BRPI0708030A2 - systems and methods for prioritizing mobile media player files - Google Patents

Abstract

SYSTEMS AND METHODS FOR PRIORITIZING MOBILE MEDIA PLAYER FILES. Modalities of systems and methods for prioritizing mobile media player files are disclosed allowing for the automated addition and / or deletion of media items to a mobile media player. In some modalities, a statistical method can be provided to infer which media items in a mobile media player should be deleted based, for example, on the user's taste data. In some embodiments, new media streams can be loaded into a user's mobile media player by creating one or more playlists from a playlist builder. The execution list (s) can be created using the user's taste data. Classifications can also be created to determine an order for deleting media items currently on a mobile media player and / or for adding new media items to the device.

Description

"SYSTEMS AND METHODS FOR PRIORIZING MOBILE MEDIA PLAYER FILES"

BRIEF DESCRIPTION OF DRAWINGS

Understanding that the drawings represent only certain preferred embodiments of the invention and that, therefore, should not be construed as limiting their scope, preferred embodiments will be described and explained with specificity and further detail through the use of the accompanying drawings, in which:

Figure 1 is a flowchart of one embodiment of a method for auto-deleting media items from a portable media player.

Figure 2 is a flowchart of an implementation of a probability computation process.

Figure 3 is a flowchart of another embodiment of a method for the automated deletion of media items from a portable media player.

Figure 4 is a flowchart of one embodiment of a method for automatically deleting media items from a portable media player and for selecting and adding new media items to the player.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

In the following description, certain specific programming details, software modules, user selections, network transactions, database queries, database structures, etc. are provided for a complete understanding of the preferred specific embodiments of the invention. However, skilled in the art realize that modalities can be performed without one or more of the specific details, or with other methods, components, materials, etc.

In some cases, well-known structures, materials or operations are not shown or described in detail to avoid obscuring aspects of the preferred embodiments. Furthermore, the features, structures or characteristics described may be combined in any suitable manner in a variety of alternative embodiments.

In some embodiments, the methodologies and systems described herein may be performed using one or more digital processors, such as the types of microprocessors commonly found in PCs, laptops, PDAs, and all kinds of other desktop or portable electronics.

Modalities of systems and methods for prioritizing mobile media re-producer files are disclosed. Some embodiments of the invention may provide systems and methods for the automated deletion of files, such as media item, from a mobile media player. In some embodiments, a statistical method may be provided to infer which media item in a mobile media player should be deleted based on, for example, taste data. Deleting these files provides free space for new media items to be placed in the media player. One or more explicitly specified user parameters, playback scenarios and / or user preferences may be combined to rank items from the highest to lowest deletion priority to satisfy a required space constraint.

Some embodiments of the invention may also, or may alternatively, provide methods or systems for filling a mobile media player with media items. In some embodiments, the autofill operation may be automatically performed at a designated time and / or prior to a significant event, such as early in the day or prior to a trip. New media items loaded on the device may be selected to be responsive to the user's taste preferences, in some cases influenced by the anticipated event. New media items can also be selected, and current items on the deleted device, based in part on the mix of user taste data accumulated on the device since the last user history fill and community taste data in a host server of the system.

In one embodiment of a system according to the invention, one or more mobile media players are provided. Each mobile media player is configured to play media items in a media item playlist. A playlist modification component is provided that can be configured to receive a media item playlist from a given user's mobile media player to analyze user taste data associated with the media player. to modify the media item playlist using user taste data. A network streaming component can also be provided to transfer media items to the mobile media player and to delete media items from the mobile media player. according to the playlist of the modified media item. If desired, the playlist of the modified media item can also be transferred to the mobile media player.

The playlist modification component can also be configured to classify media items in the media item's playlist according to user taste data analysis. In such embodiments, the network broadcast component may also be configured to delete media items from the mobile player according to a similar classification. The playlist modification component can also comprise a playlist builder configured to construct one or more playlists of user taste data. The runtime builder can be configured to sort media items into at least one playlist of user-like data. When media items in a playlist are thus classified, the network broadcast component can be configured to transfer media items to the mobile media player according to its rating. Therefore, the system can classify media items for deletion and / or addition on a mobile media player.

