Search Intent Taxonomies
Oct 21 | 2008
fleeQ search engine will give you exactly what you want.
“The perfect search engine,” said Google co-founder Larry Page, “would understand exactly what you mean and give back exactly what you want.” How to define ‘want’ in search? It is intent. Simply, Larry is stating that search engines must be capable of predicting intent. How can intent be best anticipated, and discovered through Search? Is it important to improve search services to address user intent better? Is there more than
“one way” to do so?
Intent:
Broder, a highly respected and knowledgeable IBM engineer in the scientific analysis community[1] states that user intent, in regards to Information Retrieval (IR) and specifically web search, classifies under three taxonomies, informational, transactional, navigational. Informational is defined as the intent to acquire some information assumed to be present on one or more web pages. (e.g. search terms “how do i bake
a cake” or “origin of the species”).
Web search terms signify intent, and trigger the results which search engines return, yet there is often a cross over when it comes to search terms and intent. This can be easily illustrated with the search term “cat food”, where one might assume that the search results providing locations and websites to purchase “cat food” may be relevant. Yet, the individual may actually be looking for a song by the title “cat food” or even the
process that “cat food” is made. Maybe the individual is looking for the web site www.friskies.com, or even a blog called www.ieatyourcatfood.com. The “cat food” example shows intent can be elusive as a butterfly.
Intent is not always fully exposed through a user’s choice of search words or phrases. The search engine can, in such a case, only try its best to provide relevance, with oftentimes very little information about the ‘perfect intent’ in the mind of the user. This means, for web search users, there will always be some plain luck often involved in receiving relevant search results, responsive to their intent.
Relevancy:
Is there more than one way search engines convert search terms into relevant results? Absolutely, and some of them are well defined in the book, “Google’s Page Rank and Beyond: The Science of Search Engine Rankings” (2). This book goes in depth as to how a search engine determines what results to show a user. The methods are far reaching and the effects of each algorithm are even more astounding when we look at the various aesthetic display techniques used by various engines.
Effectively, each of the engines display their results in a variable manner as to respond to the desires of the user, attempting to reach a level of information retrieval that coincides with the original intent of the user. Alas, the reality is this task is impossible to do perfectly, because it is virtually impossible to build intuitive perfection into mathematic algorithms, as people do not always submit the best search terms to signify their intent. That being said, some search engines do predict intent very well, others less so.
Many search engines use arcane methods of predicting relevance. One of the most popular is the “democratic method”, which is a factual, almost second-nature of the web linking model. Tim Berens-Lee designed the hyperlink technology at CERN. Imagining the hypertext to link one document to another. “Democracy” comes in the form of links, in that, other publishers/webmasters who find your content valuable and relevant to a subject will link from their website to your website. Google uses this “democracy” model, and although it can be fooled or cheated to drive the display rank of a page higher in the organic results provided by their search engine, it generally works for the most part.
Although the mechanisms are a bit antiquated, the masses have accepted the democratic method as a standard which they are accustomed to seeing results delivered for informational web searches. Even the “democratic method”, though organic, must still function within the algorithm paradigm, that a users search intent can become more and more fuzzy the more vague their query terms. In theory, the more search terms generated in a search query, the user’s intent should become more clear, and for the most part the results more relevant.
As for adding more terms (what google expects the user to do) will provide fewer organic results from the search engine, thereby narrowing the users targets and options. The dilemma of intent, however, is still not served completely with only organic results, nor even with contextual advertising results on the top and side of the google search results pages.
Google’s, as with many search services, approach towards predicting intent and displaying results is to deliver results that can be monetized by the Google’s investment in indexing content into organic results which feature their respective ads on their websites. Obviously as challenging as it is to predict intent, no one style of search service result deliveries can or should satisfy everyone. Some people are fine with the organic result and contextual ad mix delivered by Google and other traditional search
engines.
However, the antiquated assumptive behavior of the traditional search engines like Google, MSN, and Yahoo, where they believe that interactive user choice and action is irrelevant, may not satisfy the contextual needs of everyone at any given moment. In effect, what fleeQ has built is a new way of delivering deeper and richer context and relevancy without giving up what most users already expect from their search experience. This is a much better method to use in determining intent than simply organic results as ranked links on a page, served up with ads on the side.
Taxonomies:
Back to Broder’s three search taxonomies: having discussed Informational, let us briefly define Transactional and Navigational. Transactional is the intent of the web user to perform some web-mediated activity. (e.g. search terms “buy running shoes seattle” or “download metallica album”), Navigational is an immediate intent is to reach a particular site. (e.g. search terms “msn.com” or “cnn website” or “Oregon State University”)How can we appease these unknown needs of the user?
To improve web user satisfaction with a search experience, the burden to satisfy web users is upon search service providers to contemplate that these three search taxonomies must be pro-actively available and receptive to meet intent, as intent of a web user may variably shift between any of the three at any moment. A search service should effectively deliver a blend of the informational and transactional taxonomy results in manner that encapsulates user intention via an easy to digest search result display model. To do this, requires a search engine display format to offer the best results possible to a user.
