JP7402444B2 - Message distribution system and message distribution method - Google Patents

Description

The disclosure herein relates to information communication technology, and particularly to a message distribution system and a message distribution method.

With the vigorous development of the Internet and mobile networks, electronic products (mobile phones, tablet terminals, desktop computers, notebook computers, smart home appliances, etc.) with network access functions have become widely used by the public. Furthermore, a large amount of information exists on the network and can be easily obtained. Some apps and platforms constantly distribute messages to users' devices through push notifications to get people's attention.

Current push notifications are sent automatically or manually to bring users back to the platform and create or increase affinity with the platform. Current systems are not personalized and are limited to streamers a user follows, high-performing streamers, or offers and promotions.

However, if the content or timing of the notification is inappropriate, it may fail to attract people's attention and may sometimes irritate the user. For this reason, personalized notifications that provide notifications with the right content at the right time have become an extremely important issue.

A message distribution system according to an embodiment of the invention includes a collection unit and an evaluation unit, the collection unit configured to collect feedback regarding the message from a user terminal, and the evaluation unit connected to the collection unit. and configured to evaluate the preference of the user terminal for the message based on the feedback, the feedback including response action feedback and/or influence action feedback, and the response action feedback is configured to evaluate the preference of the user terminal for the message. and the influencing action feedback includes actions after the manipulation.

A message distribution method according to another embodiment of the present invention includes the steps of collecting feedback regarding a message from a user terminal, and evaluating the preference of the user terminal for the message based on the feedback, includes response action feedback and influence action feedback, the response action feedback includes an operation on the message by the user terminal, and the influence action feedback includes a behavior after the operation.

The technology disclosed herein can improve the ease with which a user can find the content he or she is looking for, improve the satisfaction level of the message for the user, and optimize the user experience.

1 is a schematic diagram showing the configuration of a communication system 1 based on some embodiments of the present invention. FIG. 1 is a block diagram of a message distribution system 100 according to some embodiments of the invention. 1 is a schematic diagram illustrating the configuration of a communication system 1' according to some embodiments of the present invention; FIG. 1 is a block diagram of a message distribution system 100' according to some embodiments of the invention. FIG. FIG. 2 is a block diagram of a preference evaluation element 124 according to some embodiments of the invention.

FIG. 1 is a schematic diagram showing the configuration of a communication system 1 based on some embodiments of the present invention. The communication system 1 includes a user terminal 10 and a server 20. The user terminal 10 and the server 20 are connected via a network 90, such as the Internet. Server 20 includes a message distribution system 100. The message distribution system 100 is configured to send a message M to the user terminal 10 and receive feedback F for the message M from the user terminal 10. In some implementations, message M can be a text message, email, speech bubble, push message, push notification, etc. In some implementations, the content of the message M can be text, images, sound, audio, video, live streaming, podcasting, sales, etc. In some implementations, the user terminal 10 may be a terminal used by a user. The terminal may be an electronic product. The user may be called a viewer, streamer, podcaster, audience, listener, or the like. The communication system 1 may include a plurality of user terminals 10, and one user terminal 10 is shown in FIGS. 1 and 2 for simplicity.

FIG. 2 is a block diagram of a message distribution system 100 according to some embodiments of the invention. The message distribution system 100 includes a collection unit 110, an evaluation unit 120, and a push unit 130. The collection unit 110 is configured to collect feedback F for the message M from the user terminal 10 . The evaluation unit 120 is connected to the collection unit 110 and configured to evaluate the preference P of the user terminal 10 for the message M based on the feedback F. The push unit 130 is connected to the evaluation unit 120 and configured to push the message M to the user terminal 10 .

As shown in FIG. 2, the push unit 130 pushes the message M to the user terminal 10, and the collection unit 110 collects the feedback F for the message M from the user terminal 10. The collection unit 110 transmits the feedback F for the message M to the evaluation unit 120, and the evaluation unit 120 evaluates the preference P of the user terminal 10 for the message M based on the feedback F. Thereafter, the push unit 130 receives the preference P from the evaluation unit 120. If the user terminal 10 has the preference P for the message M, the push unit 130 pushes the message M to the user terminal 10. On the other hand, if the user terminal 10 does not have preference P for the message M, the push unit 130 stops pushing the message M to the user terminal 10.

The feedback F can refer to an action that the user terminal 10 takes in response to the message M. In some implementations, the feedback F may include response action feedback Fr and influence action feedback Fi. The response action feedback Fr can refer to an operation performed by the user terminal 10 on the message M. The influencing action feedback Fi can refer to the behavior after the operation. When the user terminal 10 takes an action on the message M, the user terminal 10 transmits data to the message distribution system 100. Different data correspond to different action types. The evaluation unit 120 evaluates the preference P for the message M based on different types of the data.

In some embodiments, the operations include clicking, swiping, deleting, ignoring, muting, turning off notifications, expanding, minimizing, and other operations that the user terminal 10 may perform on the message M, and combinations thereof. may be included. For example, the operation may include an operation of clicking the message M, or an operation of swiping and deleting the message M. In some embodiments, the operation may include ignoring the message M, turning off the push notification function of the terminal, stopping delivery of the push notification of the message M, and the like. In some implementations, the message M may be expanded from a simplified message M to a complete message M. The operation may further include an operation of expanding and clicking the message M, an operation of expanding and swiping the message M to delete it, and the like.

