Consumer item matching method and system

• US 7,003,515 B1
• Filed: 05/16/2002
• Issued: 02/21/2006
• Est. Priority Date: 05/16/2001
• Status: Active Grant

• Current Assignee: Pandora Media LLC (Liberty Media Corporation)
• Original Assignee: Pandora Media Incorporated (Liberty Media Corporation)
• Inventors: Westergren, Timothy B., Stearns, Jeffrey P., Kraft, Jonathan M., Glaser, William T.
• Primary Examiner(s): Choules, Jack M.
• Application Number:
  US10/150,876
• Time in Patent Office: 1,377 Days
• Field of Search: 707/5, 707/6, 707/104.1
• US Class Current: 707/723
• CPC Class Codes: G06F 16/634 Query by example, e.g. query by humming
• CPC Class Codes: G06F 16/639 using playlists
• CPC Class Codes: G06F 16/683 using metadata automatically derived from the content
• CPC Class Codes: Y10S 707/914 Video
• CPC Class Codes: Y10S 707/916 Audio
• CPC Class Codes: Y10S 707/957 Multidimensional
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99936 Pattern matching access
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A method of determining at least one match item corresponding to a source item. A database of multiple items such as songs is created. Each song is also represented by an n-dimensional database vector in which each element corresponding to one of n musical characteristics of the song. An n-dimensional source song vector that corresponds to the musical characteristics of a source song is determined. A Distance between the source song vector and each of database song vector is calculated, each distance being a function of the differences between the n musical characteristics of the source song vector and one of source database song vector. The calculation of the distances may include the application of a weighted factor to the musical characteristics of resulting vector. A match song is selected based on the magnitude of the distance between the source song and each database songs after applying any weighted factors.

Claims

1. A computer implemented method of determining at least one match song that corresponds to a source song, comprising the steps:

creating a database comprising multiple songs, each song in the database represented by an n-dimensional database vector corresponding to n musical characteristics of the song;

determining a n-dimensional source song vector that corresponds to n musical characteristics of the source song;

calculating a first distance between the source song vector and a first database song vector, the distance being a function of the differences between the n musical characteristics of the source song vector and the first database song vector, wherein the step of calculating the first distance includes application of a weighting factor;

calculating a second distance between the source song vector and a second database song vector, the distance being a function of the differences between the n musical characteristics of the source song vector and the second database song vector;

selecting the at least one match song based on the magnitude of the first distance and the second distance.

2. A method as in claim 1 wherein the steps of calculating the second distances further includes application of a weighting factor to the difference between certain of the n musical characteristics of the second database song vectors and the source song vector.

3. A method as in claim 1 wherein a scaling factor is applied to a musical characteristic n of the source song vector before calculating a first or a second distance.

4. A method as in claim 3 wherein a scaling factor is applied to a musical characteristic n of the first database song vector before calculating a first or a second distance.

5. A method as in claim 1 further comprising, in addition to calculating the first distance and the second distance:

calculating a third distance between the source song vector and a third database song vector, the distance being a function of the difference between the n musical characteristics of the source song vector and the third database vector;

selecting the at least one match song based on the magnitude of the first distance, the second distance and the third distance.

6. A method as in claim 1 wherein a plurality of songs in the song database are ranked based on the calculated distances.

7. A method as in claim 1 wherein n is at least 100.

8. A method as in claim 2 wherein the certain n musical characteristics are predefined at the time of creation of the database.

9. A computer implemented method as in claim 1 wherein the steps of calculating the second distances further includes application of a weighting factor to the difference between certain of the n characteristics of the second database item vectors and the source item vector.

10. A computer implemented method of choosing a recommended song from a database that corresponds to a source song designated by a user comprising the steps of:

identifying a group of characteristics that correspond to the source song;

selecting one or more characteristics from the group;

generating weighting values based on the characteristics;

assigning the weighting values to characteristics of the source song;

comparing the weighted characteristics of the source song and characteristics of the songs in the predetermined database; and

selecting a recommended song based on the comparison.

11. A method as in claim 10 wherein the step of identifying a group of characteristics comprises evaluating at least 100 musical qualities of the song.

12. A method as in claim 10 wherein the step of selecting the characteristics comprises receiving user preference input.

13. A computer implemented method for selecting songs for a user comprising the steps of:

user-selecting a group of songs from a database, each song including multiple characteristics and a numerical value corresponding to each characteristic;

calculating a vector pair corresponding to the group, comprising the steps of;

calculating a first vector from averages of the numerical values of the song characteristics of the songs in the group;

calculating a second vector based on a calculation of the deviation of the numerical values of the song characteristics of the songs in the group;

selecting additional songs from the database based on a relationship of the vector pair and numerical characteristics of other songs in the database.

14. A method as in claim 13 wherein the step of calculating the vector pair further includes application of a weighting factor to some of the numerical values of the song characteristics.

15. A method as in claim 14 wherein the weighting factor is selected by the user.

16. A method as in claim 15 wherein the database contains songs analyzed and assigned numerical values corresponding to characteristics by a listener.

17. A computer implemented method for generating recommended items for a user comprising the steps of:

collecting selection data regarding user preference for particular items;

grouping the items into preference clusters;

calculating a vector pair for each of the preference clusters comprising the steps of;

calculating a first vector relating to averages of numerical values corresponding to characteristics of each of the items in the preference cluster;

calculating a second vector corresponding to the deviation of the numerical values of the characteristics;

selecting items from a database based on a relationship of the vector pair and characteristics of other items in the database.

18. A method as in claim 17 wherein the item is selected from the group consisting of music, videos, wine, films, books and video games.

19. An apparatus comprising:

a processor; and

a memory for storing computer readable instructions that, when executed by the processor, cause the computer to perform the steps of;

receiving user input corresponding to an item preferred by a user;

determining a n-dimensional source item vector that corresponds to n characteristics of the source item;

accessing a database comprising information about multiple items, each item in the database represented by an n-dimensional database vector corresponding to n characteristics of the item;

calculating a first distance between the source item vector and a first database item vector, the distance being a function of the differences between the n characteristics of the source item vector and the first database item vector, wherein the step of

calculating the first distance includes application of a weighting factor;

calculating a second distance between the source item vector and a second database item vector, the distance being a function of the differences between the n characteristics of the source item vector and the second database item vector;

selecting a recommended item based on the magnitude of the first distance and the second distance;

outputting data to the user regarding the recommended item.

20. An apparatus as in claim 19 wherein the item is from the group consisting of music, videos, video games and wine.

21. A computer implemented method of dynamically determining focus trait characteristics of a source song comprising the steps of:

generating an underlying database of ratings of musical characteristics of the source song;

presenting the musical characteristics to a user;

using the user'"'"'s selection of the musical characteristics of the source song to choose a weighting vector for use in subsequent matching.