Patent Application
20150181271
The present invention relates to data processing, and more particularly relates to a method and a device for providing targeted content.
Streaming media are multimedia that are constantly received by and presented to an end-user while being delivered by a media provider. Streaming media are usually transmitted over HTTP protocol. There are several HPPT-based media streaming communication protocols. Among those protocols, HTTP Live Streaming (also known as HLS) is implemented by Apple Inc. as part of their QuickTime and iOS software. It works by breaking the overall stream into a sequence of small HTTP-based file downloads, each download loading one short chunk of an overall transport stream. As the stream is played, the client may select from a number of different alternate streams containing the same material or content encoded at a variety of data rates, allowing the streaming session to adapt to the available data rate. At the start of the streaming session, it downloads an extended M3U (m3u8) playlist containing the metadata for the various sub-streams which are available. Herein, a playlist is a list of entries or identifiers specifying locations of one or more multimedia files. An entry or an identifier uniquely specifies a local pathname (either absolute pathname or relative pathname), e.g. “C:\My Music\newsong.mp3” and can specify a web address or URL (uniform resource locator), e.g. “http://www.mysong.com/mysong.avi”.
Based on media streaming technologies, Video on demand (VOD) or audio and video on demand (AVOD) systems can be built. The systems allow users to select and watch/listen to video or audio content on demand on the Internet. In order to generate revenue, a service provider adds or embeds advertisement into requested content.
It is desired to have a method for providing targeting advertisement content.
According to an aspect of the present invention, it is provided a method for providing content for a video. The method includes steps of receiving, from a client device, a message comprising a first content identifier uniquely identifying a first piece of content for the video and a first relevancy score associated with the first piece of content, wherein the first piece of content and the first relevancy score are determined based on a user identifier that identifies a user and a video identifier that identifies the video; determining a second piece of content and a second relevancy score associated with the second piece of content based on the user identifier, the video identifier and playback context parameter including at least one of geographical location, device type and time; and if it is determined that the second relevancy score divided by computation cost for inlaying the second piece of content into the video is larger than the first relevancy score divided by computation cost for inlaying the first piece of content into the video, using the second piece of content for the video.
According to a variant, it further comprises a step of determining relevancy scores of the first piece of content and the second piece of content to the video as the first score and the second score.
According to a variant, it further comprises steps of determining relevancy scores of the first piece of content and the second piece of content to the video; and weighting the relevancy scores of the first piece of content and the second piece of content with computation costs for the first piece of content and the second piece of content so as to get the first score and the second score.
According to a variant, the message is in the form of URL, it further comprises a step of modifying the URL by replacing the first content identifier with a second content identifier that uniquely identifies the second piece of content so as to redirect the client device to consume the second piece of content.
According to another aspect of the present invention, it is provided a device for providing content for a video. The device comprises a network interface for receiving, from a client device, a message comprising a first content identifier identifying a first piece of content for the video and a first relevancy score associated with the first piece of content, wherein the first piece of content and the first relevancy score are determined based on a user identifier that identifies a user and a video identifier that identifies the video; and a CPU for determining a second piece of content and a second relevancy score associated with the second piece of content based on the user identifier, the video identifier and playback context parameter including at least one of geographical location, device type and time; and it is determined that the second relevancy score divided by computation cost for inlaying the second piece of content into the video is larger than the first relevancy score divided by computation cost for inlaying the first piece of content into the video, using the second piece of advertisement for the video.
According to a variant, the CPU is further used for determining relevancy scores of the first piece of content and the second piece of content to the video as the first score and the second score.
According to a variant, the CPU is further used for determining relevancy scores of the first piece of content and the second piece of content to the video; and weighting the relevancy scores of the first piece of content and the second piece of content with computation costs for the first piece of content and the second piece of content so as to get the first score and the second score.
According to a variant, the message is in the form of URL, the CPU is further used for modifying the URL by replacing the first content identifier with a second content identifier that uniquely identifies the second piece of content so as to redirect the client device to consume the second piece of content
It is to be understood that more aspects and advantages of the invention will be found in the following detailed description of the present invention.
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, will be used to illustrate an embodiment of the invention, as explained by the description. The invention is not limited to the embodiment.
In the drawings:
The embodiment of the present invention will now be described in detail in conjunction with the drawings. In the following description, some detailed descriptions of known functions and configurations may be omitted for clarity and conciseness.