In some embodiments, the playlist modification component may be further configured to compare user taste data with user taste data from a previous session to identify modified user taste data. Then, the playlist modification component can modify the media item playlist using the modified user taste data.

In one embodiment of a method according to the invention, a media item playlist is received from a mobile media player. User taste data associated with a mobile media player user can also be received and analyzed. Then, a recommended playlist can be generated from user taste data analysis. Then the recommended playlist can be compared to the media item playlist and modified using the comparison with the media item playlist. This comparison and modification may involve, for example, removing media items from the recommended playlist that are already in the media item playlist. Then, the media items in the recommended playlist, or a subset of the media items in the recommended playlist, can be classified. This classification may comprise the order in which media items are to be essentially transferred to the mobile media player. Of course, this does not necessarily mean that all items in the sorted list must be transferred. Space factor factors, special user demands, etc., may require less than all classified media items to be transferred. Media Items in the Media Item Playlist can also be sorted in the order in which they are to be removed from the mobile media player. However, again, this does not require all classified items to be removed. Then one or more media items may be deleted from the mobile media player in an order according to their rating, and one or more media items may be transferred to the mobile media player in an order according to their rating.

The playlist of the media item can also be modified to reflect media items removed from the mobile media player and media items transferred to the mobile media player. Then the playlist of the modified media item can also be transferred to the mobile media player.

As used herein, a "user recommender" is an integrated module in a user community whose primary function is to recommend users to other users in that community. There may be a set of items in the community for community users to interact with. There may also be an item recommender to re-order other items to users. Examples of recommending systems that may be used in conjunction with the embodiments herein are described in US Patent Application 11 / 346,818 entitled "Recommending System for Identifying a New Set of MediaItems Responsible for an Input Set of Media Items and Knowledge Base Metrics", and patent application US 11 / 048,950, entitled "Dynamic Identification of a New Set of Media Items Responsive to the Input Mediaset", both of which are incorporated herein by reference.

As used herein, the term "media data item" shall encompass any media item or representation of a media item. A "media item" shall cover any type of media file that may be represented in a digital media format, such as a song, movie, image, eBook, newspaper, TV / radio show segment, game, etc. Thus, the term "media data item" is intended to encompass, for example, playable media item files (for example, an MP3 file), as well as metadata identifying a reproducible media file (for example, metadata). that identify an MP3 file). Therefore, it should be apparent that in any fashion that provides a process, step or system using "media items", that process, step or system may use instead a representation of a media item (such as metadata), and vice versa.

Examples of processes and systems for the automated removal of media items from a mobile media player will now be discussed in more detail. Typically, the memory available for storing media items on a mobile media player is a limited resource that must be managed. When loading new media items, a user must often decide whether to overwrite, or otherwise remove, media items currently stored on their device to make room for new media items. The process for uploading new media items can be accelerated and simplified for a user if each media item on a device, or a subset of media items on a device, can be automatically classified according to the current user's interest in retaining it. lo. Given the classification, existing items can be automatically deleted from least desirable to most desirable until the amount of space required or a limit of desirability is reached. Then, the free space is available to upload new media items. Examples of processes and systems for automatically adding media items to a mobile media player will be discussed below.

Examples of methods for inferring free space based on regression statistics are described here. An example regression model for appending a selection probability to each item on the requesting device will be introduced. Then, an example method for estimating model parameters responsive to user preferences will be outlined. Finally, examples of a deterministic and randomized scheme for deleting items using statistical regression values will be described. Inferring "free space" on a mobile device by statistical regression: Consider that the total space available for storing media items on a dis- positive is St bytes and the device memory currently includes Hi1, m2, ..., mN items requiring space:

<formula> formula see original document page 6 </formula>

The task is to determine which items ITii1l mi2 ..... miL discard, the space used by the remaining i-tens

<formula> formula see original document page 6 </formula>

available for unpublished items.

Generalized Linear Regression Model:

One approach to doing this uses standard statistical regression with a generalized modelolinear. The decision to delete an item is represented as a binary random variable y, such that a sample instance "y = 1" means delete item m, and "Yi = 0" means retain item m ,. Assume that the value of y is predicted by a measurable covariate vector X = [X1, X2 ..... x «], which need not be binary.

