Detailed Description
Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure have been shown in the accompanying drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.
It should be understood that the various steps recited in the method embodiments of the present disclosure may be performed in a different order and/or performed in parallel. Furthermore, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present disclosure is not limited in this respect.
The term "including" and variations thereof as used herein are intended to be open-ended, i.e., including, but not limited to. The term "based on" is based at least in part on. The term "one embodiment" means "at least one embodiment," another embodiment "means" at least one additional embodiment, "and" some embodiments "means" at least some embodiments. Related definitions of other terms will be given in the description below.
It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.
It should be noted that references to "one", "a plurality" and "a plurality" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.
The names of messages or information interacted between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
Referring to fig. 1, a flow of one embodiment of an information push method according to the present disclosure is shown. The information pushing method as shown in fig. 1 comprises the following steps:
Step 101, first pushing related information of the first adjustment mode information and the first candidate information is imported into the target model to obtain a first score corresponding to the first adjustment mode information.
In this embodiment, the execution body of the information pushing method may import the first adjustment mode information and the first pushing related information of the first candidate information into the target model to obtain a first score corresponding to the first adjustment mode information.
In this embodiment, the first adjustment manner information may be used to indicate a manner of adjusting the first to-be-adjusted characteristic value of the first candidate information. Here, the number of the first adjustment mode information may be at least two.
In this embodiment, the first candidate information may be information for pushing. The type of the first candidate information is not limited herein.
In this embodiment, the first pushing related information may include information related to pushing the first candidate information. The information item specifically included in the first push related information may be set according to an actual application scenario, which is not limited herein.
In this embodiment, the target model may be used to characterize a correspondence between the first pushing related information and the first adjustment mode information, and the first score. It will be appreciated that for each adjustment mode, the target model will generate a first score for that first adjustment mode information.
In this embodiment, the target model may be a model constructed based on a neural network. The specific structure of the target model may be set according to the actual application scenario, and is not limited herein.
Step 102, selecting target first adjustment mode information from at least two first adjustment mode information according to each first score.
In this embodiment, the execution body may select the target first adjustment mode information from the at least two first adjustment mode information according to each first score.
In this embodiment, the execution body may select the first score from at least two first scores generated by the target model as the target first score. Then, the first adjustment mode information corresponding to the target first score is used as target first adjustment mode information. In other words, the first adjustment mode information may be selected as the target first adjustment mode information from the at least two first adjustment mode information based on the first score.
And 103, adjusting the first characterization value to be adjusted of the first candidate information according to the target first adjustment mode information, and generating a first adjusted characterization value.
In this embodiment, the executing body may adjust the first to-be-adjusted characteristic value of the first candidate information according to the target first adjustment mode information, to generate the first adjusted characteristic value.
As an example, the first value to be adjusted may be 10. The first adjustment information may indicate an increase of 2, and then the sum of the first to-be-adjusted characterization value 10 and 2 may be used as the first adjusted characterization value, i.e. 12.
In this embodiment, the characterization value of the first candidate information may be used as a criterion for pushing the first candidate information. As an example, when information is pushed, first candidate information related to a user feature or a user search word may be recalled according to content relevance, and then, a characterization value of the first candidate information may be referred to determine whether to display the recalled first candidate information, or determine to display ranking of the first candidate information.
Step 104, pushing the first candidate information based on the first adjusted characterization value.
In this embodiment, the executing body may push the first candidate information based on the first adjusted characterization value.
In this embodiment, the pushing of the first candidate information based on the first adjusted characterization value evaluation information may be implemented in various manners, which is not limited herein.
In some embodiments, at least one of, but not limited to, determining whether to present the first candidate information or determining a rank of the first candidate information to be presented in the information sequence to be pushed based on the first adjusted characterization value may be included.
As an example, after recalling a plurality of candidate information according to the search term or the user attribute, a predetermined number of candidate information may be selected in order of the first characterization value from the high value to the low value, and the selected candidate information may be the candidate information to be presented. And then, the candidate information to be displayed can be ranked, and the ranking basis can comprise a first adjusted characterization value.
It should be noted that, in the method provided in this embodiment, the first push related information and the first adjustment mode information are used as input to import the target model, and then the target model directly outputs the first score of the first adjustment mode information, so that the first adjustment mode information matched with the first push related information can be determined, the first characterization value to be adjusted can be adjusted for the first push related information, thereby improving the accuracy of the generated first adjusted characterization value and further improving the accuracy of push.