According to an embodiment of the present invention, the principle of the present invention is used to improve a following system, in which insertion of at least one piece of targeted advertisement is decided once and fixed for a requested video content. In some systems, e.g. HTTP based media streaming system, a client device may request for, get and store a playlist from a server and in a later time the content is played by running the playlist on the client device requesting for the playlist or a different device that receives the playlist from the client device. But in the traditional system the playback is not bound to a given playback time and does not make use of information about the client environment. The present embodiment provides a situation where information about the client environment is used to dynamically adapt the advertisement content to be played (one or more piece of advertisement content among all pieces of advertisement content associated with the requested video content). According to the embodiment of the present invention, in response to a request for a video content, e.g. a movie from a user, a server generates a playlist or a file including identifiers specifying both the movie and one or more pieces of recommended advertisement; the user can store the playlist and use the playlist to play back the movie in a later time on the client device requesting for the playlist or a different client device. Herein, the advertisement as well as its score (e.g. relevance score) are decided twice separately when the user sends the request for the movie and when the user uses the playlist to play back the movie, wherein the user environment or playback context (e.g. geographical location) is taken into account for deciding the advertisement when the user uses the playlist to play back the movie. Besides, if two pieces of corresponding recommended advertisement decided at above two time points are different and the score of the advertisement decided at the time of requesting to play back is much larger than the score decided at the time of requesting for the movie, the advertisement decided at the time of requesting to play back is used as the advertisement during the play back.
In the above system, a multi-level ad-selection method requests at least twice for a same set of parameters (user_id, content_id), and determines whether the latter determined advertisement is more relevant and suitable to play than the former one. Besides, the method takes into account of additional playback context parameters (such as the client's geographical localization) that may have changed compared to that at the time of the first advertisement selection. According to a variant, because advertisement inlay process, advertisement overlay insertion process (e.g. used for logo insertion), or other types of advertisement approach (e.g. switching to some pre-computed video part) incurs cost, the method further weights latter ad-decision with the cost it incurs (e.g. to compute for inlay advertisement into the movie in some advertising methods), and use the latter one to play if it is beneficial compared to the first ad-selection.
Further, the computation cost may be different at the playlist generation time and at the playback time for reasons below. It is one of motivations to enable this two-level recommendation mechanism because the computation cost is hard to predict at the playlist generation time: 1) batch processing: any server can process different ads in a batch queue for example, if the server is in an idle state; 2) low cost compute instance: one way to spent less money on buying instance is to bid a price on unused cloud provider capacity. However, we cannot be sure that this instance will be available. Therefore, if such instance is being used instead of one of the normal price, the computation cost decreases. 3) cache operation: a score may take into account that the computed ad is available at the playlist generation time, thus computation cost is low. At the playback time, a cache algorithm decided to delete the computed ad, thus the score changes since the ad must be recomputed (computation cost increases).
An example that is put into a framework where advertisement is inlaid into the movie is shown below. In other words, the advertisement is placed on the placeholder of at least one image of the movie.
Turning back to
The second level of advertisement determination (steps 5 to 8) occurs at playing time. In order to play the requested movie, the client device sends URLs in the playlist one after another to the URL playback server. The URL playback server obtains the user ID, the movie ID, playback context and the previously calculated advertisement and its score. For example, the movie ID and the playback context are obtained in a way that they are sent together with the URL (or before the transmission of the URL) from the client device to the URL playback server. And the user ID, previously calculated advertisement and its score are obtained from the URL. Similarly as before, the playback server calls the recommender engine and gets back a possibly different advertisement and a new score. If advertisements generated by the different recommendations steps 3 and 7 are different, and the score of the latter advertisement is significantly higher than the former one (e.g. bigger than a predetermined threshold value or threshold percentage), the playback server modifies original URL to replace original advertisement and original score with the new advertisement and new score or generates a new URL from the newly returned 2-tuple, and redirecting the request to the client device so that the client device is redirected to consume the new advertisement. If the difference in score is not significant, the original URL is kept and not substituted because computing a new chunk (i.e. inlaying the new advertisement into the movie) urgently is costly and it does not bring significant improvements in the advertisement targeting. According to a variant, other than sending information about the playback context in a separate message, the URL generated at the playlist requesting time is modified or complemented with playback context. The complemented URL sent by the client must be well understood and decoded by the URL playback server. It can be based on a predefined URL semantic generated by the client or depending on placeholders generated at the playlist requesting time but filled by the client device at the playback time. An example is shown below:
(http:..movie/ads/{score}/{playback context}/{playlist date}/{last ad seen})
Besides, in order to implement the above method, the client device is able to capture playback context and to replace fields of the URL before sending the URLs to the URL playback server. According to another embodiment, the client connects to a local HTTPS proxy which replaces the fields of the URL on behalf of the client device (e.g. local squid which cooperates with the client to get the HTTPS traffic).
In a variant embodiment, other than using URLs in the steps 4, 5 and 8, messages with same parameters as the URLs are used.
Further, a variant embodiment consists in having the client device report the playback context in HTTP headers rather than modifying the URL.
In the above embodiment, all placeholders in the movie are used to present the same advertisement. In a variant embodiment, each placeholder is associated with meta data describing scene context. So for each placeholder, a 2-tuple (ad, score) is determined at the steps 3 and 7. And consequently, in the steps 2 and 6, parameter about meta data of the placeholder is also sent, and the meta data is also considered when determining the advertisement in the recommender engine.