In some embodiments, the evaluation unit 120 evaluates the preference P of the user terminal 10 based on the operation. For example, when the user terminal 10 operates the message M and opens the application, this indicates that the user terminal 10 has the preference P for the message M. On the other hand, if the user terminal 10 operates the message but does not open the application, this indicates that the user terminal 10 does not have preference P for the message M. For example, when the user terminal 10 clicks on the message M, the evaluation unit 120 evaluates that the user terminal 10 has a preference P for the message M, and the evaluation unit 120 evaluates that the user terminal 10 has a preference P for the message M. If the message M is deleted by swiping, the user terminal 10 evaluates that the message M does not have preference P. When the user terminal 10 operates the message M in a certain manner, the user terminal 10 transmits data to the message distribution system 100. Different data correspond to different operation types. The evaluation unit 120 evaluates the preference P for the message M based on different types of the data.

In some implementations, the preference P may include a preference P for simplified messages M and a preference P for complete messages M. For example, if the user terminal 10 expands the simplified message M into the complete message M and clicks on the message M, the user terminal 10 expands the simplified message M into the complete message M and clicks on the message M. indicates that it has the preference P for . On the other hand, if the user terminal 10 expands the simplified message M into a complete message M but does not click on the message M, the user terminal 10 expands the simplified message M into a complete message M, but if the user terminal 10 does not click on the message M, the user terminal 10 shows that it has a degree P, but no preference for a complete message M. If the user terminal 10 does not expand or click on the message M, the user terminal 10 has no preference P for the simplified message M and the complete message M. shows.

In some implementations, such actions include idleness, searching, visiting, following, purchasing, submitting, watching, commenting, supporting, marking, collecting, claiming, or any other actions that the user may take with the app. Combinations may also be included. That is, the action may be an action taken by the user terminal 10 after operating the message M, or the state of the user terminal 10. For example, the user terminal 10 may be idle for a while on the front page and then exit the application. In some implementations, the user terminal 10 can search for streamers and events, access streamer profiles, and explore sections. In some implementations, the user terminal 10 can enter a live room to chat with the streamer and send gifts to the streamer. In some implementations, the user terminal 10 can comment on articles and events. In some implementations, the user terminal 10 can purchase gifts, points, and memberships at an online store. In some implementations, the user terminal 10 can follow particular streamers, support streamers, and mark users as loyal fans or favorite streamers.

In some embodiments, the evaluation unit 120 evaluates the preference P of the user terminal 10 based on the behavior. When the user terminal 10 spends time within the application and takes some action, this indicates that the user terminal 10 has a preference level P for the message M. On the other hand, if the user terminal 10 is in an idle state and closes the application without doing anything, this indicates that the user terminal 10 does not have preference P for the message M. For example, when the user terminal 10 opens an app and searches for a streamer, the evaluation unit 120 evaluates that the user terminal 10 has preference P for the message M. On the other hand, if the user terminal 10 opens the app but immediately closes the app without viewing any distribution, the evaluation unit 120 determines that the user terminal 10 has a preference P for the message M. Evaluated as not having any. When the user terminal 10 performs a certain action after opening the message M, the user terminal 10 transmits data to the message distribution system 100. Different types of behavior correspond to different data. The evaluation unit 120 evaluates the preference P for the message M based on different types of the data.

In some implementations, the collection unit 110 dynamically collects feedback F for messages M from the user terminal 10. More specifically, the collection unit 110 not only collects real-time feedback F, but also collects historical feedback F. That is, the feedback F includes real-time feedback F and historical feedback F. More specifically, the response action feedback Fr further includes real-time response action feedback Fr and historical response action feedback Fr. Similarly, the influencing action feedback Fi further includes real-time influencing action feedback Fi and historical influencing action feedback Fi.

As shown in FIG. 2, the evaluation unit 120 further includes a message analysis element 121, a frequency analysis element 122, and a relevance analysis element 123. The message analysis element 121 is configured to analyze, filter and/or screen user messages Mu left by the user terminal 10 within the app. The frequency analysis element 122 is configured to identify the frequency Freq of the feedback F of the user terminal 10' with respect to the message M. The relevance analysis element 123 is configured to analyze the relevance R between the message M and the feedback F.

In some implementations, the message analysis element 121 is configured to identify the content of the user message Mu left by the user terminal 10. The user message Mu may be an article, comment, speech bubble, pictogram, etc., and the content of the user message Mu may be a text, image, sound, voice, video message, etc. The message analysis element 121 may be configured to determine whether the user message Mu (eg, part of the chat content) is positive, negative, or neutral. In some embodiments, the message analysis element 121 includes a keyword identification unit that identifies keywords in the user message Mu, and evaluates the preference P based on the keywords. For example, the message analysis element 121 identifies positive content such as "do", "good", "like", "like", "interesting", and a smiling emoticon included in the user message Mu, and User message Mu can be evaluated as positive. On the other hand, the message analysis element 121 identifies negative contents such as "no", "bad", "dislike", "boring", and sad emoticons included in the user message Mu, and determines whether the user message Mu is negative. It can be evaluated as being accurate. In some implementations, if no positive or negative content is identified, the message analysis component 121 may evaluate the user message Mu as neutral.

