a query is run automatically to find tweets about Obama. The Quick Search command you can create by following the step-by-step instructions is also useful for in-context searching, which I describe in the section Summize Smart Searches.

Selecting the Add Keyword Search command

The Summize Quick Search command is universal and only needs to be created once. Luckily, Firefox makes this creation process a piece of cake: right-click with your mouse in the Summize search input box on any Summize web page and select “Add a Keyword for this Search…” from the context menu.

Now you may try your newly created Quick Search command by carrying out a query from the address bar: Type “s [keywords]” without the double quotes and press Enter.

Summize Smart Searches
With a few simple steps you can perform a Summize search for any word on any web page displayed in Firefox. The steps to create the Summize Quick Search command that I described in the previous section are required to make this Smart Searching functionality work. First, install the SmartSearch Firefox add-on.

Note that the default, version 3.10, is meant to be used with Firefox 3 beta. Users of older versions of Firefox should install version 3.7, which you can find here: SmartSearch add-on for Firefox 2.

Restart your browser.

Next, open the SmartSearch Settings dialog box and put a check mark in the option Show “Search Web for …” item

Next right-click or select any word(s) that you want to search for on Summize, and select Search for [keyword] on…“, then select @Summize. The following screenshot shows a fun recursive search right from my WordPress editor window, looking for tweets about Summize:

This nifty SmartSearch in-context search functionality works immediately by right-clicking individual words, without the need to select a word with your mouse first. Alternatively select multiple consecutive words, right-click on the selection, and follow the same steps.

My compliments to Ben Goodger and Chris Povlrk for providing the excellent SmartSearch add-on, and of course to the Summize folks (@abdur, @gregpass, @ericcj, @jayvirdy), who in my opinion really created even more than the Google of Twitter. Chapeau bas!

Written by CleverClogs

May 10th, 2008 at 3:03 pm Opens AnswerTips for All Web Publishers

with 3 comments

As of today made their AnswerTips service available to anyone with a blog or website. Just follow the simple steps lined out on the AnswerTips configuration page, insert the code into your site and you’re good to go. There are several icons available: smaller ones, larger ones, animated and non-animated.


These are the icons you can choose from:


In anticipation of the official press release that came out today, I had another go at AnswerTips myself to make sure the service was still working as expected, and then realized that any double-click inside an AnswerTip itself in turn spawns an AnswerTip to be opened in the same space. Now this is really useful because you can now stay on the original page and perform a gradually more precise contextual search narrowed-down through the topic that you’re interested in.

In an email exchange I asked Liz Cohen of if web users who happen to dislike AnswerTips could disable it altogether, and it turns out they can, by checking off the AnswerTips section on the Preferences page:


Fred Wilson, whose investment company Flatiron Partners used to have in their portfolio, turned AnswerTips on today as well on his blog A VC. For more details visit his blog post This Blog Is AnswerTips Enabled.
If you’re especially interested in how performs on the stock market now that this wonderful news was brought out, you may want to follow the developments on the Yahoo! Finances Message Board for (sorry, no feed).

Then some links for those who have missed my earlier posts containing more historical background and ideas:

And as usual a topic news radar based on Google Blog Search if you want to keep track of the blog posts about AnswerTips, presented using Grazr:

Written by CleverClogs

February 20th, 2007 at 4:46 pm

Instant On-site Facts: AnswerTips Enabled on CleverClogs

with 5 comments

I’m very excited to introduce AnswerTips here on CleverClogs. Quietly launched on the CBSNews website last October, the service required some additional fine-tuning before gave me explicit permission last week to embed their code into my blog and to explain what it does.

Before you continue to read, please try AnswerTips for yourself: double-click any word on CleverClogs. You’ll notice that a nice call-out pops up showing the page for that topic. Here’s a screenshot of the Bay Area AnswerTip, for example, taken from my previous blog post Tagged Twice: the Blog-Tag Game:


This screenshot not only demonstrates how visually attractive AnswerTips is, but also that it is capable of performing a contextual analysis: as you can see I double-clicked the word "Bay", adjacent to "Area" and logically AnswerTips presents me with the AnswerTip for "Bay Area".