In some embodiments, the step 102 may include determining the first adjustment mode information corresponding to the highest first score as the target first adjustment mode information.
It should be noted that, by determining the first adjustment mode information corresponding to the highest first score as the target first adjustment mode information, the accuracy of adjustment can be improved, and further, the accuracy of the generated first adjusted characterization value is improved, thereby improving the accuracy of information pushing.
In some embodiments, the first push related information belongs to push related information. I.e. the first push related information may inherit the related situation in the push related information.
In some embodiments, specific items in the pushing related information may be set according to an actual application scenario, which is not limited herein.
In some embodiments, the push related information may include, but is not limited to, at least one of push basis information, push history information.
It should be noted that, by using the push history information as the push related information, the reply of the adjustment mode information may be determined by using the push history information as the criterion. Therefore, the adjustment amplitude of the characterization value to be adjusted can be determined by referring to the pushing history information of the candidate information, and the accuracy of adjusting the characterization value to be adjusted is improved.
In some embodiments, the push basis information may include, but is not limited to, at least one of attribute information of the candidate information, attribute information of the candidate information owner.
In some embodiments, the attribute information of the candidate information may include, but is not limited to, at least one of a characterization value of the candidate information, a formal type of the candidate information (e.g., picture, text, or video, etc.), a domain type of the candidate information (e.g., education, sports, literature, etc.).
In some embodiments, the attribute information of the candidate information owner may include, but is not limited to, at least one of a name, a type, etc. of the candidate information owner. The candidate information owner may include an owner of the candidate information.
In some embodiments, the push history information may include, but is not limited to, at least one of push process history information, historical operation information on pushed candidate information.
It should be noted that, the push related information includes push history data determined in real time, and the characterization value to be adjusted can be adjusted according to the real-time push environment, so that the instantaneity of the adjusted characterization value is improved, namely, the method can adapt to the latest push environment to adjust the characterization value, and the accuracy of adjustment is improved.
In some embodiments, the push process history information may include, but is not limited to, at least one of a start push time, a duration of the start push time to a current time, a number of pushes, a total cost since the push of the candidate information.
In some embodiments, the historical operation information for pushed candidate information may include, but is not limited to, at least one of a user representation of a user operating on the candidate information, a first number of operations performed on the candidate information, and a second number of operations performed on the candidate information.
Here, the operation content of the first operation may be set according to an actual scene, and is not limited herein, and the first operation may be, for example, an operation of triggering candidate information to browse the candidate information.
The operation content of the second operation may be set according to an actual application scenario, and is not limited herein, and the second operation may be, for example, an operation of triggering a preset control in the candidate information.
In some application scenarios, the first push related information may belong to push related information. The second push related information may also belong to push related information.
In some embodiments, the first push related information comprises at least one of first push basis information and first push history information, wherein the first push basis information comprises at least one of attribute information of first candidate information and attribute information of a first candidate information owner, and the first push history information comprises at least one of first push process history information and history operation information of first pair of pushed candidate information.
In some embodiments, the second push related information comprises at least one of second push basis information and second push history information, wherein the second push basis information comprises at least one of attribute information of second candidate information and attribute information of a second candidate information owner, and the second push history information comprises at least one of second push process history information and history operation information of second pair of pushed candidate information.
In some embodiments, at least two adjustment modes may be set according to practical situations, which is not limited herein.
As an example, the adjustment range and the adjustment amplitude, the adjustment range is-10 or more and 10 or less. The minimum adjustment unit is 0.1. Therefore, 200 adjustment amplitudes can be set, wherein the adjustment amplitudes represent that the characterization value to be adjusted is improved when the adjustment amplitudes are positive numbers, and the adjustment amplitudes represent that the characterization value to be adjusted is reduced when the adjustment amplitudes are negative numbers.
In some embodiments, the target model may be obtained through a model training step. Here, the model training step may be implemented by the flow shown in fig. 2. The flow shown in fig. 2 may include:
step 201, importing second pushing related information of the second adjustment mode information and the second candidate information into the model to be trained.
The second adjustment mode information in the model training step is different from the first adjustment mode information in the information pushing method corresponding to fig. 1. As an example, the first adjustment mode information may be two, the second adjustment mode information may be two, the first adjustment mode information is respectively increased by 1 and decreased by 1, and the second adjustment mode information may be respectively increased by 1 and decreased by 1.
The second candidate information in the model training step is the candidate information in a different stage from the first candidate information in the information pushing method corresponding to fig. 1. The information content of the first candidate information and the second candidate information may be the same. In some application scenarios, the candidate information A can be used as the second candidate information in the model training stage or used as the first candidate information after the model training is completed.