The covariate X is prognostic and probabilistically. This is,

<formula> formula see original document page 6 </formula>

where X0 = 1 is considered for notation convenience and F (x) is an increasing mo-notonic function ranging from F (-oo) = 0 to F (oo) = 1. The covariate X can be any property measurable of items that reflect the user's taste that increases or decreases the likelihood of deleting an item.

Recalling that y is a binary random variable, we can show that E {y | X, /?}

<formula> formula see original document page 6 </formula>

This is mathematically equivalent to saying that the random variable y is described by the equation:

<formula> formula see original document page 6 </formula>

where e is a random variable with cumulative distribution F (y).

Multinomial Generalization of the Linear Model:

The basic linear model can be further extended into a linear model by nonlinear covariate functions simply by replacing the covariate vector X with an unpublished nonlinear covariate vector

<formula> formula see original document page 6 </formula>

in the general formulation of the model so that

<formula> formula see original document page 6 </formula>

This nonlinear model can be much more difficult to solve if one seeks to find an ideal multinomial function G (X) 1 instead of just the parameter vector β. The computational environment R supports this nonlinear extension in the generalized linear model solver, where G (X) is a multinomial vector function.

Estimating Model Parameters:

In the most common case, the vector β of the a priori model parameters is not known, but rather, an estimated β 'is derived given a set of examples (yi, xi), (y2, X2) .... . (,Ν, Χν). N »Μ. Typically this is done using maximum probability methods, which incorporate specific mathematical techniques to deal with certain details that arise in real applications. However, it may be unlikely in the present application that there is a set of examples (yi, xi), (y2, x2),, (Υν, Χν) that can be used to derive an estimate for the parameter vector β '= Ijff11l β \. .... ^ k]. Instead, one should either specify ad hoc values for β or use alternative methods to produce these values. A practical method is to quantify informal logical rules for retaining or deleting items.

Arithmeticization of Logic Expressions:

A simple approach is to consider that X variables are values for a set of X metrics that can be computed for each media item and that there is a ológica (Χ) function that informally describes the rules in terms of metrics for deciding whether an item. must be retained or deleted. Using the standard methods of Boolean algebra, you can reduce Γ (Χ) to a logical form of "sum of products":

Γ (Χ) = Z1 (X) V Z2 (X) V ... V Zk (X)

where each Zi (X) is a conjunction of some subset of the X metrics, for example:

Zi (X) = X2 A X5 A X7

We can translate the Boolean function Γ (Χ) into a multinomial function weighted by the X values of the X metrics.

BG (X) = P1Z1 + β2ζ2 + ... + yffkzk -1/2

where each Zi is the product of the metric values in the corresponding conjunctionZil for example:

Zi (X) = X2X5X7.

There are several ways to choose the βλ values for this translation of the Boole-ana Γ (Χ) function. One way, if we consider that each Xi is a limited nonnegative quantity, is to let each β \ be the reciprocal of the product from the upper limits of the metric value in the corresponding product, for example:

Bi = 1 / [sup (x2) sup (x5) Sup (x7)].

This monotonic model can be extended to include negated logic variables - "χ, by representing the metric value for a logic variable negated as-Xi = SUp (Xj) - Xi. Some Example Metrics:

As developed herein, the model may include, as a component, any metric that is predictive of whether an item should be retained or deleted. A predictive covariate class are popularity metrics, including:

1) The total number of reproductions p (t) at any point in time, perhaps normalized by the expected total number of reproductions P, (ie, p (t) / P).

2) The recent playback rate r (t) - [p (t) -p (t-At)] / At.

3) The recent skip rate s (t) = [q (t) -q (t-At)] / At, where q (t) is the number of times the listener prematurely finished playing the item.