Historical Background
Two years ago I wrote a rather extensive review of in the blog post Instant Reference Library Becomes Free: 1-Click Answers—how AnswerPages were now being offered for free and what the desktop utility 1-Click Answers was like in every-day use. After this I engaged in an ongoing and at-times intense conversation with product manager Gil Reich. Although the name AnswerTips didn’t surface yet in these chat sessions, we definitely talked at length about the pros and cons of turning 1-Click Answers into a universal feature that would allow web publishers to offer topic definitions on their web sites. 1-Click Answers pointed in the right direction, but we agreed it wasn’t ideal yet: people would still need to install a piece of desktop software, they’d have to memorize the Alt-Click keyboard-mouse combination and it seemed more natural to offer an inline pop-up instead of opening a new browser window with the topic page.

During 2005 made available what is now known as the AnswerLinks feature: web site owners could point to individual AnswerPages and make the links stand out by using specific attributes of the <a href> element, such as the green dashed underlining and a mouse cursor in the shape of a question mark. Still, this feature requires some effort on behalf of the web site author and it doesn’t seem to have been widely adopted so far.

What I was asking for was basically a Javascript implementation of the 1-Click Answers behavior: people would be able to alt-click or double-click on any word in a browser window and then miraculously a widget with the relevant topic page would appear. No installation should be required on the end user’s side. Web publishers would only have to insert a piece of Javascript in the <head> section of their source code and needn’t worry about proper HTML mark-up after that.

Having dismissed the whole idea over time as being unfeasible, I still couldn’t resist bringing up the idea once again in a chat with Gil last August. Gil replied immediately, mumbling something along the lines of: " … wait until you see what we’re working on."

After Gil’s unmistakably clear hint I didn’t let go of the idea and kept reminding him that I was still interested. Then ‘finally’, last October, Gil pointed me to the CBS News site and said "go to their site, and click on any word". I was flabbergasted and kept nudging Gil to let me blog about it. Last week I received the long-awaited go-ahead—some browser-specific glitches had been ironed out and I could get started.

Implementing AnswerTips on CleverClogs took me less than two minutes: I had thoroughly studied the instructions that Michal Gorlin Becker of had sent me. It was a mere cut and paste operation: I selected the piece of Javascript and put it into my TypePad templates, made sure I adjusted the value of the ANSW.nafid variable to the unique one that Michal assigned to me, rebuilt my blog and that was all there was to it. also provided an animated GIF that I put in my sidebar to indicate that my site now offers double-click functionality.


Pre-selecting from Homonymic Definitions
Some topics, names and abbreviations are so ubiquitous that you may want to point to a particular topic page: in such cases publishers can override the default behavior and use a disambiguating AnswerTip hyperlink using the following HTML code:

<a href="" 
class="bulbLink" onclick="return ANSW.b5.SendQuery
(this,'harry-potter');">Harry Potter</a>

Applied to the name of Harry Potter a direct link to his topic page would then appear as Harry Potter

Some afternotes

Of course AnswerTips are enabled on the website itself. I trust they will soon put up a page that lists other AnswerTips-enabled web sites.

Other tools that I find very useful are the Firefox extension (Windows, Mac OS X and Linux), which enables AnswerTips on any website, and
1-Click Answers (Windows, Mac OS X), which enables AnswerTips within all of your desktop applications.

It’s probably needless to say that I’m very impressed with AnswerTips. Please let me know if you agree that it’s an improvement to CleverClogs. I will definitely pass on your remarks to the development team.

Written by CleverClogs

December 22nd, 2006 at 4:25 pm

Posted in Search Engine News

Search Clouds Increase Relevance of Search Results

with 3 comments

On reflection it’s actually striking that so far Google hasn’t embraced any kind of topic clustering functionality outside of its ’Similar Pages’ feature. I’m not the only one to feel this need: recently Arc90 lead architect Joel Nagy also investigated into this area and quietly published about the research he did into contextual relevance of pages found through search queries, with ’Search Clouds’ as the central idea.