In some embodiments, the evaluation unit 120 evaluates the preference P of the user terminal 10 based on the user message Mu. For example, if the user terminal 10 comments on a streamer in a streaming room and the comment is positive, meaning that the user terminal 10 likes the streamer, the evaluation unit 120 It is evaluated that the user terminal 10 has a preference level P for the message M. On the other hand, if the user terminal 10 comments on the streamer and the comment is negative, this indicates that the user terminal 10 does not have preference P for the message M. In some implementations, if the user message Mu is neutral, the user message Mu does not contribute to the preference P.

In some implementations, the frequency analysis element 122 may determine the frequency Freq of the feedback F. More specifically, the frequency analysis element 122 can specify the frequency Freq of the response action feedback Fr and/or the influence action feedback Fi. The frequency analysis element 122 is configured to calculate the frequency Freq of operations and/or actions of the user terminal 10' and determine whether the frequency Freq is positive, negative, or neutral. Good too.

In some embodiments, the evaluation unit 120 evaluates the preference P of the user terminal 10 based on the frequency Fre from the frequency analysis element 122. If the user terminal 10 frequently performs a specific operation and/or action, this indicates that the user has a preference P for the message M. For example, if the user terminal 10 views a streamer after clicking the message M each time, the frequency analysis element 122 analyzes the frequency Freq as positive, and the evaluation unit 120 determines that the user terminal 10 views the message M. It is evaluated that it has a preference P for M. In some implementations, the frequency Freq may be referred to as the number of times a particular operation and/or action occurs. If the number of times is greater than, within, or less than a specific range, it indicates that the frequency Freq is high, medium, or low. When the evaluation unit 120 receives the data of the user terminal 10, the evaluation unit 120 evaluates the similarity of the types of the data with respect to the message M. If there is similar or the same type of data for the message M, the frequency analysis element 122 starts calculating the frequency Freq. Different frequencies Freq can contribute as different increases or decreases to the preference P.

In some implementations, the frequency analysis element 122 may further detect continuity of the operations and/or actions. For example, if the user terminal 10 frequently watches a certain streamer almost every day, but the user terminal 10 does not watch the streamer for one day, the frequency analysis element 122 determines that the frequency Freq is continuous. Detect that it is not. In such a case, the user terminal 10 may be busy and forget to watch the streamer, so the push unit 130 may push the message M or a reminder message to the user terminal 10. However, if the user terminal 10 does not click on the message M and no longer watches the streamer, the frequency analysis element 122 analyzes that the frequency Freq is negative. On the other hand, if the user terminal 10 clicks on the message M and starts viewing the streamer, the frequency analysis element 122 can analyze that the frequency Freq is positive.

In some implementations, the frequency analysis component 122 may further detect consistency of the operations and/or behaviors. For example, if the user terminal 10 views a certain streamer every time the user terminal 10 clicks the message M, but the user terminal 10 does not follow the streamer, the frequency analysis element 122 may discrepancies can be detected. In such a case, the push unit 130 may push a message M' related to the message M, such as recommending following the streamer or receiving a free gift by following the streamer. In some embodiments, the collection unit 110 further collects feedback F' for the message M', and the evaluation unit 120 further evaluates the preference P' of the user terminal 10 for the message M'. can do.

In some implementations, the relevance analysis component 123 may be configured to identify the relevance R between the message M and the feedback F. More specifically, the relevance analysis element 123 identifies the content of the message M and the operation and/or action of the user terminal 10, and determines the relevance R between the message M and the feedback F. Constructed to judge. The relevance analysis element 123 may be configured to determine whether the relevance R is relevant or unrelated.

In some embodiments, the evaluation unit 120 evaluates the preference P of the user terminal 10 based on the relevance R from the relevance analysis element 123. In some cases, the behavior of the user terminal 10' is unrelated to the content of the message M. For example, the message M may recommend viewing a live streaming of a streamer on the user terminal 10. Although the user terminal 10 clicks on the message M, the user terminal 10 instead of viewing the recommended streamer views another streamer. In such a case, the relationship R between the message M and the behavior is irrelevant, and the feedback F does not contribute to the preference P. On the other hand, when the relevance R is relevant, the feedback F can contribute to the preference P. Thereby, the accuracy of the preference P can be improved.

In some embodiments, when the relevance R is relevant, it may be further classified as positively relevant or negatively relevant. For example, the message M may recommend viewing a live streaming of a streamer on the user terminal 10. After clicking and viewing the streamer, the user terminal 10 may register the streamer as a channel or cancel the channel registration. The relevance analysis element 123 can determine that the relevance R is positively relevant or negatively relevant, and can contribute to the preference P by increasing or decreasing, respectively.