Another class of predictive covariate is based on similarity between an item ie some unpublished candidate items j1t j2, ..., jm- A type of similarity metric is based on metadata. Suppose we have a set of I metadata items (for example, gender, age, label) for each item and we denote the metadata item values for item j as m1i, m2i, ..., mli. We can define metadata similarity between items i and j as

n (i, j) = msim (m1i, m1j) + msim (m2i, m2j) + ... + msim (mii, mij)

where msim (mii, m ,,) evaluates to "1" if mu and m ,, are considered to be similar, and "0" otherwise. Given item i and some unpublished candidate items j1? j2, ..., jm, we can use as another component of the model:

4) Metadata similarity q-mean n (i) = {[n (ij1) q + n (i, j2) q + ... + n (i, jI) q] / l} 1 / q

Note that for q = 1, q -> 0, q = -1, eq — ► q-mean reduces to arithmetic mean, geometric mean, harmonic mean and maximum (norm), respectively. In addition, you can also take the minimum of n (lji) as a similarity metric.

In addition to metadata similarity, in those cases where we have an independent disimilarity measure 0 <p (i, j) <1 between items i and j, we can define

5) The relational similarity of q-mean μ (ί) = {[M (i, ji) q + M (i, j2) q + ... + M (i, ji) q] / l} 1 / q

Finally, if we let U (i, j) denote the vector of similarities μ (i, Ι), 11 i, j for i-tem i, we can define the cosine similarity between U (i, j) and U ( j, i) how

p (i, j) = (U (i, j), U (j, i)) / IIU (U) II || U (j, i) ||

In some cases, you may want to include only the similarities of component μ (i, Ι) for the unpublished items ji, j2 ..... jm in the similarity vectors U (i, j) and U (ji> - Using these similarities of component p (i, j) we can define

6) The cosine similarity of q-mean p (i) = ([P (Ij1) "+ p (i, j2) q + ... + p (iji) q] / l} 1 / q

Other metrics and techniques for combining measurable metrics on an item in a probability value, including those involved in the arithmeticization of a logical functionΓ (Χ) that informally describes whether an item should be retained or deleted, are also in the spirit of the deduction method. the probability that an item should be retained or deleted, as described herein.

Item Deletion Methods:

Since we have a coefficient vector β, we can determine which items to let in one of two ways:

Deterministic Deletion:

In the deterministic approach, the actual decision D deletes the item Hi1 is determined by E {yi | Xi, /?} = F (^ Xi) predicted as

/ 1 user specifies which item to delete

|

di = <0 user specifies which item to hold

\ 1 if E {yi | Xi /?}> B, otherwise

where Xi is the vector of the measured factors for item m ,, and the limit 0 <B <1 is a desirable limit to retain items, or is selected such that the resulting free space S1 is large enough to satisfy all other space requirements that may be imposed externally. Random selection

The random approach uses the binary random variable y directly. That is, the decision d, deleting the item mi is the random variable.

/ 1 user specifies which item to delete

|

d, = <0 user specifies which item to hold

|

\ yIt otherwise

where Prfyi = 1 | Xi) = F (^ Xi). Given FOffXi), standard computation techniques can be used to generate sample instances of a binary random variable y, with required binary distribution Prfyi = 1} and Pr {y, = 0}.

Because the deletion is a random process, the amount of free space resulting from Sf is also a random variable in this approach. Satisfy a desired goal for the resulting amount of free space Sf> S | may be more complex than the deterministic approach. This can be handled by generating a sample instance vector.

Y (n) = [yi, y2 ..... yN],

making the indicated deletions and evaluating the resulting free space Sf. Then the process can be repeated as necessary until the requirement Sf> S | be satisfied. Alternatively, sample instance vectors Yi1i, Y * 2) can be sequentially generated, making no deletions until the deletions specified by the particular instance Y (m) result in sufficient free space to satisfy the requirement Sf> S | Illustrative methods and systems are described earlier for intelligently deleting media items on a mobile player device based on the user's preferences for freeing up device storage space for new media items. Some modalities use predictive covariate Xi = [X1l x2, ..., Χκ], which are metrics for a user's taste preferences, to arrive at a probabilistic decision y about each media item m on the device must be deleted. The covariate Xi can include metrics of local taste data uploaded from the mobile player to the system host, as well as community taste measures gathered and maintained by the system host of a user community. Similarly, the method may be incorporated into the mobile device and triggered by metrics supplied to it by the system host, or may be resident in the system host, and the final deletion decision transmitted to the mobile device.

