TWI637629B - Video stream delivering method and apparatus using the same - Google Patents

Abstract

The invention provides a video stream distribution method and a device thereof. The foregoing method includes: transmitting a plurality of first video stream streams to the plurality of video stream terminal devices according to the plurality of first tile feature values; collecting a plurality of users from the video streaming terminal device in the preset time interval for the foregoing a plurality of point reading records performed by the video streaming terminal device; analyzing the foregoing point reading record to find a plurality of second party feature values; and updating the first piece feature value based on the second piece feature value to generate a plurality of Updating a tile feature value; and transmitting a plurality of second video streams to the aforementioned video stream terminal device according to the updated tile feature value.

Description

Video stream distribution method and device thereof

The present invention relates to a video stream distribution method and apparatus thereof, and more particularly to a video stream distribution method and apparatus for updating a party policy through a learning function.

At present, the multimedia on-demand video (Multimedia on Demand, MOD) video streaming system has been able to collect information from users in each video server in the Internet Protocol Television (IPTV) system. Read the information and refer to the system information to generate a pie strategy. However, since the user's needs and viewing behavior change at any time, the original parameters and feature values are given and cannot be changed in time. Therefore, after a period of time, all parameters and feature values may not be consistent with the user. Demand status. If you want to change it, managers need to study for a while to find the appropriate parameter values and eigenvalues. What's more, the original parameters and eigenvalues may not meet the user's needs at all, so it is necessary to generate new parameters and eigenvalues to meet the current situation. In this case, the manager must re-edit or re-write the scheduler, which is a waste of time.

In view of this, the present invention provides a video streaming method and apparatus thereof, which can continuously update a party policy based on collected user click records, so that the user can preferentially see the video that best matches his viewing preferences. Streaming.

The present invention provides a video stream distribution method, including: transmitting a plurality of first video streams to a plurality of video stream terminal devices according to a plurality of first party feature values; collecting a plurality of users from the video stream terminal devices a plurality of point reading records for the foregoing video streaming terminal device in a predetermined time interval; analyzing the foregoing point reading record to find a plurality of second party feature values; and updating the foregoing based on the second piece feature value a slice feature value to generate a plurality of updated tile feature values; and a plurality of second video stream streams are sent to the aforementioned video stream terminal device according to the updated tile feature value.

In addition, the present invention also provides a video streaming chip device, including a storage circuit, a transceiver circuit and a processor. The storage circuit stores a plurality of modules. The transceiver circuit is connected to a plurality of video streaming terminal devices. The processor is connected to the storage circuit and the transceiver circuit, and accesses the module to perform the following steps: sending a plurality of first video streams to the video stream terminal device according to the plurality of first component feature values; The streaming terminal device collects a plurality of point reading records of the plurality of users for the foregoing video streaming terminal device in a preset time interval; analyzing the foregoing point reading record to find a plurality of second party feature values; The second tile feature value updates the first tile feature value to generate a plurality of updated tile feature values; and dispatches the plurality of second video stream streams to the aforementioned video stream terminal device according to the updated tile feature value.

Based on the above, the present invention provides a video stream distribution method and a device thereof, which can analyze and learn a user's point-and-click record in a video streaming terminal device (for example, a MOD set-top box), thereby finding a user's preferred viewing. The feature stream type, time period, language and other feature values, and the patching strategy is updated to allow the video stream having the aforementioned feature value to be preferentially displayed to the user for viewing, so that the user does not need to click through multiple times to find the desired Watch the video stream. Thereby, the user experience can be improved.

The above described features and advantages of the invention will be apparent from the following description.

FIG. 1 is a schematic diagram of a video streaming party device and a video stream terminal device according to an embodiment of the invention. In this embodiment, the video stream distribution device 100 includes a storage circuit 110, a transceiver circuit 120, and a processor 130. The video streaming device 100 can be configured, for example, on a multimedia content transmission platform maintained by a telecommunication service provider, and can be used to determine a distribution policy of a plurality of video streaming terminal devices 20, wherein each video streaming terminal device 20 is, for example, deployed. A MOD set-top box, an IPTV device, and/or an OTT (over-the-top) device in a plurality of users' homes or the like, but the present invention is not limited thereto.

