System and method for mining playlist data for use in providing digital media or other content

Patent metadata

Field	Value
Publication number	US-11126629-B2
Application number	US-201715667209-A
Country	US
Kind code	B2
Filing date	Aug 2, 2017
Priority date	Aug 2, 2017
Publication date	Sep 21, 2021
Grant date	Sep 21, 2021

Abstract

In accordance with an embodiment, described herein is a system and method for mining playlist data, for use in providing advertisement-supported media content or targeted advertisements. A media server provides media content items as one or more media streams. Each of a plurality of playlists is associated with theme words provided by a title and/or description of the playlist. A playlist can also be associated with one or more theme words associated with the at least one media content item included in the playlist based on inclusion of the at least one media content item in the playlist. A dataset of media content item-theme word pairs provides an association of a particular media content item included in a particular playlist, with a theme word associated with that playlist. A score is assigned to each of the media content item-theme word pairs based on their appearances across several playlists. Advertisement-supported media content or targeted advertisements can be selected based on the dataset of media content item-theme word pairs, and, for example, fetched from an advertisement server, for insertion into a media stream.

First claim

What is claimed is:

1. A system for delivering media content with targeted advertisements, comprising:
a media server including an application interface that receives requests from client devices for media content, and a media content buffer that is populated with media content data provided thereon and associated with a plurality of media content items to be communicated to client devices, and a playlist database providing a playlist data describing a plurality of playlists, wherein each playlist includes at least one media content item of the plurality of media content items, and wherein each playlist is associated with one or more theme words;
wherein one or more of the client devices has a user interface adapted to display media options and to determine a user interaction indicative of selection of a particular media option, and the selection is used as a command to the media server to download, stream, or otherwise access a corresponding particular media content item or a media stream;
an advertisement server that provides advertisements for insertion into the media stream;
a keyword server that mines the playlist data and generates keywords based on and associated with tracks or other media content items delivered to the client devices and theme words derived from the mined playlist data describing playlists in which the tracks or other media content items appear;
wherein the system is configured to:
communicate, from the media server to the keyword server, the playlist data provided by the playlist database for each of a plurality of playlists, and a user listening history data provided by a user listening history database for each of a plurality of users that includes a history of media content items streamed to the users during a period of time;
generate by, and store, at the keyword server, a dataset of track-word pairs as a first lookup table, based on the playlist data received from the media server, by extracting theme words from titles and descriptions of playlists as indicated by the playlist data, and pairing the theme words for each playlist with media content items within the playlist;
generate by, and store, at the keyword server, a dataset of user-word pairs as a second lookup table, based on the user listening history data received from the media server and by reference to a set of media content items streamed to each particular user as indicated by their user listening history data, and the theme words associated with each media content item within the set as indicated by the dataset of track-word pairs;
wherein the lookup table of track-word pairs associates of a particular media content item included in a playlist with a particular theme word determined from the one or more theme words associated with the playlist, and wherein the lookup table of user-word pairs associates each particular user with a plurality of theme words determined from and associated with one or more media content items delivered to the user during a period of time;
assign a score to each track-word pair by determining a frequency of appearances of the track-word pair across the plurality of playlists associated with the theme words;
assign a score to each user-word pair by determining a frequency of appearances of each theme word associated with the media content items delivered to the particular user over a period of time;
while delivering a media stream to the particular user at a particular client device,
evaluate, by the keyword server referencing the lookup table of track-word pairs and the lookup table for the user-word pairs, the score assigned to each track-word pair, and the score assigned to each user-word pair, to determine one or more keywords from the one or more theme words, for use in selecting an advertisement from the plurality of advertisements to be fetched from the advertisement server, based on:
the one or more media content items delivered to the particular user in the media stream preceding the advertisement, and at least one of:
the score assigned to each track-word pair associated with the one or more media content items delivered to the particular user in the media stream preceding the advertisement, or
the score assigned to each user-word pair associated with the one or more media content items delivered to the particular user in the media stream preceding the advertisement; and

insert the selected advertisement into the media stream.

2. The system of claim 1 , wherein the one or more theme words associated with a playlist is derived from one or more of
a title of the playlist,
a description of the playlist, and
one or more theme words associated with at least one media content item included in the playlist.

3. The system of claim 1 ,
wherein the score assigned to each track-word pair is assigned based on a combination of two or more statistical calculations for the track-word pair, and
wherein the relative weighting of the combination is adjustable for each advertiser associated with the plurality of advertisements.

4. The system of claim 3 , wherein the score assigned to each track-word pair is assigned based on a combination of a pointwise mutual information (PMI) calculation and a chi-square calculation for the track-word pair.

5. The system of claim 1 , wherein the system selects the one or more keywords for use in selecting the advertisement based on highest score for track-word pairs associated with a last k media content items streamed to the user.

6. The system of claim 5 ,
wherein the advertisement server includes a real time advertisement server; and
wherein the real time advertisement server of the system selects the advertisement by
receiving an ad request,
querying the first lookup table maintaining the dataset for track-word pairs for the last k media content items streamed to the user,
calculating N most relevant theme words for the last k media content items, to be selected as the one or more keywords, and
selecting the advertisement based on the N most relevant theme words.

7. The system of claim 1 ,
wherein the system selects the one or more keywords for use in selecting the advertisement based on highest score for track-word pairs associated with the user.

8. The system of claim 7 ,
wherein the advertisement server includes a real time advertisement server and a batch advertisement server;
wherein an advertisement is selected for an advertiser by one of the real time advertisement server and the batch advertiser server;
wherein the real time advertisement server of the system selects the advertisement by
receiving an ad request,
querying the first lookup table maintaining the dataset for track-word pairs for the last k media content items streamed to the user,
calculating N most relevant theme words for the last k media content items, to be selected as the one or more keywords, and
selecting the advertisement based on the N most relevant theme words; and

wherein the batch advertisement server of the system selects the advertisement by
receiving an ad request,
querying the second lookup table maintaining the dataset for the user-word pairs for the user,
receiving N most relevant theme words for the user, as the one or more keywords, and
selecting the advertisement based on the N most relevant theme words.

9. A method for delivering media content with targeted advertisements, comprising:
providing a media server including an application interface that receives requests from client devices for media content, and a media content buffer that is populated with media content data provided thereon and associated with a plurality of media content items to be communicated to client devices, and a playlist database providing a playlist data descr

Assignees

• Spotify Ab

Inventors

• Mpossi Margreth
• Rabadi Michael

Classifications

• G06Q30/0255 Primary

  based on user history · CPC title

• G06F16/48

  Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually · CPC title

• G06Q30/02

  Marketing; Price estimation or determination; Fundraising · CPC title

• G06F16/951

  Indexing; Web crawling techniques · CPC title

• G06F16/40

  of multimedia data, e.g. slideshows comprising image and additional audio data (retrieval of still image data G06F16/50; retrieval of audio data G06F16/60; retrieval of video data G06F16/70) · CPC title