An example of a method for automatically deleting media items from a mobile media producer is shown in the flowchart of figure 1. As shown in the figure, media item playback data is recorded on a media player. Then, a user makes a request to transfer new media items and / or to create free space on the device in step 504. Then, the media items in the device are sorted based on the assigned probability values, such as indicated at 506. Then the process of deleting items starts at 508, starting from the highest (or lowest, depending on how the rating is structured) ranked media item from the list. Once a media item has been deleted, the system can Check to see if a limit value has been reached at 512. If the limit value has been reached, the sequence ends. If not, a check is made if the required space on the device is available as indicated in 510. If the requested space is not yet free, the next item in the list is deleted. If so, the sequence ends.

An example of a probability computation process is shown in the flowchart of figure 2. As shown in the figure, a probability model is selected at 514. Then local and / or community metrics are applied to determine a vector. covariate coefficient, as indicated in 516. Then, a deletion probability is assigned to each media item in the portable media player at 518. Then, the deletion values are applied in the deletion method as previously described. - written in 520.

Another example of a method for automatically deleting media items from a mobile media player is shown in the flowchart of figure 3. As shown in the figure, media item playback data is recorded to a portable media player in step 602. A user then makes a request to transfer new media items and / or create free space on the device in step 604. Then, probability values for deleting each of the media items, or a subset of the media items, Portable media producers are accessed at 606. Then, the process of randomly deleting items based on computed probabilities starts at 608. Once a media item has been deleted, the system can check to see if a threshold value has been reached at 612. If the limit value has been reached, the sequence ends. If not, a check is made if the required space on the device is available, as indicated in 610. If the requested space is not yet free, the next item in the list is deleted. If so, the sequence ends.

Now methods for auto-filling a mobile media player will also be described in more detail. In one embodiment, each of a plurality of portable media players has access to a central server or the like. Each user can initially obtain and install software on a local computer. Each user can also provide taste data to the server.

After the above-described initialization phase, a user can establish communication between the host server and its media player, typically by connecting the media player to a networked computer to access the host server. The user can then issue a "start" command to start the auto-fill process. Alternatively, the system can be configured to automatically start the process by being connected to a media player. In some versions, this step may also include retrieving local user taste data such as playback counts, playback sequences, media item ratings, etc. of the mobile device. Still other examples of user taste data are presented elsewhere in this disclosure.

In some versions, a list of media items currently on the device may also be transferred to the host server. Then, the host server may supply the user's taste data to a standalone playlist builder system, which may use all retrieved user taste data, historical user taste data, and / or the list of user items. media from the mobile device to provide the host server with playlists of media items responsive to the user's preferences. These playlists can comprise media items in all user-available media libraries known to the playlist builder system. Typically, these libraries will include media items licensed to the user and stored on the server for user access. .

Once the host server has been supplied with playlists, the host server can retrieve a list of media items currently on the mobile device (if the list has not been previously retrieved). Then, the list of new media items to be loaded on the device is reconciled with the list of media items currently on the device to determine how much space should be freed up on the device to accommodate the new media items and which media items. media must be deleted. Then the deleted media items for deletion and the new media items are loaded onto the device.

Although the preceding steps can be performed repetitively to infinity without additional explicit user intervention, from time to time the user may wish to adjust user preferences to change autofill behavior. For example, a user may wish to provide the user with special instructions, such as instructions for marking certain media items stored on the requesting Device as "non-select" or "always delete". Alternatively, a user may wish to provide additional user taste data by associating user likes with a specific activity, such as data in the form of a request that a selection of media items or a genre of media items be to "run" , "dinner" or "work". As yet another example, a user may wish to provide temporal user taste data by associating a selection of media items or a media item genre with a future date or event.