The storage circuit 110 is, for example, a memory, a hard disk, or any other component that can be used to store data, and can be used to record a plurality of code codes or modules. The transceiver circuit 120 includes, for example, a transmitter circuit, a receiver circuit, an analog to digital converter, a digital to analog converter, a low noise amplifier, a mixer, a filter, an impedance matching circuit, a transmission line, a power amplifier, one or more The antenna circuit and the internal storage medium are used to enable the video streamer device 100 to perform a wired or wireless signal transmission or reception function with the video stream terminal device 20, but the present invention is not limited thereto.

The processor 130 is coupled to the storage circuit 110 and the transceiver circuit 120. The processor 120 is, for example, a general purpose processor, a special purpose processor, a conventional processor, a digital signal processor, a plurality of microprocessors, one or more microprocessors combined with a digital signal processor core, and controls. , Microcontroller, Application Specific Integrated Circuit (ASIC), Field Programmable Gate Array (FPGA), any other kind of integrated circuit, state machine, based on advanced reduced instructions Advanced RISC Machine (ARM) processors and similar products.

2 is a flow chart of a video streaming method according to an embodiment of the invention. The method of this embodiment can be performed by the video stream distribution device 100 of FIG. 1, and the details of the steps of the method will be described below with the components shown in FIG.

In step S210, the processor 130 dispatches a plurality of first video stream streams to the plurality of video stream terminal devices 20 according to the plurality of first tile feature values. In an embodiment, the first component feature value is, for example, a movie attribute, a category, a language, and the like initially set by an administrator according to an original party policy, such as a comedy movie, an action movie, a children movie, a drama movie, and a horror movie. , variety shows, English, Japanese, Mandarin, Taiwanese, etc., but not limited to this.

For example, if a manager wants to send an English comedy movie to each video streaming terminal device 20, the manager can use "English" and "comedy film" as the first feature value of the first party, and according to the video. The streaming terminal device 20 delivers the first video stream, and the first video stream is a video stream having attributes such as "English" and "comedy". As another example, if the manager wants to deliver the variety program of the Mandarin language to each of the video streaming terminal devices 20, the manager can use "Mandarin" and "variety program" as the first feature value of the first party, and according to Each of the video streaming terminal devices 20 delivers a first video stream, and the first video stream stream is a video stream having attributes such as "Mandarin" and "variety show".

Next, in step S220, the processor 130 collects, from each of the video streaming terminal devices 20, a plurality of click recordings performed by the user for each of the video streaming terminal devices 20 within a predetermined time interval. In various embodiments, the aforementioned predetermined time interval is, for example, days, weeks, months, or even years. In addition, the aforementioned review record may include, for example, all behaviors each time the user uses the MOD, including the boot time, the shutdown time, the time spent on the home page, the selected channel, the selected movie, the viewed movie/ The type of program, when to watch, how long to watch, whether there is fast forwarding, whether there is playback, etc., but is not limited to this.

Thereafter, in step S230, the processor 130 analyzes the aforementioned point-and-click record to find a plurality of second tile feature values. In various embodiments, the processor 130 may analyze the aforementioned point-and-click records based on a multi-layer neural network, nearest neighbor method, simple Bayesian classification, decision tree, support vector machine, and latest deep learning algorithms. (learning) the second feature value, but is not limited thereto. Specifically, since the first video stream delivered by the video streaming device 100 according to the dispatching strategy formulated by the manager may not meet the needs of the user, the user may find out what he wants to watch. Video streaming to watch. In this case, the processor 130 can learn (learn) the feature value, the category, the language, and the like of the video stream that the user prefers by using the received point-and-click record, that is, the second piece feature value. However, the invention is not limited thereto.

In step S240, the processor 130 updates the foregoing first tile feature value based on the foregoing second tile feature value to generate a plurality of updated tile feature values, and then, in step S250, the respective tile features according to the updated tile feature value. The streaming terminal device 20 delivers a plurality of second video streams.

