A reissued patent for Apple (AAPL), just made public today, covers a way of employing profiles of users' interests to help rank the relevance of search results. The language of the claims is broad enough in scope as to potentially cover what such companies as Google (GOOG), Amazon (AMZN), and even Facebook do. If unchallenged, it could provide Apple with a significant bargaining chip in business negotiations with many other companies.
A patent holder can ask the US Patent and Trademark Office to reissue a patent to correct mistakes in the original or, if filed within two years of the original patent's grant, to broaden the claims and, therefore, what the patent covers. Apple's original patent number 6,202,058 was titled System for ranking the relevance of information objects accessed by computer users. Originally filed in April 1994, it was granted on March 13, 2001.
On March 12, 2003, one day before the critical two-year anniversary, Apple filed for a reissue. The pharmaceutical industry more commonly uses reissued patents because of the length of time it takes to understand what new compounds can do and the number of previously unconsidered applications that can come to light. Reissues are less common in high tech, possibly because the speed of change and the brief time to go from invention to marketed product doesn't offer as much advantage.
However, between 1994 and 2003 was an enormous stretch in computer technology and was enough of a span for Apple to see the explosion of the Web and the rapid ascent of online commerce, search engines, Google, Amazon, early social networks, and far more. The ability to expand a patent that could cover matching information with stored consumers' interests could be a strategic gold mine.
The reissued patent, number RE41,899, has the same title as the '058 patent, but now takes precedence. Look at the first claim:
In a computerized information access system, a method for presenting items of information to users, comprising the steps of: a) storing user profiles for users having access to the system, where each user profile is based, at least in part, on the attributes of information the user finds to be of interest; b) determining an attribute-based relevance factor for an item of information which is indicative of the degree to which an attribute of that item of information matches the profile for a particular user; c) determining a measure of correlation between the particular user's interests and those of other users who have accessed said item of information; d) combining said relevance factor and said degree of correlation to produce a ranking score for said item of information; e) repeating steps b, c and d for each item of information to be presented to said particular user; and f) displaying the items of information to the user in accordance with their ranking scores.Aside from the general broad application of user profiles to finding information, here are some details that strike me as being of particular importance:
- A "computerized information access system would cover virtually anything, including relational databases, flat file storage, Web sites with HTML sites, e-commerce catalogs, or collection of advertisements.
- "[P]resenting items of information to users" is also incredibly broad. There's no indication that the user has to actively seek information, and so it would cover any system that pushed information to people (advertising network) or one in which someone sought information (search engine).
- "[E]ach user profile is based, at least in part, on the attributes of information the user finds to be of interest" doesn't say that the user has to fill in a list of interests, though that clearly would be part of it. I suspect that collecting information about users, like watching what they click on and storing it, would count as a way of building a profile on the attributes of information that people find of interest. In other words, this could cover every type of behavior watching that marketing systems do.
- Determining what interest "attributes" information possesses and then comparing those to a given user's profile for a degree of match covers much of what happens on the Web.
- The system correlates one person's interests to those of other users and what they've looked at, creating a learning method that has become common in information retrieval of all sorts.
- The system combines both the direct comparison of information attributes to a single user's profile and the correlation between what interests that user and what information others, with similar interests, have looked at to create a relevancy score. There's no demand that the two be combined in any particular way, and it might include ignoring one or the other.
A method of presenting documents from a document collection to a user, the method comprising: storing a user profile vector for the user, the user profile vector in a vector space derived from terms contained in the document collection and including a plurality of weights, each weight associated with a term in the document collection; selecting a plurality of documents from the document collection, each document associated with a document vector in the term vector space; for each selected document: determining a relevance score, the relevance score based on a relationship between the user profile vector and the document vector associated with the selected document; determining a correlation score between the user and other users corresponding to the selected document; and combining the relevance score and the correlation score to determine a final ranking score for the selected document; and presenting the selected documents to the user according to the final ranking scores.A collection of documents? Sounds like many things, including a set of product descriptions or an assortment of Web pages that you've indexed against keywords (which, after all, help indicate user interests). The problem for other companies is not having a collection of documents, per se, but of leveraging information about individual users -- a big and important activity for Google -- to help decide what documents to deliver. The rankings by interest probably don't have to be the sole criteria in use.
Now notice that all this now dates back to 1994, before Google, Amazon, Facebook, ad networks, Internet commerce, or many other things that we take for granted. This patent almost predates Web search engines. Not a bad thing to have in your back pocket when Amazon and Google, among others, are your major competitors.
- Google Plans a New Type of Extended Display Search Ad
- Google Develops Storage Technology; Time for HP, EMC and Others to Get Nervous
- Twitter's Patent Strategy: Don't Bother.
- Patent Gives Facebook Broad Power Over Mobile Location Networking