A search service should utilize the best boat to take the user on a fishing trip. Allow the search engine to fish their intent, through behavior, and then predict what type of fish they want to pick from the net. To provide this opportunity, our fleeQ search technology sought to improve how results are displayed by utilizing intelligent taxonomy synonyms, selecting the most appropriate search engine to request from, thereby better addressing the taxonomy dilemma to provide the most relevant results to a web user.
Selectively omitting search engines that are geared towards non taxonomy relevance, our fleeQ technology effectively pattern matches the results based upon consumer desire, past history if available, and allows the consumer choice to define the scope and relevancy of results through their actions. Thus allowing us to reflect the isolation of their intent through behavior analysis.
We have discussed informational and transactional taxonomy, but there is a third leg of Broder’s stool, the navigational aspect of how users interact with search. Albeit, many search engines provide navigational matches in the organic search results, it is not possible to truly deliver a navigational intent with out disrupting the users expected outcome. The navigational aspect of search has hereto been addressed by Google
through the “I’m feeling lucky” button found on the Google home page. Although it delivers a relevant and contextually appropriate destination URL based upon the Google search results algorithm, it is is lacks providing other results from the other taxonomies. It operates on only one Broder taxonomy leg as it navigates the browser to a URL page result only. You won’t see the Google search results page at all. FleeQ search, however displays the whole Broder stool, with results of three taxonomies simultaneously, a relevant navigational result, transactional and informational results and advertisements.
DISPLAY OF RESULTS:
FleeQ search delivers a taxonomy trifecta of results into a single action which integrates a non-pretentious “tabbed” and super imposed display approach to search result delivery. A user can interact with the various search tabs, based on their intent of wanting informational or transactional results, or with the page result displayed behind the search results, if they possibly they are looking for relevant navigational results. Furthermore, users can interact with contextual advertisement results incorporated into our display model, for possibly more transactional and informational results. Search engine services can never truly capture the illusive butterfly of user intent. fleeQ was designed to deliver, in a single action, a vehicle by which a search engine can deliver their true search intent.
What’s more, is that although the “democratic” method for displaying organic website link results is seemingly fair to publishers, and therefore assumed to be thereby most relevant by application of results to users, this process can be spoofed or cheated easily. This means that search results that could be available to users, from web publishers who lack either sophistication in the use of deceptive link-farm technology or
exposure and links from other websites necessary to rank higher in the organic results, will likely never appear as an “I’m feeling lucky” target URLs result for a Google web user.
It should not be seen as so radical a thought that both web users and web publishers should interface on the net in an egalitarian manner. A paradigm of the net should not forever remain vested in “Who is more important in web search, user or publisher? ” Why not both be held important? There is no actual justification for not empowering web
publishers (of varying scales) an opportunity for more egalitarian participation in search result display opportunities to web users. Status quo search delivery means that many perfectly relevant navigational or organic link results (for users) are not frequently delivered due to the lack of a level playing field for web publishers, inherent in the continuing operation of major search engine/s.
The fleeQ / adUup system is the first solution that comes close to truly actualizing and delivering a broader search result display solution in effort to raise the status quo of the display innovation stagnant search engines. The status quo is unsatisfactory for both user and web publisher, in that efforts of many publishers to produce and deliver quality content, products, and services goes unnoticed by the user who must rely on results being delivered to them by the status quo limitations of so-called “democratic” search engines.
adUup/fleeQ technology levels the search driven playing field, it is more egalitarian for both web user and publisher, in that it allows web-masters and publishers who have already traffic to participate in an opportunity for visibility before a greater potential of relevant web users. This is finally offered without the need publishers to purchase sponsored links so as to increase likelihood of exposure to relevant users. This is exposure issue is a critical concept to understand, publishers, who do not have much traffic, need it desperately, and having a mechanism by which they can get relevant contextual targeted traffic at no cost until now has been almost impossible.
Greater exposure opportunities to traffic is effectively what we deliver through our proprietary Ancillary Ad System. Like other search services, advertisements are a financially reality to operate adUup and fleeQ and therefore necessary, but not necessarily an evil. It is not inexpensive to deliver the search service, as the the cost of servers, and bandwidth is a market commodity, and not inexpensive. One of the unseen activities and costs of search services, is the creative minds behind the code. The best engineers needed to design search algorithms, for better and better targeting, is an extremely expensive payroll cost.
As an example of the value of algorithms, take a look at the value of a company, PowerSet. Launched in the first quarter of 2007, Powerset’s inventory is simply a set of algorithms and code on top of the wiki data, wiki data that is available freely to the entire world. Anyone can download the wiki data, because its open domain information. What the engineers at Powerset did do was to create some cool associate, contextual algorithm inventories, that sort and predict intent of the web user.
They did it pretty good, and recently, Microsoft bought them for over $100 million dollars. The “Power-soft” purchase most certainly exemplifies the enormous value of code, in and of itself, in “predicting intent” for search service providers. Powerset didn’t have any revenue, they didn’t have any traffic of any significance, the only thing they had was their algorithms. Access to the minds, the engineers that create algorithms is not inexpensive, as is continually proved in acquisitions, such as this one by Microsoft.