Therefore, the video streaming method and apparatus provided by the present invention can continuously adjust the dispatching strategy according to the preference of the user, so as to achieve a better sending effect, thereby providing a better user experience.

In a different embodiment, the processor 130 may perform the step S240 according to different manners, which are described in the following several embodiments, but are not intended to limit the possible embodiments of the present invention.

Embodiment (1): In this embodiment, the processor 130 may take out one of the second tile feature values (hereinafter referred to as a specific feature value), and determine whether the first tile feature value includes the specific feature. value. If so, the processor 130 ignores the particular feature value; if not, the processor 130 adds the particular feature value to the first tile feature value to generate the updated tile feature value.

For example, assume that the first tile feature value includes attributes such as "comedy piece", "action piece", and "child piece", and the processor 130 finds a "drama piece" based on the collected point-reading record (ie, second The feature value of the tile is also a favorite attribute of the user. In this case, since the first feature feature value does not include a "drama", the processor 130 may add a "drama" to the first tile feature value to generate an updated tile feature value. The update feature feature values in this example include "comedy film", "action film", "child film" and "drama film".

Then, the processor 130 can send a second video stream having attributes such as "comedy piece", "action piece", "child piece" and "drama piece" to each video streaming terminal device 20, thereby providing the user with the video stream. A movie that better matches its preferences.

On the other hand, if the second component feature value found by the processor 130 based on the collected click record is already present in the first tile feature value (eg, "action slice"), the processor 130 may directly This second fragment feature value is ignored.

Embodiment (2): In this embodiment, the processor 130 may take out one of the second tile feature values (hereinafter also referred to as a specific feature value), and determine the specific feature value and any of the first tiles. Whether there is a affiliation between the feature values, and if so, the specific feature value is added to the first tile feature value according to the affiliation to generate the updated tile feature value.

In various embodiments, the administrator can set the affiliation between the feature values by himself. In order to make this concept clearer, an example is provided below for explanation.

Please refer to Table 1 and Table 2 below, wherein Table 1 includes, for example, the initially set first tile feature value, and Table 2 includes, for example, a second tile feature value analyzed by the user's click record analysis.  <TABLE border="1" borderColor="#000000" width="85%"><TBODY><tr><td> First dispatch feature value</td><td> times</td></tr ><tr><td> Comedy Movies</td><td> 10020 </td></tr><tr><td> Action Movies</td><td> 3800 </td></tr>< Tr><td> Drama Film</td><td> 4586 </td></tr></TBODY></TABLE> Table 1  <TABLE border="1" borderColor="#000000" width="85%"><TBODY><tr><td> Second feature value</td><td> times</td><td> Time period </td><td> times </td></tr><tr><td> comedy film</td><td> 23020 </td><td> 7:00-12 </td>< Td> 16023 </td></tr><tr><td> 13:18-18</td><td> 4230 </td></tr><tr><td> 19:00-6 /td><td> 2767 </td></tr><tr><td> Action Movies</td><td> 12800 </td><td> 7:00-11 </td><td> 875 </td></tr><tr><td> 13:18 </td><td> 7925 </td></tr><tr><td> 19:00-6 </td ><td> 400 </td></tr><tr><td> Drama Film</td><td> 54586 </td><td> 7:00 to -12</td><td> 10123 < /td></tr><tr><td> 13:18 </td><td> 4230 </td></tr><tr><td> 19:00-6 </td>< Td> 40233 </td></tr></TBODY></TABLE> Table 2  

As can be seen from Table 1, the number of viewings of comedy films is higher than that of action movies and children's films. However, as can be seen from Table 2 (which can be analyzed by algorithms such as multi-layer neural networks), there are more time periods "7: -12", "13: -18" and "19" - 6 points" and other second feature values. Moreover, Table 2 also shows that the comedy film has a higher number of plays in the period "7:00 - 12 o'clock", which may reflect that the user has a "comedy piece" attribute in the time period "7 o'clock - 12 o'clock". The video stream is streamed for viewing. Therefore, the processor 130 can regard the time period "7 o'clock - 12 o't" as a specific feature value subordinate to the "comedy piece", and accordingly distribute the second video stream. For example, the processor 130 can forward the video stream having the "comedy piece" attribute to the user in the period "7 o'clock - 12 o'clock", but the present invention is not limited thereto.