Here, the second push related information may be used to indicate information related to the push of the second candidate information.
Here, the model to be trained may be an initial model which is not trained, or may be a model which has been trained but not trained.
Here, the model to be trained may be used to characterize the correspondence between the first push related information, the first adjustment mode information, and the first score. It will be appreciated that for each adjustment mode, the model to be trained will generate a first score for the first adjustment mode information.
Here, the model to be trained may be a model constructed based on a neural network. The specific structure of the model to be trained can be set according to the actual application scene, and is not limited herein.
Step 202, selecting target second adjustment mode information from at least two second adjustment mode information according to each second score,
Here, the manner of selecting the target second adjustment manner information according to the second score may be the same as or may be different from the manner of selecting the target adjustment manner information according to the first score.
And 203, adjusting the second characterization value to be adjusted of the second candidate information by using the target second adjustment mode information, and generating a second adjusted characterization value.
Here, the second of the second to-be-adjusted characteristic values is to be consistent with the second expression in the second candidate information, and it is understood that the second candidate information does not have the first to-be-adjusted characteristic value.
And 204, adjusting parameters of the model to be trained based on the pushing effect information.
The pushing effect information is used for representing the pushing effect of pushing the second candidate information based on the second adjusted representation value.
In this embodiment, the specific representation manner of the pushing effect information may be set according to an actual application scenario, which is not limited herein.
In some application scenarios, the pushing effect information may be represented by a ratio of the first operation times to the display times. As an example, the first operation may be an operation that the user clicks on the second candidate information, or may be an operation that the user clicks on a preset control in the second candidate information.
In this embodiment, the parameters in the model to be trained may also include, but are not limited to, at least one of weights in the model to be trained and bias terms in the model to be trained.
In this embodiment, parameter adjustment of the model to be trained based on the pushing effect information may be implemented in various manners, which is not limited herein.
By way of example, back propagation, gradient descent, etc. may be used to adjust parameters of the model to be trained.
In some embodiments, the stopping condition of the model training may be set according to the actual application scenario, which is not limited herein. As an example, the stopping condition of model training may include, but is not limited to, at least one of a model parameter update magnitude being less than a preset update magnitude threshold, a number of model updates being greater than a preset update number threshold.
In this embodiment, the model to be trained after the parameter adjustment may be used as the target model, or the model to be trained after the parameter adjustment may be trained to obtain the target model.
It should be noted that, according to the model training method provided by the embodiment, the pushing related information of the to-be-trained model is imported and can be used for representing the pushing environment of the candidate information, the scoring of the adjustment mode information of the pushing environment is based on the actual pushing effect information, and the parameters of the to-be-trained model are adjusted, so that the scoring is changed towards the direction of improving the pushing effect evaluation, and therefore the correlation between the representation value and the pushing related information or the correlation between the representation value and the pushing effect can be avoided from being calculated from the content, the adjustment mode information is judged directly from the angle of the pushing effect information, and the complexity of determining the more accurate adjustment mode information is reduced.
In some embodiments, the step 202 may include selecting the target second score from the second scores by using a random selection manner that sets the probability of selection, and determining the second adjustment manner information corresponding to the selected target second score as the target second adjustment manner information.
In some embodiments, a selected probability may be set for each score. The manner in which the selection probability is set may be various.
In some application scenarios, the specific value of the first preset probability may be set according to an actual application scenario, which is not limited herein. As an example, the first preset probability may be set to 90%. The first preset probability is less than 1.
Here, the probability of determining the highest second score as the target second score is the first preset probability. It will be appreciated that in general, the number of highest scores is one.
It should be noted that, by adopting a random calculation method for setting the selection probability, the exploratory of the information of the various adjustment methods to be selected can be increased. Avoiding the model from being trapped in local optimum by selecting the highest second score each time.
In some application scenarios, the selected probabilities of the respective non-highest second replies may or may not be equal.
As an example, the selected probability corresponding to the highest scoring may be set as the first preset probability. And determining the ratio of the difference value of 1 and the first preset probability to the number of scoring of the non-highest scoring as a second preset probability. The second preset probability may be the selected probability of each non-highest scoring. In this case, the probabilities of the respective non-highest second scores being chosen are equal.
In other words, if the product of the number of highest scores and the first preset probability is taken as the first product. The probability of determining other second replies than the highest scoring as target second scoring is a second preset probability. The product of the number of replies other than the highest scoring and the second preset probability is taken as the second product. The sum of the first product and the second product is equal to 1.