No matter how precise your keyword query, search engines don’t always produce the type of results that is relevant for your research goal. So far I’ve constructed hundreds of compound and advanced search queries myself and each time I’m surprised about the noise that inevitably slips into the search results. Some of these erroneous results can only be detected with the human brain. Yesterday, for example, Fred Zelders sent me an email message to inform me about an irrelevant search result in one of the blog search feeds in RSSonate, my RSS-in-the-blogosphere monitor. I really needed to read the post to which he referred twice before I could confirm he was correct in his observation:


Surprisingly RSS in this case stands for ’Random Shutdown Syndrome’, an ailment from which quite a few Macbooks notebook computers seem to suffer, where apparently they start to randomly reboot themselves. On further investigation I was actually stunned to discover that AbbreviationZ (an afiiliate of lists 40 additional ways to resolve the abbreviation RSS.

Fred’s feedback is expecially valuable to me because he is very knowledgeable about RSS and OPML technology. I know he subscribes to RSSonate ever since I launched it and in turn I often visit FeedFiles—a comprehensive repository of RSS tools—to see if he or his son perhaps already reviewed.the RSS tools I’m discovering. Fred’s blog Fred on OPML is in English, his other writings seem to all be exclusively in Dutch.

Over the years I’ve seen a couple of initiatives and attempts to cluster search results around topics of interest. Clusty immediately comes to mind of course. Another way to improve the precision/noise ratio is by using advanced syntax techniques, like category, tag or in-title search.The major search engines and some of the blog search engines offer query modifiers like these to improve the search results. Other engines allow you to indicate the scope of your search: for example, commercial vs non-commercial search results. If I recall well Yahoo! offers this, but as a search engine it just won’t click with me.

On reflection it’s actually striking that so far Google hasn’t embraced any kind of topic clustering functionality outside of it’s ’Similar Pages’ feature. I’m not the only one to feel this need: recently Arc90 lead architect Joel Nagy also investigated into this area and quietly published about the research he did into contextual relevance of pages found through search queries, with ’Search Clouds’ as the central idea. Joel mocked up this impression of what a Google search query containing the words ’Nintendo’ ’Wii’ ’launch’ and ’date’ could look like if it were accompanied by Search Clouds:


I like the implications of Joel’s research: it clearly shows how useful it would be to have his concept of Search Clouds implemented in search engines, giving an immediate clue which page could be most relevant to my research goal. Imagine that in my search engine profile I could also maintain a persistent list of tags and keywords; Search Clouds would then be able to visually indicate to what extent a search result is likely to match my profile…

The Arc90 sandbox Arc90 Lab has been on my radar ever since I discovered their Link Thumbnail feature, about which I blogged a few months ago in my blog post Pull Quote Mystery. It’s rewarding and encouraging that Chris LoSacco from Arc90 indeed followed up to the promise he made in the comments section on CleverClogs, saying Joel would continue to improve the usability of Link Thumbnail: in the blog post Updated : New Link Thumbnail Goodness Chris announces that they now rely on the new thumbnail service WebSnapr instead of Alexa.

Note that there’s no download or web service to Joel’s project yet. I would appreciate it if he’d continue working on Search Clouds. I’d be delighted to provide extensive feedback if and when he needs it. Tell me, what techniques do you use to fine-tune your search results?

Technorati Tags: , , , , , , , , ,

Written by CleverClogs

October 23rd, 2006 at 4:14 pm

Google Ajax Search

without comments

This is where I’ll be playing with the new Google Ajax Search applet.

I’m trying to find out what makes the code appear and work fine on my home page, but not on the individual-entry pages.

I’ve now moved the code from my sidebar to the main container to see if that makes a difference.

UPDATE Sept 12, 2006 – 11:24 PM GMT+2:
I restored the original setup. Moving the code from one column to another didn’t help me much.