Similarly, based on the contents of Table 2, the processor 130 can then deliver the video stream with the "comedy" attribute to the user during the time period "13: 18:00", and deliver it at the time "19:00-6". The video stream having the "drama" attribute flows to the user, but the present invention is not limited thereto.

In addition, although the above periods are regarded as subordinate to the video stream type (for example, comedy, action, and feature films, etc.), the manager can also adjust the affiliation between the time slot and the video stream type according to the needs. (For example, set the video stream type to the slave time period).

On the other hand, if there is no affiliation between the specific feature value and any of the first tile feature values, the processor 130 may determine whether the first tile feature value includes the specific feature value, and if so, the specific The feature value is ignored, otherwise the specific feature value is added to the first tile feature value. An example of the same is provided below.

Please refer to Table 3 below and Table 4, wherein Table 3 includes, for example, the first set of feature values of the initial settings, and Table 4 includes, for example, the second tile feature values analyzed by the user's click record analysis.  <TABLE border="1" borderColor="#000000" width="85%"><TBODY><tr><td> First dispatch feature value</td><td> times</td></tr ><tr><td> Action Movies</td><td> 19030 </td></tr><tr><td> Comedy Movies</td><td> 12800 </td></tr>< Tr><td> horror film</td><td> 9860 </td></tr></TBODY></TABLE>Table 3  <TABLE border="1" borderColor="#000000" width="85%"><TBODY><tr><td> Language</td><td> Second Part Feature Value</td><td> Times</td></tr><tr><td> Mandarin </td><td> Comedy </td><td> 18123 </td></tr><tr><td> Taiwanese< /td><td> Grassroots</td><td> 16123 </td></tr><tr><td> Mandarin</td><td> Action Movies</td><td> 14230 </ Td></tr><tr><td> Mandarin</td><td> Children's Film</td><td> 12767 </td></tr><tr><td> Taiwanese</td> <td> Variety class</td><td> 4230 </td></tr><tr><td> Taiwanese</td><td> Action movie</td><td> 2767 </td> </tr></TBODY></TABLE> Table 4  

As can be seen from Table 3, the number of views of the action movie is higher than that of the comedy film and the horror film. However, as can be seen from Table 4, the second feature values of the language "Mandarin" and "Taiwanese" are added. However, unlike Table 2, there seems to be no obvious affiliation between language and video stream types, but a parallel relationship. Moreover, since the original first feature value does not include specific feature values such as "Mandarin" and "Taiwanese", the processor 130 can add "Mandarin" and "Taiwanese" to the first feature value. . The first feature values in this example include "comedy film", "grassroot", "action film", "children", "drama", "variety", "Mandarin" and "Taiwanese". .

On the other hand, if the specific feature value learned by the processor 130 is "comedy", "grassroot" or other members already existing in the first feature value, the processor 130 can directly ignore it.

Next, the processor 130 may determine whether the number of the first tile feature values has reached a predetermined number, and if so, delete one of the first tile feature values to generate an updated tile feature value. In particular, when the video streamer device 100 continuously learns and operates for a period of time, the tile feature value considered at the time of dispatching may substantially contain all possible tile feature values, which will cause it to be sent. The film strategy is like a dummy. In view of this, the processor 130 may delete one or more of the first tile feature values in time when the number of first tile feature values reaches a preset number (eg, an arbitrary integer) (eg, has a minimum usage) One or more of the frequencies, but not limited to this, to generate updated tile feature values. Thereby, the video streaming device 100 can continuously deliver the video stream to the user in a party policy that best matches the user's preference.

Embodiment (3): In this embodiment, the video streaming terminal device 20 can be distinguished to correspond to a plurality of regions (for example, north, central, south, provinces, autonomous regions, counties, and the like of a country, but The plurality of device groups, which are not limited thereto, the point record can be divided into a plurality of record groups corresponding to the foregoing region, and the second tile feature value is divided into a plurality of feature value groups corresponding to the regions group.