In the above embodiment, the geographical location is used to represent playback context. It shall note the playback context is not limited to the geographical location, and other kinds of information may also be taken into account, for example, device type, time etc. For example, an advertisement for a tea brand recommended at 5 pm would not be recommended at 8 pm when a pizza advertisement suits better; a promotion advertisement (i.e. pizza) determined at the playlist requesting time may expire at the playback time.
In a variant embodiment, the score returned by the recommender engine/ad-auction is weighted with the cost (cloud compute cost) necessary to generate the corresponding targeted chunk (i.e. computation for inlaying the advertisement into the portion of the movie). In general the cost of the second recommendation is higher: at the playback time the system has less time to generate the corresponding chunk and thus requires more cloud-compute power. A simple quality score/predication score would ignore this behavior, which may end up in situations where the system systematically takes only the second recommendation into account and consequently significantly increases the cloud-cost. In this variant the system divides the score returned by the recommender engine/ad-auction with the cloud-compute cost: weighted_score=score/cloud_cost. The resulting weighted score is included in the URL of the chunk. Doing so allows to compare the cost of the two recommendations: second_score/second_cost>?first_score/first_cost. So the system uses the advertisement with higher weighted score.
It is noted that the word “register” used in the description of memories 310 and 320 designates in each of the mentioned memories, a low-capacity memory zone capable of storing some binary data, as well as a high-capacity memory zone, capable of storing an executable program, or a whole data set.
Processing unit 311 can be implemented as a microprocessor, a custom chip, a dedicated (micro-) controller, and so on. Non-volatile memory NVM 310 can be implemented in any form of non-volatile memory, such as a hard disk, non-volatile random-access memory, EPROM (Erasable Programmable ROM), and so on. The Non-volatile memory NVM 310 comprises notably a register 3201 that holds a program representing an executable program comprising the method according to the invention. When powered up, the processing unit 311 loads the instructions comprised in NVM register 3101, copies them to VM register 3201, and executes them.
The VM memory 320 comprises notably:
In this embodiment, the network interface 313 is used to implement the different transmitter and receiver functions of the receiver device.
According to a particular embodiment, all entities in the
Other device architectures than illustrated by
Specifically, as to the client device, its CPU 1) uses the network interface to send request and receive response, 2) uses the NVM to store the received playlist and 3) decodes data specified by the URL so as to present the content to the user.
As to the playlist generation server, its CPU 1) uses the network interface to receive the request for playlist from the client device, send request with parameters of user ID and movie ID to the recommender engine, receive a response including two parameters indicating the advertisement and its score and send a generated playlist to the client device, 2) obtains the user ID and the movie ID from the client device and 3) generates the playlist based on the received response.
As to the recommender engine, its CPU 1) uses the network interface to receive the request from the playlist generation server, send the response including two parameters indicating the advertisement and its score to the playlist generation server, receive a request with parameters of the user ID, the movie ID and the playback context from the URL playback server and send a response with two parameters indicating the advertisement and its score to the URL playback server, 2) determines the advertisement and its score based on the user ID and the movie ID conveyed in the request sent from the playlist generation server by using a first determination algorithm and determines the advertisement and its score based on the user ID, the movie ID and the playback context conveyed in the request sent from the URL playback server by using a second determination algorithm.
As to the URL playback server, its CPU 1) uses the network interface to receive a URL from the client device, wherein, for example the URL comprises parameters indicating the user ID, the movie ID, one or more pieces of advertisement, their corresponding scores and playback context, 2) uses the network interface to send a request with parameters of user ID, the movie ID and the playback context to the recommender engine, 3) uses the network interface to receive a response with parameters indicating one or more pieces of advertisement and their scores from the recommender engine, 4) determines whether or not the received scores from the recommender engine are larger than the scored conveyed in the URL with a predefined value or by a predefined percentage value, 5) if the determination is positive, modifies the URL to replace the original advertisement and scores with the ones in the response from the recommender engine; and if negative, does not modify the URL, and 6) uses the network interface to send the modified URL to the client device so as to redirect the client device to consume the new advertisement. According to a variant embodiment, the CPU weights the scores with computation cost for generating the advertisement, and compares the weighted scores for deciding which advertisement is going to be used.
A number of implementations have been described. Nevertheless, it will be understood that various modifications may be made. For example, elements of different implementations may be combined, supplemented, modified, or removed to produce other implementations. Additionally, one of ordinary skill will understand that other structures and processes may be substituted for those disclosed and the resulting implementations will perform at least substantially the same function(s), in at least substantially the same way(s), to achieve at least substantially the same result(s) as the implementations disclosed. Accordingly, these and other implementations are contemplated by this application and are within the scope of the invention as defined by the appended claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 13306848.6 | Dec 2013 | EP | regional |