As indicated earlier, some system modalities may be configured to assume that the host server has access to a runtime builder system that automatically generates playlists based on user preferences that are answered in response to a query by the host server. Examples of such systems known to those skilled in the art include US Patent 6,953,532, entitled "Auto Playlist Generator" and 6,526,411, entitled "System and Method for Creating Dynamic Playlists". Additional examples can be found in US2002 / 0082901, entitled "Relationship Discovery Engine", 2003/0221541, "AutoPlaylist Generation with Multiple Seed Songs" and 2005/0235811, entitled "Systems for and Methods of Selection, Characterization and Automated Sequencing" of Media Content ". Each of the foregoing published patents and patent applications are incorporated herein by reference. In preferred embodiments, the playlist builder will accept user's taste data in multiple forms, such as one or more of the following:

1. Tags - Tags are pieces of information that are separate from, but related to, an object. In the practice of collaborative categorization using freely chosen keywords, tags are descriptors that individuals assign to objects.

2. Media Items or Media Item Lists - Playlists, Movies, etc. which are the same type as the items being transferred to the portable device.

3. Special Item Lists - For example, items that you explicitly ban, items that a smart system that uses user feedback data implicitly ban, items that you explicitly request to be on your device.

4. Metadata - Text-based data that is regularly associated with an item to classify it (for example, gender, year, date of purchase).

5. Ratings - User-defined value that judges the quality of the item.

6. Events - Time-based resources that can be inferred from a calendar (for example, a Yahoo® calendar).

7. Serendipity - User-defined value that indicates the user's preference for non-popular content in their streaming stream.

8. - Playback Models - A model of a user's item playback behavior, such as the number of times a user plays a media item before getting tired of it, the user's expected playback count profile for an item over time. , and / or percentage of a media item playlist that a user produces before switching to another.

9. - Continuous Playback Stream - Timestamp recording of media playback interaction, including statistics such as completion, skipped, restarted, and / or deleted items.

10. Play Count - Number of times a media item has been played back over a given period of time.

11. Time Data - Preferences implicitly or explicitly associated with an instance of time or event, such as time of day or day of week.

12. Signed Influences - User-defined external taste inputs that help define the desired experience (eg, friend's likes, expert-defined media items such as a DJ, etc.).

13. Environmental Noise - User ambient noise level perceived by the identifier.

14. Discovery Value - User-defined value to indicate media items.

never reproduced by the user that can be introduced into the media consumption experience.

15. Rediscovery Value - User-defined value to indicate the media items that the user has not recently played, which can be entered in the experiment.

30 media consumption.

Given one or more playlists generated by a playlist builder system in response to a query, then the host server can determine the set of media items included in those playlists and perform a self-fill process, an example of which is shown in figure 4. Initiation 702 of the

The autocomplete process may be in response to a single user action, such as connecting the mobile device to the host server, or receiving an explicit "start" indication by the user. Then a playlist builder can receive user taste data 720 and use the data to generate one or more playlists, as indicated in 704.

Once the playlist (s) have been provided, the host server may enumerate 706 the set of media items in those playlists and then reconcile that set. media items with the currently-in-place media item set. The host can determine which of the new media items are already in the device and removes them from the smooth of items to be loaded. In some embodiments, these media items may also be removed from the deletion consideration. Then the host server can use a method to effectively classify media items on the device considering the likelihood that they should be deleted based on how compatible they are with the user's taste data supplied to the playlist builder system. . This rating can then be used to select the media items to load on the device. The host server can also classify media items in a similar way, considering the likelihood that they should be deleted from the re-producer.

Using rating information for new media items, the host server can, as shown in 710, select the item with the highest probability that should be loaded from the list of new media items. Then items can be deleted 712 from the media producer until sufficient space is available for the most compatible of the new media items using the method described below, after which media items are loaded 714 into the media player. This process of selecting 710, deleting 712, and loading 714 is repeated until, for example, the time available for the autofill operation has elapsed, as indicated in 716, all device items have been deleted, as indicated in 718, or the entire set of unpublished media items has been loaded on the device, as indicated in 722. Alternatively, in circumstances where there is no time limit on the upload process, the 712 deletion process may delete enough items from the device to free enough space to load tan -the items unreleased for device capacity will accommodate, or a preselected number of items. A user can also enforce that a particular subset of memory be used for unpublished items and free that amount of space. Then the loading process 714 can load as many new media items as available device space allows in a single operation.

Finally, for media players who also accept runtime transfers, the playlist (s) can be edited to remove all unpublished items that may not load, and then the ( run list (s) loaded into the device as indicated at 724 and 726 respectively.