Then, the processor 130 may extract the first feature value group in the feature value group, where the first feature value group corresponds to the first device group after the device group (including, for example, the video stream in the A province) Terminal device 20). Then, the processor 130 may update the first tile feature value to the plurality of first update feature values (for example, "comedy piece" and "grassroot") of the first device group according to the first feature value group. Where the first updated feature value is included in the aforementioned updated feature value.

Similarly, the processor 130 may extract the second feature value group in the feature value group, where the second feature value group corresponds to the second device group after the device group (for example, including the video stream in the B province) Terminal device 20). Then, the processor 130 may update the second tile feature value to a plurality of second update feature values specific to the second device group according to the second feature value group (eg, “comedy film”, “grassroot”, The "action piece" and the "child piece", wherein the second updated feature value is included in the updated feature value.

Afterwards, the processor 130 may send the first part of the second video stream (for example, the video stream having the "comedy piece" and the "grassroot" attribute) to the first device group according to the first updated feature value, and according to The second updated feature value sends a second portion of the second video stream to the second device group (eg, a video stream having "comedy", "grassroot", "action", and "child" attributes) . That is, the processor 130 may separately find the video stream type that the user in the A province and the B province prefers to watch based on the click recording records of the video streaming terminal device 20 belonging to the A province and the B province, and according to Individually determine the most appropriate placement strategy for users in Province A and Province B. In this way, the processor 130 can send a corresponding video stream to the viewer according to the different needs of the users in each region, thereby improving the user experience.

In other embodiments, the relevant personnel may also change the above provinces into counties, municipalities, autonomous regions, or other similar geographic regions as needed.

In summary, the present invention provides a video and audio streaming method and apparatus thereof, which can analyze and learn a user's point-and-click record in a video streaming terminal device (for example, a MOD set-top box), thereby finding a user. The feature value of the video stream type, the time period, the language, and the like are preferred, and the party strategy is updated to preferentially display the video stream having the foregoing feature value to the user for viewing, so that the user does not need to click multiple times. Find the video stream you want to watch. Thereby, the user experience can be improved.

Moreover, the present invention can also remove infrequently used feature values in a timely manner, so that the video streamer device continuously delivers the video stream to the user in a tile policy that best matches the user's preference. In addition, the present invention can also distribute video and audio streams that best suit the viewing habits of the region to users in different regions, so that users in various regions can enjoy video streams that match their preferences without being used by users in other regions. The influence of viewing preferences.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

100‧‧‧Video streamer device

110‧‧‧Storage circuit

120‧‧‧ transceiver circuit

130‧‧‧Processor

20‧‧‧Video Streaming Terminal Equipment

S210~S250‧‧‧Steps

FIG. 1 is a schematic diagram of a video streaming party device and a video stream terminal device according to an embodiment of the invention. 2 is a flow chart of a video streaming method according to an embodiment of the invention.

Claims (12)