In some implementations, the feedback F may further include time information. The evaluation unit 120 further evaluates the preference P based on the time information. More specifically, the evaluation unit 120 further evaluates the preference level P based on the time information of the response action feedback Fr and the influence action feedback Fi. The time information may refer to time zones such as evening, morning, noon, night, weekdays, weekends, and holidays. Further, the time information may indicate a period involved in the application, such as 1 minute, 10 minutes, or more than 1 hour. The time information may refer to the time zone of the user terminal 10, such as Japan time or US time. For example, since the user terminal 10 tends to click on the message M and watch live streaming at night, the evaluation unit 120 determines that the user terminal 10 operates the message M and performs the action at night. It is evaluated that it has the preference P. Since the user terminal 10 always swipes and deletes the message M in the morning, this indicates that the user terminal 10 does not have preference P for the message M in the morning.

In some implementations, the feedback F may further include location information. The evaluation unit 120 further evaluates the preference P based on the position information. More specifically, the evaluation unit 120 further evaluates the preference P based on the position information of the response action feedback Fr and the influence action feedback Fi. The location information may refer to a country, city, town, home, workplace, restaurant, etc. Further, the location information may refer to a means of transportation such as a car, bus, train, ship, or airplane. For example, since the user terminal 10 clicks on the message M to watch the streaming at home, the evaluation unit 120 evaluates that the user terminal 10 has the preference P for operating the message M at home. . Since the user terminal 10 always swipes and deletes the message M at the workplace, this indicates that the user terminal 10 does not have a preference P for the message M at the workplace. In some implementations, the push unit 130 may further push the message M to other users with similar location information. For example, when a number of users in Taipei City visit Arthur's profile, the push unit 130 may push the message M recommending other users in Taipei City to watch the streamer.

In some implementations, the feedback F can further include device information. The evaluation unit 120 further evaluates the preference P based on the device information. More specifically, the evaluation unit 120 further evaluates the preference P based on the device information of the response action feedback Fr and the influence action feedback Fi. The user terminal 10 can generate an application using a terminal such as a mobile phone, a tablet, or a smart home appliance. The device information may refer to the current mode of the terminal, such as power off mode, standby mode, idle mode, busy mode, etc. Further, the device information may indicate a mode in which the user terminal 10 is set, such as an incoming call rejection mode, silent mode, or silent mode. For example, since the user tends to swipe and delete the message while in busy mode, the user terminal 10 indicates that it does not have preference P for the message M when in busy mode.

In some implementations, the device information may include an operation time between when the message is read and when the message is operated. The evaluation unit 120 can evaluate the preference P based on the operation time. In some embodiments, if the message M is manipulated immediately after it is read, the user terminal 10 tends not to check the content of the message M or does not like any message. may be indicated. For example, if the user terminal 10 clicks on the message M in less than 1 second, it may indicate that the user terminal 10 tends to always click on every message M without checking the message M. Therefore, the feedback F may not contribute to the preference. If the user terminal 10 swipes and deletes the message M in less than 1 second without checking the content of the message M, the user terminal 10 may indicate that it does not wish to receive any messages. .

In some implementations, the evaluation unit 120 may further evaluate the preference P based on a combination of the above-mentioned information. For example, if the user terminal 10 located in New York City frequently clicks on the message M to view live streaming in the evening, the evaluation unit 120 may collect the time information of "evening" and the location information of "New York City". It is evaluated that the user terminal 10 having the message M has a preference P for the message M. In some implementations, the combination of the above information may be determined according to actual needs.

In some embodiments, the preference P may be evaluated based on the retention rate of the user terminal. The retention rate can refer to the rate at which the user terminal 10 returns and opens the application after receiving the message M. In some implementations, the retention rate may be the number of times an app is opened or the length of engagement with an app over a period of time. In some embodiments, the fixed period of time may be 3 days, 7 days, 14 days, etc. For example, if the number of times the app is opened or the length of time associated with the app increases during the seven days after receiving the message M, this indicates that the user terminal 10 has the preference P for the message M.

In some implementations, the preference P may include other factors, such as push notification click rate, push notification cancellation rate, session length for opening and engaging with apps due to push notifications, and frequency of opening apps. May be evaluated based on criteria. When the click rate of the push notification becomes high, it indicates that the user terminal 10 has the preference P for the message M. When the push notification cancellation rate becomes high, it indicates that the user does not have the preference P for the message M. If the number of times the application is opened by clicking on the message M increases, this indicates that the user terminal 10 has the preference P for the message M. If the length of time involved in the application increases, this indicates that the user terminal 10 has the preference P for the message M. When the frequency of opening the app increases, this indicates that the user terminal 10 has the preference P for the message M.

In some implementations, the evaluation unit 120 may further include a preference evaluation element 124. The preference evaluation element 124 is configured to evaluate the preference P based on the feedback F using the criteria described above. In some implementations, the preference evaluation element 124 may evaluate the preference P by each of the criteria described above or by a combination of the criteria described above. FIG. 5 is a block diagram of preference evaluation element 124 according to some embodiments of the invention. The preference evaluation element 124 includes a criterion calculation section 1241, a criterion evaluation section 1242, a criterion module 1243, and a weighting module 1244. The criterion calculation unit 1241 is configured to calculate the preference P based on the criterion or the combination of criteria. The criterion evaluation unit 1242 is configured to evaluate the criterion and the weighting corresponding to each criterion in order to calculate the preference P. The criteria module 1243 is configured to store the criteria for calculating the preference P. The weighting module 1244 is configured to store the weighting for each criterion.

