[August 25, 2011 -- 12:30 PM -- T. Hayes    ©2012  Codentity LLC]

The Quest For The Ultimate Search Engine

Imagining A More Efficient Search Methodology

Where Current Technology Might Be Improved

Questions Going Forward

Real-World Scenarios

The Power Of A Unique Personal Profile

Music Search Embodiments

Anticipated Actions

The Quest For The Ultimate Search Engine

Historically, typical Web search events have been formed from “scratch” each time a user entered his query keywords at the home page of the chosen search engine.  Search engine products had no way of positively identifying a specific individual, and then restricting search results to topics, or range values, or other data that the user was CERTAIN to want.

There have been variations on the keyword theme with the addition of enhancements like collaborative filtering and, in some cases, an attempt to guess a user’s future intentions based on the examination of a transaction log that records previous search keywords, or the address and content of previous Web destinations executed on a particular computer.

Internet search is an amazing tool and most people are probably satisfied, if no longer impressed by what it can do. However, in our opinion, the “ultimate search engine” has yet to be deployed.  We think there is room for improvement and, in this blog report, we present our case for building superior Web search and music search products—NOW.

[Discover an advanced Codentity music search initiative here.]

Imagining A More Efficient Search Methodology

In order that everyone may benefit from faster, more relevant, more consistent search operations, shouldn’t someone build a demonstrably superior search product?  Could that superior product be based on a “Persistent and Unique User Profile” methodology?

In this report, the term persistent is intended to mean a consistently reproducible process, while the term unique refers to an absolute identification method (such as that potentially embodied by the linking of a mathematically distinctive number with an individual’s true name, and possibly a corresponding user alias).  And, finally, the term profile indicates a compiled, stored data repository consisting of personal preference declarations and other biographical information which can be linked to the unique user—for recall on demand.

To elaborate further, the ultimate search engine should be able to:

Would this methodology save time, and promote greater precision in search results?  We assert that, if properly implemented, it would produce immediate benefits.

Imagine Web search software that could always positively identify, and then persistently connect with, a demonstrably unique user anywhere in the world.  Imagine if that user had access to interface tools that let him or her select, and then optionally apply, a layered and configurable level of filtering across their choice of demographic and psychographic attributes.

Imagine further how much more efficient a search process could be if every time a unique user logged on to a search engine, the search engine had immediate access to that user’s declared preferences—and those preferences could be used to filter IN, or filter OUT, data hits across a diverse range of attribute tests associated with the query topic.

Imagine a search product that correctly predicts anticipated query results based in part on the biographical data stored in each unique user’s profile.

The design concepts for this type of methodology have been published in the form of a patent disclosure, and they are waiting for the product development team at an aggressive, objective-inspired company.

Where Current Technology Might Be Improved

The motivated reader might pose the question: “Don’t search engines today have the ability to identify a user?”  A sincere answer to that question would depend on one’s definition of the term “identify”.

While it can be true that most major search engines seek to identify a user in some manner whenever a user arrives at the search home page (SHP), this so-called “identification” is most-often made by evaluating a stored “cookie” on the user’s computer.  The cookie is a small file that tells the search engine that the connected COMPUTER has visited the site previously.

In this sense the user—an individual person—is not really identified, and, instead, his device is merely recognized.  However, some search engines are taking the identification process one or two steps further.

Users can opt to sign-in at Web search sites.  They enter a name, password, and possibly additional non-intrusive information.  This is a way for the user to notify the Web site that he or she is back, and is willing to allow the Web site to maintain a history of that user’s page and topic visitations, as they traverse the world-wide Web.

This methodology certainly adds several degrees of enhancement to the identification process.  But, since there is most probably no irrevocable verification of that "signed-in" user’s name, and no way to guaranty successful analysis of that user’s preferences by observing only where he or she has been, this level of identification cannot be precise.

Contemplate this simple analogy:

Commuter A, a man who rides the train to work each weekday, might visually recognize a certain woman, Commuter B, as the same woman who boards his train car each morning.  But recognizing her on sight is not equivalent to actually identifying her by name.

And yet, even if Commuter A somehow learned the name of Commuter B, this would not be the same as actually knowing her—i.e., understanding her personal preferences on a range of topics from gourmet foods to financial instruments to political philosophies.

It is our hope and assertion that as search methodologies evolve, they should seek to become better at KNOWING a customer (the browsing user).

By the term knowing, we intend to represent a meaning that combines the implicit and explicit definitions of the following words and phrases at the same time:  understanding, predicting, suggesting, locating, rating, classifying; ignoring unwanted aspects, prioritizing desired attributes, responding dynamically to the needs of the user, how to initiate collaboration with each unique user, allowing profile access to user-designated participants as desired, and so on.

There is no doubt that today’s search engine products from Google™ and Microsoft® are quality tools—especially when compared with the state-of-the-art in years past.

But is Google search as good as it could be?  Is Microsoft Bing search as good as it could be?  Should Google search be better?  Should Bing be better?  Can Google and Bing be improved?  These questions may be relevant, but they are also too vague and forgiving.

Questions Going Forward

Looking ahead, here are the questions that, if answered and acted upon, might be more beneficial to the millions of unique users in the world-wide search community:

Recently, Google™ announced improvements to their “personalized Web search" tools.  And, Microsoft Bing™ has promoted refined “music search” capability.  This is good news.  But do their announced improvements go far enough?

Real-World Scenarios

Out in the “Webosphere”, millions of individuals turn to search engines every minute of every day and, as described in the paragraphs above, for most people, the search experience is likely not uniquely tailored to their requirements.  It probably involves recreating their search data, and reshaping their search preferences (if any preferences can be known with precision), with almost EVERY SEARCH EVENT.