It seems the TypePad comment-form template module contains or calls a piece of code that prevents the Google Ajax Search box from being displayed. When I disable this particular module, the Google search box pops up just fine. Something tells me the onLoad() event handler, on which Google Ajax Search relies, is reset through the authentication script in the comments section.

I’ll submit a TypePad support ticket and use this blog post to report the outcome.

UPDATE Sept 12, 2006 – 11:43 PM GMT+2: Literal text of my support request ticket:
"As you can see at the top of the sidebar of my tech blog CleverClogs (, I’m trying to implement Google Ajax Search. I’ve got it working just fine on the main page, but on subpages for individual entries the search box refuses to display. The Google Ajax Search API key is valid for my domain and its subpages.

I did A LOT of background research before submitting this request for help. I’m now pretty sure the cause of the problems lies in the default comment-form template module. If I prevent that module from loading within the individual-entry template, Google Ajax Search appears just fine. It looks as if the comment-form steals away the onLoad event handler.

I think once we’ve got this sorted out, Six Apart will be able to offer an extremely attractive Google Ajax (Blog) Search widget that can be implemented in a typical sidebar TypeList without much difficulty (except for a Google stylesheet that needs to load from the page header of course).

Is there anyone on your team with whom I could talk this particular issue through?

I hope you will consider to take the time to handle my request.

With the best of intentions,

Marjolein Hoekstra"

UPDATE Sept 13th, 2006: 05:20 AM GMT+2 My TypePad ticket is picked up by Jen, with the remark that there’s a glitch in the way I coded the closing <div> and if fixing that might make a difference. Jen also asks how I implemented Google Ajax Search. I immediately fixed the typo but it doesn’t make a difference. It makes sense to build a new weblog from scratch. My first document on that site will be a list of instructions on how to implement Google Ajax Search on a TypePad blog.

UPDATE Sept 13th, 2006: 11:20 AM GMT+2 The new blog is live: Google Ajax Search. Please check it out if you want to know the details of this particular project.

UPDATE Sept 13th, 2006: 07:05 AM GMT+2 Submitted my reply to Jen:

"Hi Jen,

Thanks for your fast response. Embarrassing glitch you found in that /div tag, but unfortunately fixing that didn’t lead to resolving the issue.

Your kind reply inspired me to dig pretty deep today and I figured I wouldn’t be a blogger if I wouldn’t publish my findings. I worked on this all day, so it better be good this time.

So, please go here: to see a journal of my findings debugging this divine widget and then move over to the ‘real’ thing here: a new blog completely devoted to Google Ajax Search:

Note that there are two blog posts there, of which the oldest one remains on the home page as a featured post. I figured TypePad Pro users would most easily find it this way. This post has comments enabled.

In the sidebar you will find a Recent heading that mentions the other post I just created, with comments disabled. You’ll immediately notice the different effect that enabling comments has on the Google Ajax Search box.

I hope you like what I’ve been trying to do. Please note I’m a very avid TypePad user, very loyal and very eager to learn (and probably most from my own mistakes).

Do let me know what you think of all this. Feel free to pass this on if there’s anyone else on the team who might be interested as well.


(P.S. in some situations it can be much more efficient to exchange ideas over IM. Should you or someone else be inclined to this: I’m Chopianissima on most IM/Skype systems)

FINAL UPDATE: September 20th, 2006: on Friday, Sept 15th, TypePad Support got back to me with their final, rather disappointing verdict: they say they are currently not able to alter the comments script so that it will work with Google AJAX Search. A Javascript nitwit, I have now put all my hopes in Marc Lucovsky and his development team at Google. I posted a note on the Google Ajax Search API Developer support forum pointing to my new Google AJAX Search blog on the TypePad system and mentioned the problems I was running into while implementing it. Then Mark was generous enough to not only correct my code so that it reflects the API’s current feature set (it’s adjusted now), but also to start investigating if the issues I’m having can be fixed on the short term. Hopefully this story will have a happy ending after all.