A video streaming method includes: transmitting a plurality of first video streams to a plurality of video stream terminal devices according to a plurality of first party feature values; collecting a plurality of users from the video stream terminal devices And a plurality of point reading records for the video streaming terminal devices in the preset time interval; analyzing the point reading records to find a plurality of second party feature values; and updating the second piece feature values based on the second piece feature values The first slice feature values are used to generate a plurality of updated tile feature values, including: extracting one of the second tile feature values; and determining the one of the second tile feature values Whether there is a dependency relationship between any of the first tile feature values; if yes, adding the one of the second tile feature values to the first tile feature values according to the dependency relationship to generate the Updating the tile feature values; if not, determining whether the first tile feature values include the one of the second tile feature values, and if so, the second tile feature values The person ignores, otherwise the second party The one of the values is added to the first piece feature values; determining whether a quantity of the first piece feature values has reached a predetermined number, and if so, the first piece feature values One of them is deleted to generate the updated tile feature values; And transmitting, by the updated image feature values, the plurality of second video streams to the video streaming terminal devices. The method of claim 1, wherein the video streaming terminal devices are divided into a plurality of device groups corresponding to a plurality of regions, the point-by-point records being classified as corresponding to the plurality of regions Record group, the second tile feature values are divided into a plurality of feature value groups corresponding to the regions, and the step of updating the first tile feature values based on the second tile feature values The method includes: extracting a first feature value group in the group of feature values, where the first feature value group corresponds to a first device group after the device group; according to the first feature value group The group updates the first fragment feature values to a plurality of first update feature values that are specific to the first device group, wherein the first updated feature values are included in the updated feature values; and the features are extracted a second group of feature values in the group of values, wherein the second group of feature values corresponds to a second group of devices after the group of devices; and the group of the second group of feature values The second tile feature value is updated to a plurality of the second device group And updating the feature values, wherein the second updated feature values are included in the updated feature values. The method of claim 2, wherein the step of dispatching the second video streams to the video stream terminal devices according to the updated feature values comprises: determining, according to the first updated feature values The first device group dispatches a first portion of the second video streams; And transmitting a second portion of the second video streams to the second device group according to the second updated feature values. The method of claim 1, wherein the updating the first tile feature values based on the second tile feature values to generate the updated tile feature values comprises: extracting the second parties One of the slice feature values; determining whether the first tile feature values have included the one of the second tile feature values; if so, the second tile feature values are Ignore; if not, add the one of the second tile feature values to the first tile feature values to generate the updated tile feature values. The method of claim 1, wherein the one of the first component feature values has a lowest frequency of use among the first component feature values. The method of claim 1, wherein the preset time interval is at least a plurality of days. A video streaming chip device includes: a storage circuit for storing a plurality of modules; a transceiver circuit for connecting a plurality of video streaming terminal devices; and a processor connecting the storage circuit and the transceiver circuit for accessing The modules perform the following steps: And transmitting, by the plurality of first component feature values, the plurality of first video streams to the video streaming terminal devices; collecting, by the video streaming terminal devices, the plurality of users for the audio and video in a preset time interval a plurality of point reading records performed by the streaming terminal device; analyzing the point reading records to find a plurality of second piece feature values; updating the first piece feature values based on the second piece feature values to generate And a plurality of updated tile feature values, including: extracting one of the second tile feature values; determining the one of the second tile feature values and any of the first tile features Whether there is a dependency relationship between the values; if yes, adding the one of the second component feature values to the first tile feature values according to the affiliation to generate the updated tile feature values; Determining whether the first piece feature value includes the one of the second piece feature values, and if so, ignoring the one of the second piece feature values, and vice versa One of the second feature values is added to the first a piece of feature value; determining whether a quantity of the first piece feature values has reached a predetermined number, and if so, deleting one of the first piece feature values to generate the updated piece features And outputting a plurality of second video streams to the video stream terminal devices according to the updated tile feature values. The device of claim 7, wherein the video streaming terminal devices are divided into a plurality of device groups corresponding to a plurality of regions, the click recording records being classified as corresponding to the plurality of regions Record group, the second tile feature values are divided into a plurality of feature value groups corresponding to the regions, and the processor is configured to: extract a first feature of the feature value groups a value group, wherein the first feature value group corresponds to a first device group after the device group; and the first feature feature values are updated to be specific to the first feature value group a plurality of first updated feature values of the first device group, wherein the first updated feature values are included in the updated feature values; and a second feature value group of the feature value groups is extracted, wherein the The second feature value group corresponds to a second device group after the device groups; and the second component feature values are updated to be specific to the second device group according to the second feature value group a plurality of second updated feature values, wherein the second updated feature value packets Contained in these updated feature values. The device of claim 8, wherein the processor is configured to: dispatch a first portion of the second video streams to the first device group based on the first updated feature values; And transmitting a second portion of the second video streams to the second device group according to the second updated feature values. The device of claim 7, wherein the processor is configured to: extract one of the second component feature values; and determine whether the first component feature values include the And the one of the second tile feature values; if yes, the one of the second tile feature values is ignored; if not, adding the one of the second tile feature values to the one The first tile feature values are used to generate the updated tile feature values. The device of claim 7, wherein the one of the first component feature values has a lowest frequency of use among the first component feature values. The device of claim 7, wherein the preset time interval is at least a plurality of days.