As shown in FIG. 5, the criterion evaluation unit 1242 acquires the weighting corresponding to the criterion from the criterion module 1243 and the weighting module 1244. The criterion evaluation unit 1242 can evaluate the corresponding weighting for each criterion and determine an appropriate weighting for each criterion. After the weighting for each criterion is determined, the criterion evaluator 1242 may update the weighting module 1244. Next, the criterion calculation unit 1241 can obtain the criterion and the corresponding weighting, and calculate the preference P. For example, the criterion evaluation unit 1242 may acquire several criteria such as criterion C1, criterion C2, criterion C3, etc. from the criterion module 1243. In some implementations, the criterion C1 may be the click rate of push notifications, the criterion C2 may be the length of sessions engaged with the app through push notifications, and the criterion C3 may be the frequency of opening the app. It's okay.

Once the criteria are determined, the criteria evaluation unit 1242 can evaluate the weighting for each criterion. For example, the criterion evaluation unit 1242 can evaluate weighting W1, weighting W2, and weighting W3 for criterion C1, criterion C2, and criterion C3, respectively. In some implementations, the weightings W1, W2, W3 may be, for example, 0.3, 0.3, 0.4 by default, where the sum of the weightings W1, W2, W3 is 1. In some implementations, the default value of the weighting may be determined according to actual needs. In some implementations, the weightings may be dynamically changed as needed.

In some implementations, the criteria evaluator 1242 may evaluate the weightings according to real-world data from a real-world database. As shown in FIG. 5, the reference evaluation unit 1242 can acquire the real environment data from the real environment database. In some implementations, the real-world data may include real-world data of the app. For example, the actual environment data may be the length of time that the user terminal 10 is involved in the application, or the total amount of profit that the user terminal 10 contributes to the application. More specifically, the benefit contributed by the user terminal 10 may be a gift or reward purchased, donated, or received by the user terminal 10, or the business value of the app. The criterion evaluation unit 1242 can adjust the weighting of each criterion when the actual environment data decreases or does not increase as expected. The criterion evaluation section 1242 may further operate the criterion calculation section 1241 to generate a formula for calculating the preference P. The formula may be as follows.

(Math. 1) Preference P = W1* C1 + W2* C2 + W3*C3

In some implementations, the preference evaluation component 124 may further include a training pairing component 1245 and a model generator 1246. The criterion evaluation unit 1242 can evaluate the criteria C1 to C3, adjust the weightings W1 to W3, operate the criterion calculation unit 1241, and update the training pairing element 1245. The training/pairing element 1245 is configured to train the reference calculation unit 1241 and the preference P updated by the feedback F, and perform pairing. The model generation unit 1246 is configured to generate a model of the preference P regarding the feedback F. In some implementations, the training and pairing element 1245 takes in the feedback F, trains the preference P updated by the reference calculation unit 1241, and pairs the preference P for the feedback F. Ring. The model generation unit 1246 acquires the preference P and the feedback F, and generates a model for the preference P and the feedback F. In some implementations, the weightings change dynamically with respect to the real-world data, for example. More specifically, the reference calculation unit 1241, the reference evaluation unit 1242, the training/pairing element 1245, and the model generation unit 1246 dynamically operate to generate a model suitable for the preference P regarding the feedback F. generate. Finally, the model generation unit 1246 outputs the trained model. According to the present invention, a more accurate preference P for the feedback F can be evaluated, and user satisfaction for the message M such as a push notification can be improved. Additionally, the click-through rate, engagement time, and business value of the app can be increased accordingly.

In some implementations, the evaluation unit 120 may be configured to identify whether there are changes in user operations and/or behaviors that lead to an increase or decrease in retention rate. In this case, the push unit 130 can direct the user terminal 10 to specific feedback by pushing a corresponding message in order to increase the retention rate or other criteria.

In some implementations, the message distribution system 100 may be action-oriented. More specifically, the message distribution system 100 can direct the user terminal 10 toward a specific action by pushing the orientation message Mo with a positive preference Po. For example, the message distribution system 100 may direct the user terminal 10 to watch a streamer's live streaming. In order to direct the user terminal 10 to watch the streamer, the push unit 130 pushes an orientation message Mo to the user terminal 10. The orientation message Mo has a positive preference Po from most users or from the same group as the user terminal 10. The push unit 130 pushes the orientation message Mo to the user terminal 10 in order to orient the user terminal 10 towards similar operations and/or actions. In some embodiments, the collection unit 110 further collects a feedback Fo for the orientation message Mo, and the evaluation unit 120 further evaluates the preference Po of the user terminal 10 for the orientation message Mo. can do.