Final UPDATE: September 22th, 2006: Mark Lucovsky left a short, but oh so meaningful comment on my Google Ajax Search blog. Mark wrote:

Take a look:

I’m so impressed. Basically, Mark signed up for a blogging system with which he wasn’t familiar, worked his way through the Advanced Templates mechanism, and solved the issue I was having to get Google AJAX Search to behave with TypePad comments (or vice versa).
I’m going to be out of town for a few days. After that I’ll fix the code to my own blogs and also update the instructions for everyone else to enjoy. Great work, Mark.

Written by CleverClogs

September 12th, 2006 at 6:27 pm

Posted in Search Engine News

Technorati Tags Delivery Secrets

with one comment

Have you always wondered which top-notch delivery protocol the Six Apart programmers implemented to get your tags stored safely and securely into the massive Technorati database servers?

The answer to that question is finally here, on Everything TypePad today: the blog post Sending your blog by special delivery features a special video documentary (1:57 mins) revealing inside secrets to get the job done:


Now that we’re talking Technorati-speak anyway: have you tried clicking on any of the tags at the footer of any of my own blog posts yet? Instead of performing a generic Technorati search, each tag hyperlink delivers Technorati search results from CleverClogs only.

The CleverClogs comments feed in my sidebar is new too, so is the coComment conversations widget. More about coComment at the Pull Quote Mystery conversation. This blog is definitely rolling!

Besides email, would you prefer other methods to connect to me? A formal contact form?  What else?

Written by CleverClogs

August 24th, 2006 at 11:17 pm

Keyword In Context Search: SurfWax LookAhead

without comments

"LookAhead for Blogs is exciting because its keyword-in-context search considerably improves the chance of readers opening posts from my site that are highly relevant to them."

Actually I was doing background research for a future blog post about the LiveSearch beta service from Yahoo!, when Gary Price of ResourceShelf caught my attention with a mention of SurfWax LookAhead in his overview article Results BEEFORE Clicking the Search Button. I had come across pretty impressive mock-up demos of LookAhead technology in the past, so I was curious to learn what’s new now.

It turned out that the SurfWax people made a bloggers’ version of their software available called LookAhead for Blogs. I signed up for the service straight away because I see enormous potential in it. LookAhead for Blogs creates a full-text index of the items in your RSS feed and makes them accessible through a keyword-in-context search applet.


Take a look at the simple search box at the top of my sidebar: just start typing in that box and you will see a dropdown box with a list of all articles that match your search string. There’s no search or OK button: the results show up immediately and dynamically thanks to the use of Ajax technology. The more items in the feed, the more hits you will see.

LookAhead for Blogs is exciting because its keyword-in-context search considerably improves the chance of readers opening posts from my site that are highly relevant to them.

To get a few background details straight I had a brief email conversation last night with Tom Holt, CEO of Surfwax, in which he told me that SurfWax is working on incorporating blog archives into its lexicon indices. Tom also pointed me to a another of their services that just went live this week—it’s called ShopEasier and it allows you to shop at for products available through sites like Froogle, Nextag, Yahoo, Amazon, Become and Shopzilla.

To implement Lookahead on your own site, sign up for a free Lookahead for Blogs account and provide the url to your RSS feed, your email address and desired password. Note that any feed may be registered only once. Immediately after completing the sign-up process you will receive an activation link by email that leads to a web page with detailed installation instructions.

Summarizing the installation instructions, you need to create a sidebar item with a piece of code that creates the search box and a script tag that initiates the search index.
This is the code I used to create my LookAhead ‘searchlet’ for CleverClogs:

<form name="search">

<input type="text" name="term" onKeyUp="changeVal();" style="width: 200px;">
<em>Powered by <a href="">SurfWax LookAhead</a> </em>

<script src="" type="text/javascript"></script>
<link href="" rel="stylesheet" type="text/css">
[cloaked ID]',, 270, 120, 'left');

The text "Powered by SurfWax LookAhead" was added by me as a courtesy.

Written by CleverClogs

August 6th, 2006 at 9:19 pm

Posted in Search Engine News