In some embodiments, media items may be randomly chosen and engineered to free sufficient storage space on the media player to accommodate new media items. In other embodiments, currently reproducing media items can be ranked in terms of how compatible they are with the criteria used to generate playlists of new media items to be loaded onto the device. Then these media items may be deleted preferably based on that rating. The same process can also be applied to playlist items to provide a reverse sort of new media items to get the most desirable items loaded first. In this way, the subset of the most desirable unpublished items will be loaded in the event that the entire set of new media items cannot be loaded due to, for example, time or space limitations.

As explained, a statistical method can be applied to infer which media items in the player should be deleted based on user preference preferences metrics. The method can model the conditional probability of the binary decision y that an item should be deleted (y = 1) in terms of a computed value X for an item η-vector X of the item:

Pr {y = 1 IX = IX11X2 ..... Xk]} = F (0G (X))

where G (X) is an m-vector of the nonlinear metric functions and β is a vector of weighting values for linearly combining the components of G (X) into a calar value. The functions of the G (X) component are of a particular class of function, such that the decision variable y has a particular type of generalized linear model.

y = F (£ G (X)) + e

where F (X) is a non-decreasing "binding" monotonic function that maps (-oo, oo) to the range (0,1), and is a zero-mean random variable that is considered to have a relevant distribution (although the process of self-deletion operate unconscious distribution of e).

Given F (X), β, and G (X) data specified, the host server sorts the media player items that are candidates for deletion and unpublished items to be loaded by the first FOffG (X) computation for each item in both sets. A common choice of F (x) is the logistic function.

F (x) = 1 / (1 + e-x)

Since F (x) is a non-decreasing monotonic function, the items can be classified by the computed values F (/? G (X)), with the values /? G (X) used as first round tiebreakers. when the F (/? G (X)) values for two items are identical and the /? G (X) values for those items differ. When both FOffG (X)) and / ffG (X) values match for both items, any preferred method can be used to rank one after another, including a random choice. Thus, the classification of media items in a set by the values of F (BG (X)) is seen identical to the classification of those items by the values of BG (X), the latter being a classification explicitly linked to computed metrics.

Another method for computing the metrics X = [X1, ..., Xk] assumes that items should be classified based on a set of item attributes a1, a2 ..... ak) as described above in the discussion of the construction of the playlist, and a set of preference values P1lP2 ..... pk ranging from 0 (strong preference) to 1 (weak preference). The degree to which an item satisfies the ak attribute can be computed as xK = h (ak) [100-90p (] + (1-h (ak)) u (pk)

where h (ak) = 1 if the item has ak attribute and h (ak) = 0 otherwise, and u (pk) = 0 if

Pk = Oe u (pk) = 5 otherwise. Then media items on the device can be classified for possible deletion by

PG (X) = 50k - (X1 + ... + xK)

Similarly, unpublished items can be sorted upside down according to the same values.

The classification of media items on the device during the reconciliation process 708 establishes the relative likelihood that the items will be deleted. In deletion process 712, items are actually deleted in a deterministic or probabilistic manner, for example by the methods described herein. The deterministic and probabilistic methods used for deletion can also be applied to process 712 to select items in the sequence of unpublished items to increase the diversity of user experiences in repeated autofilling operations.

The foregoing description fully discloses the invention, including preferred embodiments thereof. Without further elaboration, it is believed that those skilled in the art may use the foregoing description to utilize the invention to its fullest extent. Therefore, the examples and embodiments disclosed herein are to be construed as illustrative only, and not as limiting the scope of the present invention in any way.

It will be obvious to those skilled in the art that many changes can be made to the details of the above-described embodiments without departing from the fundamental principles of the invention. Therefore, it is understood that the invention should not be limited to the specific embodiments disclosed, and that modifications and other embodiments should be included within the scope of the appended claims.

Therefore, the scope of the present invention should be determined solely by the following claims.