In some implementations, the evaluation unit 120 evaluates the preference P for the message M and determines whether the push unit 130 pushes the message M to the user terminal 10. In some implementations, the push unit 130 may further push a message M' related to the message M based on the preference P. For example, the message M recommends John to watch Paul's streaming, John clicks on the streaming, and gives Paul a gift during the streaming. The next time Paul delivers again, the message M may be pushed to John. Additionally, messages M' regarding free gifts or discount information may be pushed to John in the future. In some embodiments, the push unit 130 can further recommend the recommendation message Mr to the user terminal 10 based on operations and/or actions within the application. For example, if the user terminal 10 opens the app and scrolls down the feed but ends without viewing any distribution, the push unit 130 pushes the recommendation message Mr and sends another distribution to the user terminal 10. and discount information.

In some embodiments, the preference P may be expressed as a percentage, and the default value of the preference P may be 50%. The feedback F may be classified as positive, negative or neutral feedback F. More specifically, the operations may be classified as positive, negative or neutral operations. The behavior may be classified as positive, negative or neutral behavior. When a positive, negative or neutral operation is performed, the preference P can be increased by an increment, decreased by a decrement, or kept constant. Similarly, when a positive, negative, or neutral action is performed, the preference P can be increased by an increment, decreased by a decrement, or kept constant.

If the preference P is higher than the threshold Th, the evaluation unit 120 evaluates that the user terminal 10 has the preference P, and the push unit 130 further pushes the message M to the user terminal 10 in the future. It's okay. On the other hand, if the preference P is lower than the threshold Tl, the evaluation unit 120 evaluates that the user terminal 10 does not have the preference P, and the push unit 130 pushes the message M to the user terminal 10. You may stop. Furthermore, when the preference P is within the threshold Th and the threshold Tl, the push unit 130 and the collection unit 110 continue pushing the message M and collecting the feedback F, and the evaluation unit 120 continues to push the message M and collect the feedback F, and the evaluation unit 120 continues to push the message M and collect the feedback F. It is possible to continue evaluating the preference P for the 10 messages M and determine whether the user terminal 10 has the preference P for the message M.

In some embodiments, the increment and the increment may be 3%, 5%, 10%, etc. The threshold Tl may be 20%, 30%, 40%, etc. The threshold Th may be 60%, 70%, 80%, etc. In some embodiments, the default values of the preference P, the increment, the decrement, the threshold Tl or the threshold Th can be determined according to actual needs.

FIG. 3 is a schematic diagram showing the configuration of a communication system 1' according to some embodiments of the present invention. FIG. 4 is a block diagram of a message distribution system 100' according to some embodiments of the invention. To facilitate understanding, identical reference numerals have been used wherever possible for identical elements common to these figures. As shown in FIG. 3, the communication system 1' includes a user terminal 10A, a user terminal 10B, a user terminal 10C, and a server 20. The user terminals 10A to 10C and the server 20 are connected via a network 90, for example, the Internet. The server 20 includes a message distribution system 100'. The communication system 1' may include a plurality of users, and in FIG. 3, the user terminals 10A, 10B, and 10C are shown for simplicity.

The message distribution system 100' includes a collection unit 110, an evaluation unit 120, a push unit 130, and a comparison unit 140. The collection unit 110 is configured to collect feedback F1, F2, F3 from the user terminals 10A, 10B, 10C for the message M, respectively. The evaluation unit 120 is connected to the collection unit 110 and is configured to evaluate the preference P1, P2, P3 of the user terminal 10A, 10B, 10C for the message M based on the feedback F1, F2, F3. configured.

As shown in FIG. 4, the comparison unit 140 is connected to the collection unit 110 and the evaluation unit 120, respectively, and receives the feedback F1, F2 and/or F3 from the collection unit 110 and the preference P1, P2 and/or or P3 with the feedback F1, F2 and/or F3 from the evaluation unit 120. More specifically, the comparison unit 140 compares the feedback F and the preference P collected from a plurality of users with the feedback F evaluated by the evaluation unit 120. The push unit 130 is connected to the comparison unit 140 and configured to push the message M to the user terminal 10 .

In some embodiments, the comparison unit 140 compares the feedback F1 with the preference P1 from one specific user (for example, the user terminal 10A). More specifically, the comparison unit 140 compares whether or not the preference P1 matches the feedback F1. For example, when the user terminal 10A clicks on the message M and purchases a premium membership, the evaluation unit 120 evaluates that the user terminal 10A has a preference P for the message M. Clicking on message M can be considered an affirmative action, and purchasing a premium membership can be considered an affirmative action. Therefore, the comparison unit 140 compares the feedback F and the preference P to confirm that the preference P matches the feedback F.

In some embodiments, the comparison unit 140 may be configured to detect whether the preference P1 of the user terminal 10A for the message M has changed. The comparison unit 140 may compare the real-time feedback F1 from the user terminal 10A with the historical feedback F1. If the real-time feedback F1 matches the historical feedback F1, this indicates that the preference P1 has not changed. On the other hand, if the real-time feedback F1 does not match the historical feedback F1, it can be predicted that the preference P1 for the message may have changed. For example, if the user terminal 10A has clicked on the message M in the past, this indicates that the user terminal 10A had a preference P for the message M. However, if the user terminal 10A swipes and deletes the message M in real time, this indicates that the user terminal 10A does not have preference P1 for the message M. Therefore, the comparison unit 140 compares the real-time feedback F1 of swiping and deleting the message M with the historical feedback F1 of clicking the message M, and determines whether the preference P1 for the message M has changed. Detect that.