As an example, if there are 10 replies, 10 different choice probabilities can be set, the sum of these 10 choice probabilities being 1. In this case, the probabilities of the respective non-highest second scores being chosen are not equal.
In some embodiments, the model training step includes increasing the first preset probability based on an increase in the number of model training updates.
Here, the above-mentioned first preset probability, i.e. the selected probability of the highest second answer, may be increased as the model training proceeds. In this way, the convergence rate of the model can be increased, i.e. the accuracy of the model is generally increasing as model training proceeds, i.e. the accuracy of the second scoring is also increasing, in which case the search for other adjustment modes can be reduced and the model parameters are improved based on the highest second scoring in a concentrated way. Model training is carried out by using real data, and the model to be trained is continuously applied to the scene of the real scene, so that the pushing effect of the real scene can be improved.
In some embodiments, the push effect information may include a push effect characterization value. The step 204 may include determining a loss function value according to the change of the pushing effect information of the second candidate information, and adjusting parameters of the model to be trained according to the loss function value.
In some application scenarios, the push effect information may include a push effect characterization value. As an example, the pushing effect characterization value may be a ratio of the number of presentations to the number of clicks, where the more clicks, the smaller the pushing effect characterization value, indicating that the better the pushing effect.
As an example, the square of the push effect characterization value may be taken as the loss function value. Or the product of the square of the push effect characterization value and a preset attenuation coefficient can be used as the loss function value.
As an example, the present push effect token value may be compared with the last push effect token value, and if the push effect token value increases, a loss function may be generated according to the highest score.
If the representation value of the pushing effect is increased, the target adjustment mode is adopted to reduce the pushing effect, so that scoring corresponding to the information of the target adjustment mode can be reduced. Thus, a score corresponding to the target adjustment method can be determined as the loss function value. Based on the model parameters to be trained adjusted by the loss function value, the scoring corresponding to the target adjustment mode information can be gradually reduced.
In some embodiments, the step 204 includes adjusting parameters of the model to be trained according to pushing effect information corresponding to a second candidate information set, where the second candidate information set includes at least two second candidate information to be pushed.
It should be noted that, the training model is trained by adopting the related data of the plurality of second candidate information, so that the training model learns the related features of the second candidate information set space, when the single second candidate information is processed by using the training model, the processing mode can refer to the whole situation of the second candidate information set space, and the single second candidate information is processed, namely, a certain second candidate information is processed by referring to the whole situation of the second candidate information, so that the processed characterization value of the second candidate information is matched with the whole situation, namely, the adjustment accuracy is improved.
Referring to fig. 3, the present disclosure provides a model training method that may include the flow shown in fig. 3. The flow shown in fig. 3 may include step 301, step 302, step 303, and step 304.
Step 301, importing second adjustment mode information and second pushing related information of second candidate information into the model to be trained.
The second adjustment mode information in the model training method is the adjustment mode information in a different stage from the first adjustment mode information in the information pushing method corresponding to fig. 1. As an example, the first adjustment mode information may be two, the second adjustment mode information may be two, the first adjustment mode information is respectively increased by 1 and decreased by 1, and the second adjustment mode information may be respectively increased by 1 and decreased by 1.
The second candidate information in the model training method is the candidate information in a different stage from the first candidate information in the information pushing method corresponding to fig. 1. The information content of the first candidate information and the second candidate information may be the same. In some application scenarios, the candidate information A can be used as the second candidate information in the model training stage or used as the first candidate information after the model training is completed.
Here, the second push related information may be used to indicate information related to the push of the second candidate information.
Here, the model to be trained may be an initial model which is not trained, or may be a model which has been trained but not trained.
Here, the model to be trained may be used to characterize the correspondence between the first push related information, the first adjustment mode information, and the first score. It will be appreciated that for each adjustment mode, the model to be trained will generate a first score for the first adjustment mode information.
Here, the model to be trained may be a model constructed based on a neural network. The specific structure of the model to be trained can be set according to the actual application scene, and is not limited herein.
Step 302, selecting target second adjustment mode information from at least two second adjustment mode information according to each second score,
Here, the manner of selecting the target second adjustment manner information according to the second score may be the same as or may be different from the manner of selecting the target adjustment manner information according to the first score.
And 303, adjusting the second characterization value to be adjusted of the second candidate information by using the target second adjustment mode information, and generating a second adjusted characterization value.