The industry gossip that Powerset code would be the Google Killer, ended up as a false hood. I do have to say that what they did do, albeit a small piece, they did well. They showed a new way of displaying results, and a different way of displaying them, that for me seemed pretty cool. But, in the end, PowerSet was a letdown for the fleeQ engineering team, who were looking and dreaming of the day a massive index farm (complete with spider and bots collecting the webpages of the world) would one day surface to put greater contextually and intelligence to the plethora of data that exists on the web.
As for how our advertising (which financially supports or search services) is managed for the best user experience, our search engine does one of two things: We locate a contextual advertiser who desires to advertise against that term, in which case we redirect the browser to that advertiser and display the xyz.com results atop the advertisers website inside of a SRW; OR In the case we do not locate an advertiser against the term, we send the advertiser back to the xyz.com website search page. Allowing the website to display results as if we were never even present.
Our Ancillary ad technology invention empowers adUup/fleeQ’s egalitarian search paradigm. Ancillary ad technology, (as defined in the patent written by myself, Dylan Rosario) is entitled, “Alternative method of retrieving and displaying requested URIs”. The new paradigm provides cohesive satisfaction of user search intent and addresses publisher desire for traffic/revenue via a unified search result delivery model. A model that can accommodate for the Broder’s search intent taxonomies in a single action.
A publisher who integrates our technology into their website enhances their opportunity for exposure to those web users who would have never otherwise found them in the democratic organic results at a mega search engine, nor via the “I’m feeling lucky” method of search engines, such as Google.
The way we deliver navigational traffic to a website is based upon our custom algorithm, it makes the event of searching by a user a traffic generating event. The process then delivers via the “I’m feeling lucky” model traffic to any arbitrary website that our search and targeting algorithms find contextual or relevant. In essence driving traffic to the publisher.
For instance, a participating publisher site at www.GYM.com earns an opportunity for his site to be delivered as the “I’m feeling lucky” Ancillary Ad, once his site generates three fleeQ user searches. In return, adUup locates a search event from one of our many other participating network publishers, and directs the delivery of an Ancillary ad on behalf of publisher www.GYM.com, as a viewable background page to a user searching with fleeQ in another site.
Search boxes featured on publisher sites create an opportunity to capture inbound related traffic and exposure to web users who otherwise would not have viewed their great message, product or service. Opportunity for this traffic is generated not only from other websites, but also the fleeQ.com search portal, and the fleeQ browser embedded toolbar and search boxes.
Publisher revenue opportunity (often siphoned away by status quo search engines) can now be captured through an egalitarian search delivery model that restores value opportunities to traffic otherwise lost. Although, web user search intent is illusive to capture some extents, fleeQ inherently provides traffic satisfaction for both the publisher, and navigational relevance delivery to a web user at the same time. The user gets great results from their query, and the publisher benefits further from his/her visitors.
If the publisher desires to not receive free traffic, and rather cash, we can direct that search to another publisher site and give the originating publisher in this case xyz.com credit towards receiving an Ancillary Ad from our network. Essentially, adUup utilizes the power of search as a force to drive revenues for publishers via a new paradigm. The ancillary ad. When you are at a site (e.g. xyz.com) and you search for either local (in-site) xyz.com results, or you desire web results (e.g. non-local as for example what you get from google) the publisher directs the search to our search redirector. We receive the search term directly from the xyz.com search form located throughout the website.
In the final analysis, adUup offers publishers options in regards to earning guaranteed revenue or traffic by offering web search from their website(s), has been enhanced via the fleeQ adUup technology. Furthermore by giving this new option to the publisher they do not interfere with the users search experience, because intention is taken into account more completely than that of any search engine today. This is something that no other ad network or search engine offers today. We are excited to be delivering a solution that truly “levels the playing field”.
Thanks for reading.
Dylan Rosario
adUup President - Founder and Inventor
What is search intent? What are the taxonomies of search?
http://www.sigir.org/forum/F2002/broder.pdf
Read what this guy this is the primary purpose of a search engine ….
LOL…. way wrong! This is what a marketer thinks.
http://www.build-a-home-business-website.com/purpose-of-search-engines.html
A bit more along these lines…
http://jimjansen.blogspot.com/2008/06/user-intent-in-web-searching.html
What’s next? Search 2.0 and beyond….
http://www.readwriteweb.com/archives/search_20_what_is_next.php
1)[Broder, A. 2002. A Taxonomy of Web Search. SIGIR Forum 36, 2, 3-10.]
2)Google’s PageRank and Beyond: The Science of Search
Engine Rankings (2) (Hardcover)
http://www.amazon.com/Googles-PageRank-Beyond-Science-Rankings/dp/0691122024
INTENT AND SEARCH : HOW FLEEQ ADDRESSES THE CONSUMER / USER SEARCH EXPERIENCE




Searching the Internet has never been better. One search to bring them all. Get the best of Google, Yahoo, MSN, Amazon.com, YouTube and more.