In some implementations, the comparison unit 140 compares the feedback F1, F2, F3 for the message M from multiple users, such as the user terminals 10A, 10B, 10C. If the feedback F1 is similar to the feedback F2, the user terminal 10A and the user terminal 10B may be classified into the same group. When the feedback F1 is different from the feedback F3, the user terminal 10A and the user terminal 10C may be classified into different groups. For example, if the user terminal 10A clicks on the message M to follow a streamer and the user terminal 10B also does so, the user terminal 10A and the user terminal 10B have similar preferences for the message M. Indicates that it has. In this case, the user terminal 10A and the user terminal 10B may be classified into the same group. On the other hand, if the user terminal 10C swipes and deletes the message M, this indicates that the user terminal 10A and the user terminal 10C have different preferences for the message M, and The user terminal 10C may be classified into different groups. Here, "similar" may mean that the operations and/or actions are the same, and "different" may mean that the operations and/or actions are different.

As shown in FIG. 4, the push unit 130 can push the messages M1, M2, M3 to the user terminals 10A, 10B, 10C, respectively. The collection unit 110 can collect feedback F1, F2, F3 from the user terminals 10A, 10B, 10C for the messages M1, M2, M3, respectively. The evaluation unit 120 can evaluate the preference P1, P2, P3 of the user terminal 10A, 10B, 10C for the message M1, M2, M3 based on the feedback F1, F2, F3. If the user terminal 10A has a preference P1 for the message M1 based on the feedback F1, the message M1 may be further recommended to users in the same group, such as the user terminal 10B. In some implementations, the collection unit 110 may further collect feedback for the message M1 to the user terminal 10B, and the evaluation unit 120 may further collect feedback for the message M1 of the user terminal 10B. degree can be evaluated. Furthermore, a message M' related to the message M1 may be pushed to users in the same group, such as the user terminal 10A and the user terminal 10B. On the other hand, if the user terminal 10A and the user terminal 10C belong to different groups, the message M' related to the message M1 may not be recommended to the user terminal 10C.

In some implementations, the comparison unit 140 may be configured to compare the real-time feedbacks F1, F2, F3 for messages M from multiple users, such as the user terminals 10A, 10B, 10C, and the historical feedbacks F1, F2, F3. can be compared with. If the real-time feedback F2 is similar to the historical feedback F1, the user terminal 10B can be predicted to have a preference P2 similar to the preference P1 of the user terminal 10A. Further, if the real-time feedback F2 is similar to the historical feedback F1, the user terminal 10B may be classified into the same group as the user terminal 10A. On the other hand, if the real-time feedback F3 is different from the historical feedback F1, the preference P3 of the user terminal 10C cannot be predicted based on the preference P1 of the user terminal 10A, and the user The terminal 10C may be classified into a different group from the user terminal 10A.

In some implementations, the comparison unit 140 may compare the preferences P1, P2, P3 for the message M from multiple users, such as the user terminals 10A, 10B, 10C. Even if the feedbacks F1, F2, and F3 of the user terminals 10A, 10B, and 10C are different, the preferences P1, P2, and P3 for the message M may be the same. For example, the streamer whose message M is in the user terminal 10A and the user terminal 10B is recommended. Regarding the message M, the user terminal 10A checks the profile of the streamer, and the user terminal 10B views the streamer. Although their actions are not the same, both actions are positive. Therefore, the preferences P1 and P2 for the message M are similar. When the preference level P1 is similar to the preference level P2, the user terminal 10A and the user terminal 10B may be classified into the same group. On the other hand, when the preference level P1 is different from the preference level P3, the user terminal 10A and the user terminal 10C may be classified into different groups.

In some implementations, even though the user terminals 10A, 10B, 10C have similar feedback F1, F2, F3, the preferences P1, P2, P3 for the message M may be different. For example, the message M recommends viewing of a streamer on the user terminal 10A and the user terminal 10B. Regarding the message M, the user terminal 10A comments on the streamer with an angry emoticon, whereas the user terminal 10B comments with a smiling emoticon. Even if their actions are the same, their preferences P1 and P2 for the message M are different. Therefore, for users within the same group, both may be further classified into different subgroups. For example, if the user terminals 10A, 10B, 10C have similar preferences P1, P2, P3 for the message, the user terminals 10A, 10B, 10C may be classified into the same group. When the feedback F1 and the feedback F2 are similar with respect to the similarity preferences P1, P2, and P3, the user terminal 10A and the user terminal 10B may be further classified into the same subgroup. Users within a subgroup are much more similar than users who are not within a subgroup.

In some implementations, if the message M is related to a streamer, the push unit 130 may further push feedback F for the message M from multiple user terminals 10 to the streamer. The streamer can use the feedback F as a reference to improve his or her performance. For example, the message M may recommend the viewer to check out the streamer's songs. However, most viewers do not click on the message M. This indicates that most viewers are not interested in the streamer's songs. The push unit 130 pushes the feedback F and the preference P to the streamer, so that the streamer can understand that the viewer is not interested in his song. The streamer may perform a dance performance instead of singing next time. In some implementations, the message M may include reports and suggestions for the streamer. The report may include a monthly report of the streamer, such as the streamer's received coins, streaming time, streaming valid time, etc.