Claims (21)

1. System for the automated modification of a mobile media player playlist, characterized by the fact that it comprises: a mobile media player configured to play media items in a media item playlist; execution configured to receive media item execution list, analyze user taste data associated with the mobile media producer, and modify the media item playlist using the user's taste data; is a network streaming component configured to transfer media items to the mobile media player and to delete media items from the mobile media player according to the modified media item's playlist. System according to claim 1, characterized in that the playlist modification component is configured to rank the media items in the media item playlist according to the analysis of user taste data. System according to claim 2, characterized in that the network transmission component is configured to delete media items from the mobile media player according to their ratings. A system according to claim 1, characterized in that the playlist modification component comprises an execution list builder configured to construct one or more playlists from the user's taste data. System according to Claim 4, characterized in that the playlist constructor is configured to sort media items into at least one playlist from the user's taste data. System according to claim 5, characterized in that the network transmission component is configured to transfer media items to the mobile media player according to their ratings. System according to claim 1, characterized in that the playlist modification component is configured to compare user taste data with user taste data from a previous session to identify modified user taste data, and wherein the playlist modifying component is configured to modify the media item playlist using the modified user taste data. System according to claim 1, characterized in that the network transmission component is configured to transfer the modified media item playlist to the mobile media player. System according to claim 1, characterized in that the user's taste data comprises at least one of the production count data, labels, special user instructions, metadata, ratings and reproduction statistics. System according to Claim 1, characterized in that the user's taste data comprises at least one of an indicative metric, the total number of plays for one or more media items in the media item playlist, a metric. indicative of a recent playback rate for one or more media items in the media item playlist, and a metric indicative of a skip rate for one or more media items in the media item playlist. System according to claim 1, characterized in that user taste data comprises temporal user taste data relating to a future date or event. System according to claim 1, characterized in that user taste data comprises data that associates user tastes with a specific activity. 13. Method for Automatically Modifying a Mobile Media Player Playlist, FEATURED by the fact that the method comprises: receiving a media item playlist from a mobile media player, receiving user taste data associated with a mobile media player user; generate a recommended playlist from user taste data; compare the recommended playlist with the media item playlist; modify the recommended playlist using the playlist comparison media item; remove media items from the mobile media player, where media items are removed from the mobile media player according to the user's taste data. A method according to claim 13, characterized in that it further comprises transferring media items to the mobile media player, wherein media items are transferred to the mobile media player according to the data of taste of the mobile media player. user. Method according to claim 14, characterized in that it further comprises classifying one or more media items in the re-ordered playlist in an order in which the media items are to be transferred to the media player. mobile. Method according to claim 14, characterized in that it further comprises: modifying the media item playlist according to the mobile media player's moved media items; modifying the item playlist media according to the media items transferred to the mobile media player; and transmit the modified media item playlist to the mobile media player. A method according to claim 13, characterized in that the user's taste data comprises at least one of an indicative metric of the total number of plays for one or more media items in the media item playlist of a metric indicative of a recent playback rate for one or more media items in the media item playlist, and a metric indicative of a recent ignore rate for one or more media items in the media item playlist. A method according to claim 13, characterized in that it further comprises classifying one or more media items in the media-playlist playlist in an order in which the media items are to be removed from the mobile media player. A method according to claim 13, characterized in that it further comprises: receiving one or more special instructions from the user; and apply one or more special instructions to the user on the media i-tem playlist. A method according to claim 19, wherein one or more special user instructions comprise a media item removal list to remove and a media item retention list to retain, and in which the step Applying one or more special instructions to the user in the Media Item Playlist comprises removing media items from the Media Player Remove List and preventing the removal of media items from the Media Player Retention List. 21. Method for Automatically Modifying a Mobile Media Player Playlist, FEATURED by the fact that the method comprises: receiving a media item playlist from a mobile media player, receiving user taste data associated with a mobile media player user; generate a recommended playlist from the user's taste data; compare the recommended playlist with the media item playlist; modify the recommended playlist using the comparison with the media item playlist; sort one or more media items in the recommended playlist into an order in which media items should be transferred to the mobile media player; sort one or more media items in the item playlist media in an order in which media items should be removed from the mobile media player; remove media items from the media player m level, where sãoremovidos media items from the mobile media player according to their rankings; transfer media items to the mobile media player, where mídiasão items transferred to the mobile media player according to their rankings.