Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1999-05-06
2001-06-19
Choules, Jack (Department: 2177)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000
Reexamination Certificate
active
06249785
ABSTRACT:
COPYRIGHT AUTHORIZATION
A portion of this patent document contains unpublished material that is subject to copyright. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the patent file or records of the U.S. Patent and Trademark Office, but otherwise reserves all rights to copy or reproduce the copyrighted material in any form.
BACKGROUND OF THE INVENTION
1. Field of Invention
This invention relates to an apparatus and a method for reliably predicting subjective evaluations by a person and, in particular, to an apparatus and a method for predicting an evaluation based upon links established by previous evaluations by the person and other people.
2. Prior Art
As society's inundation with information continues or even worsens, mechanisms such as web browsers have been provided to enable technically unsophisticated people to use or to manipulate complex systems. Another such mechanism is a computerized system for preselecting items for a user based upon previous choices or explicit preferences from the user. Several such systems are described in the prior art and each suffers from the inability to reliably predict subjective reaction, i.e. a response based upon taste, not upon a need for factual information.
Subjective responses generally relate to objects of art, such as books, magazines, movies, plays, or music, including objects literally of taste, e.g. beer, wine, cigars or, more broadly, restaurants. The invention is described in terms of rating books even though the invention has much broader utility.
The prior art can be broadly divided between systems that focus on the user, e.g. trying to find users with similar tastes, or on the item, e.g. a popular book is likely to be well received by a new reader. A problem with focusing on the user is the fact that individual rating patterns can vary widely. For example, many people rate things within the upper portion of a range, e.g. on a scale of 1-10, most or all ratings are in the range of 6-9. A minority use the whole scale and another minority use a lower but narrow sub-scale such as 3-7.
Focusing on the user also has the problem that the results will be distorted by eclectic tastes. Trying to find a pattern among people who seemingly have no pattern to their choices is fruitless.
Focusing on the item essentially ignores the subjective tastes of people. For a database serving a large number of users, raw popularity data may be effective for a limited number of choices. For evaluating books, for example, where few may have read a given title, the sample size is too small to validly predict choice based simply on popularity.
U.S. Pat. No. 4,870,579 (Hey) discloses a system that focuses on the user in that one or more other users are designated as predicting users because of common choices. A weighting factor is included to adjust for differences in ratings among the predicting users.
U.S. Pat. No. 5,583,763 (Atcheson et al.) discloses a somewhat more straight-forward approach. A user's preferences are compared with others. When a significant number of matches are found between the two users, unmatched entries from a second user are suggested to the first user.
U.S. Pat. No. 5,749,081 (Whiteis) discloses a system that focuses on the items. The system makes recommendations based upon how many times an item known to the user is linked with other items not known to the user. The system searches through a list of items that are paired or linked with the items chosen by the user, along with the number of times each pair occurs. The items that are most frequently linked with the user's selected items are displayed as recommended items.
In view of the foregoing, it is therefore an object of the invention to provide an apparatus and method for accurately predicting a person's subjective choices.
Another object of the invention is to provide an apparatus and method for recommending items to a person based upon accurate estimates of a favorable reaction to the recommendation.
A further object of the invention is to provide an apparatus and method for accurately predicting subjective choices that is unaffected by individual rating patterns.
Another object of the invention is to provide an apparatus and method for accurately predicting subjective choices even when the number of ratings for a given item is relatively small, e.g. less than ten.
SUMMARY OF THE INVENTION
The foregoing objects are achieved in this invention in which books or other items are paired and all possible combinations of pairs or links are identified. The number of links is reduced based primarily on similarity of rating and secondarily on frequency of rating. The remaining links are used to make a prediction based upon the average rating (
) of the books, as rated by the user, linked to the book of interest plus the difference (&dgr;) between the average rating of the book of interest, as rated by all users, minus the average ratings of the linked books, as rated by all users. The averages of the linked books are weighted by similarity. In an alternative embodiment of the invention, the prediction is based upon the average of values applied to books linked to the book of interest, where the values are based upon ratings of the linked books by the user inquiring about the book of interest.
REFERENCES:
patent: 4613946 (1986-09-01), Forman
patent: 4870579 (1989-09-01), Hey
patent: 4996642 (1991-02-01), Hey
patent: 5583763 (1996-12-01), Atcheson et al.
patent: 5749081 (1998-05-01), Whiteis
patent: 5758026 (1998-05-01), Lobley et al.
patent: 5790426 (1998-08-01), Robinson
patent: 5884282 (1999-03-01), Robinson
patent: 5933516 (1999-08-01), Tu et al.
patent: 6016475 (2000-01-01), Miller et al.
patent: 6029195 (2000-02-01), Herz
patent: 6092049 (2000-07-01), Chislenko et al.
patent: 6112186 (2000-08-01), Bergh et al.
Choules Jack
Fennemore Craig P.C.
MediaChoice, Inc.
Pham Linh M
Wille Paul F.
LandOfFree
Method for predicting ratings does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for predicting ratings, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for predicting ratings will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2509397