Yes, Google’s personalized Web search tools offer some degree of user correspondence, but the fidelity of that user-connection (and thereby the implication of a user understanding) may be unclear and based, in part, on an analysis of a computer’s browsing history (where the hardware is observed, and the true affinities of the person remain unknown).  In looking at sites visited or search terms submitted, can such tools, if fundamentally tied to hardware-tracking for visited sites, absolutely differentiate between two or more different people who share the same computer?

Consider a husband who "searches and purchases" sport-fishing accessories online—and then surrenders the family computer to a wife who searches for, and subsequently visits, Web sites that offer provocative lingerie.  This scenario is not so hard to imagine, yet it could present an evaluation problem to site-logging methods.

Would the Web tracker algorithms conclude that the unspecified (i.e., indeterminate shared) user is obviously an avid sports fisherman who also enjoys wearing intimate ladies apparel?  And if a teenaged son or daughter also has access to the same computer, could the aggregate profile assumption (for the user who, remember, is likely not uniquely identified on that computer) be erroneously classified as—a customer who revels in sport fishing, wears ladies undergarments, and listens to the reigning pop music stars of the day?

Browsing history and destination analysis appear to be fairly passive methods of making assumptions about a user’s inclinations.  In the absence of a unique user profile that persistently maintains user-configured filters with declared preferences, it is possible to create false assumptions because the software has no pre-existing knowledge of a certain unique user’s openly declared affinities.

The Power of a Unique Personal Profile

So, historical analysis may be a reasonable start, but my company’s “personal profile” queries (described in detail in our 2009 patent application) offer what we believe to be a more efficient active system.  We claim that our query formulation methods deliver true innovation via pre-qualification filters that conform to an individual’s expressed interest.

In fact, the claims in our public patent disclosure facilitate a personalized search experience based, in part, on an individual’s declared preferences, which are categorized and quantified with the collaboration of the user.  Additionally, we teach a method of persistently connecting each portable query with a demonstrably “unique” user anywhere in the world, on any platform, at any time—with or without an Internet connection.

The discussion of the primary utility embodied in our application is focused on creating a system of queries that can collectively predict an individual’s affinity for “hit music” recordings.  However, we also believe our method and system could be instrumental in development of search technology designed to accurately assess an individual’s expectations, and then forecast and shape the relevancy of the results for any Web search initiated by that user.

Music Search Embodiments

This blog report has primarily focused on certain aspects of query construction and user profiling as disclosed in our patent application—


Patent Application number: 20100063975 (with a parent filing date back to October 2004)

—and how future Web search implementations modeled on those concepts might serve to improve Web search across the board.

It should also be made clear that we believe that business operations affected by MUSIC SEARCH could also benefit greatly from our designs.

Since music search execution is the crucial first step taken by each customer in the $8 billion dollar annual U.S. music market, the improvements envisioned by our system could provide a distinct advantage to the end user, and to potential system licensees.

Motivated music sellers who wish to utilize the system to direct customers to their Web stores, where sellers can efficiently close music sales transactions, may recognize the strategic benefits offered by our system, and be willing to pay licensing and transaction fees.

Additionally, the major online music sellers might choose to use the system’s design to augment their currently deployed search methodologies.

Our disclosure envisions implementation via a Web services / eCommerce model that is likely to attract millions of potential customers who would benefit from its uniquely-customized search methods.  The system and methods in the invention collaborate with users to eliminate search fatigue because customers can intuitively locate desired content, based on calculations made from information stored in their “personal music profile.”

At the same time, music vendors would profit from the system’s ability to efficiently direct user content focus.  This precision targeting means the user is more likely to accept presented selections (recommendations), and complete his purchase of recorded music products.

The system disclosed in the invention could be crafted as the interface that fundamentally determines customer behavior with respect to initial product discovery, and their subsequent desire to acquire specific selections.  The system might quickly achieve universal adoption because it benefits both parties to a potential “search and close” transaction.

By “universal adoption”, we mean that the system could be tailored to supplement the existing query methodologies of all leading music sellers (such as iTunes®, Wal-Mart®, Best Buy®, Rhapsody™, Yahoo® and Amazon®).  With the appropriate technical and marketing incentives, the system could appeal to a majority of their established customers—while it serves to invite new shoppers.

These objectives are admittedly ambitious.  However, we believe they can be achieved because our system overcomes observed deficiencies in at least 3 key areas of the current art: customer collaboration, music affinity calculation, and profile transportability.  Put simply, our system will help participating music vendors deliver an improved customer experience.

[Discover an advanced Codentity music search initiative here.]

Anticipated Actions

It is our hope that one of the serious Web search, or music search, innovators will commit to reviewing our disclosed methodology, and in so doing, become the company inspired to create a more refined, user-connected search product.  We also contend that such a company might gain an advantage that could all but guaranty users a more constructive, more relevant, more efficient search at every execution.

Offering users such an advantage might be a sure way to build product loyalty.

From the desk of
Thomas J. Hayes
Managing Partner
Codentity® LLC

Discover an advanced Codentity music search initiative here.

The opinions expressed in this report are Copyright 2012 by Codentity LLC.  Qualified individuals who wish to respond to expressed opinions, or to correct potentially inaccurate assumptions made in this blog report, are encouraged to contact us:  editor@codentity.com

Go to Codentity home page 

NOTICE:  Microsoft and Bing are trademarks or registered trademarks of Microsoft Corporation.  Google is a registered trademark of Google, Inc.  All other trademarks, including (where applicable) listed song titles and their corresponding artist names, are the property of their respective owners.