Additionally, the systems or methods described in the above embodiments may be provided in non-transitory computer readable storage devices such as solid state storage, optical disk storage, magnetic disk storage, and the like. Alternatively, the program may be downloaded from a server via the Internet.

Although the technical contents and features of the present invention have been described above, those with ordinary knowledge in the technical field to which the present invention pertains will appreciate that many variations and modifications can be made without departing from the teachings and disclosure of the present invention. It can be performed. Therefore, the scope of the invention is not limited to the embodiments already disclosed, but includes other variations and modifications that do not depart from the invention and are included in the scope of the appended claims.

1 Communication system 10 User terminal 20 Server 100 Message distribution system M Message Mu User message F Feedback Freq Frequency P Preference 110 Collection unit 120 Evaluation unit 121 Message analysis element 122 Frequency analysis element 123 Relevance analysis element 124 Preference evaluation element 1241 Standard Calculation unit 1242 Standard evaluation unit 1243 Standard module 1244 Weighting module 1245 Training/pairing element 1246 Model generation unit 130 Push unit 140 Comparison unit

Claims (18)

A message distribution system,
a collection unit configured to collect feedback regarding the message from the user terminal;
an evaluation unit connected to the collection unit and configured to evaluate the preference of the user terminal for the message based on the feedback;
The feedback includes response action feedback and/or influence action feedback, the response action feedback includes an operation on the message by the user terminal, and the influence action feedback includes an action after the operation,
the message is a push notification, the content of the message is related to live streaming,
The action is an action performed by a user on a live broadcaster by entering a live room,
A message distribution system further comprising evaluating preference for the push notification based on the action . The message distribution system according to claim 1, wherein the evaluation unit further evaluates the preference based on the operation or the action. The feedback further includes device information, the device information includes an operation time, the operation time is from when the message is read to when the operation is performed, and the evaluation unit determines the preference based on the device information. The message distribution system according to claim 1, characterized in that the message distribution system evaluates the degree of the message. 2. According to claim 1, the feedback refers to data sent by the user terminal, and the evaluation unit evaluates the preference for the message based on different types of the data. message distribution system. 5. The behavior includes at least one of idleness, search, visit, follow, purchase, send, watch, comment, support, mark, collect, and claim . Message distribution system. According to claim 4, the feedback further includes time information and/or location information, and the evaluation unit evaluates the preference based on the time information and/or location information. message distribution system. The evaluation unit includes a message analysis element, the message analysis element configured to analyze a user message left by the user terminal, and the evaluation unit evaluates the preference based on the user message. The message distribution system according to claim 4, characterized in that:. the evaluation unit includes a frequency analysis element, the frequency analysis element configured to analyze the frequency of the feedback to the message, and the evaluation unit evaluates the preference based on the frequency; 5. The message distribution system according to claim 4, characterized in that: The evaluation unit includes a relevance analysis element, the relevance analysis element configured to identify a relevance between the message and the feedback, and the evaluation unit determines the preference based on the relevance. 5. The message distribution system according to claim 4, characterized in that the message distribution system evaluates. further comprising a push unit, said push unit being connected to said evaluation unit and configured to push said message to said user terminal if said user terminal has said preference for said message; The message distribution system according to claim 1, characterized in that: Furthermore, the preference evaluation element includes a retention rate of the user terminal, a click rate of the message, a notification release rate of the message, a session length related to the app by opening the app through the message, and the app. 2. The message distribution system according to claim 1, wherein the preference is evaluated based on at least one of the frequency of opening the message. Message according to claim 1, characterized in that the collection unit and the evaluation unit dynamically collect the feedback to evaluate the preference, and the feedback includes historical feedback and real-time feedback. distribution system. Claim further comprising a comparison unit, said comparison unit being connected to said collection unit and said evaluation unit, respectively, to compare said feedback from different user terminals and classify said user terminals into groups. Message distribution system according to item 1. A message distribution method, the method comprising:
collecting feedback regarding the message from the user terminal;
evaluating the user terminal's preference for the message based on the feedback;
The feedback includes response action feedback and influence action feedback, the response action feedback includes an operation on the message by the user terminal, and the influence action feedback includes a behavior after the operation,
the message is a push notification, the content of the message is related to live streaming,
The action is an action performed by a user on a live broadcaster by entering a live room,
A message distribution method , further comprising evaluating preference for the push notification based on the action . 15. The message distribution method according to claim 14 , characterized in that it includes the step of analyzing a user message left by the user terminal and evaluating the preference level based on the user message. 15. The method of distributing a message according to claim 14 , further comprising the step of analyzing the frequency of the feedback on the message and evaluating the preference based on the frequency. 15. The message distribution method according to claim 14 , further comprising the step of identifying a relationship between the message and the feedback, and evaluating the preference based on the relationship. 15. The method of claim 14 , further comprising pushing the message to the user terminal if the user terminal has the preference for the message.