Here, the second of the second to-be-adjusted characteristic values is to be consistent with the second expression in the second candidate information, and it is understood that the second candidate information does not have the first to-be-adjusted characteristic value.
And step 304, adjusting parameters of the model to be trained based on the pushing effect information.
The pushing effect information is used for representing the pushing effect of pushing the second candidate information based on the second adjusted representation value.
In this embodiment, the specific representation manner of the pushing effect information may be set according to an actual application scenario, which is not limited herein.
In some application scenarios, the pushing effect information may be represented by a ratio of the first operation times to the display times. As an example, the first operation may be an operation that the user clicks on the second candidate information, or may be an operation that the user clicks on a preset control in the second candidate information.
In this embodiment, the parameters in the model to be trained may also include, but are not limited to, at least one of weights in the model to be trained and bias terms in the model to be trained.
In this embodiment, parameter adjustment of the model to be trained based on the pushing effect information may be implemented in various manners, which is not limited herein.
By way of example, back propagation, gradient descent, etc. may be used to adjust parameters of the model to be trained.
In some embodiments, the stopping condition of the model training may be set according to the actual application scenario, which is not limited herein. As an example, the stopping condition of model training may include, but is not limited to, at least one of a model parameter update magnitude being less than a preset update magnitude threshold, a number of model updates being greater than a preset update number threshold.
In this embodiment, the model to be trained after the parameter adjustment may be used as the target model, or the model to be trained after the parameter adjustment may be trained to obtain the target model.
It should be noted that, according to the model training method provided by the embodiment, the pushing related information of the to-be-trained model is imported and can be used for representing the pushing environment of the candidate information, the scoring of the adjustment mode information of the pushing environment is based on the actual pushing effect information, and the parameters of the to-be-trained model are adjusted, so that the scoring is changed towards the direction of improving the pushing effect evaluation, and therefore the correlation between the representation value and the pushing related information or the correlation between the representation value and the pushing effect can be avoided from being calculated from the content, the adjustment mode information is judged directly from the angle of the pushing effect information, and the complexity of determining the more accurate adjustment mode information is reduced.
In some embodiments, the step 302 may include selecting the target second score from the second scores by using a random selection manner that sets the probability of selection, and determining the second adjustment manner information corresponding to the selected target second score as the target second adjustment manner information.
In some embodiments, a selected probability may be set for each score. The manner in which the selection probability is set may be various.
In some application scenarios, the specific value of the first preset probability may be set according to an actual application scenario, which is not limited herein. As an example, the first preset probability may be set to 90%. The first preset probability is less than 1.
Here, the probability of determining the highest second score as the target second score is the first preset probability. It will be appreciated that in general, the number of highest scores is one.
It should be noted that, by adopting a random calculation method for setting the selection probability, the exploratory of the information of the various adjustment methods to be selected can be increased. Avoiding the model from being trapped in local optimum by selecting the highest second score each time.
In some application scenarios, the selected probabilities of the respective non-highest second replies may or may not be equal.
As an example, the selected probability corresponding to the highest scoring may be set as the first preset probability. And determining the ratio of the difference value of 1 and the first preset probability to the number of scoring of the non-highest scoring as a second preset probability. The second preset probability may be the selected probability of each non-highest scoring. In this case, the probabilities of the respective non-highest second scores being chosen are equal.
In other words, if the product of the number of highest scores and the first preset probability is taken as the first product. The probability of determining other second replies than the highest scoring as target second scoring is a second preset probability. The product of the number of replies other than the highest scoring and the second preset probability is taken as the second product. The sum of the first product and the second product is equal to 1.
As an example, if there are 10 replies, 10 different choice probabilities can be set, the sum of these 10 choice probabilities being 1. In this case, the probabilities of the respective non-highest second scores being chosen are not equal.
In some embodiments, the model training method includes increasing the first preset probability based on an increase in a number of model training updates.
Here, the above-mentioned first preset probability, i.e. the selected probability of the highest second answer, may be increased as the model training proceeds. In this way, the convergence rate of the model can be increased, i.e. the accuracy of the model is generally increasing as model training proceeds, i.e. the accuracy of the second scoring is also increasing, in which case the search for other adjustment modes can be reduced and the model parameters are improved based on the highest second scoring in a concentrated way. Model training is carried out by using real data, and the model to be trained is continuously applied to the scene of the real scene, so that the pushing effect of the real scene can be improved.
In some embodiments, the push effect information may include a push effect characterization value. The step 304 may include determining a loss function value according to the change of the pushing effect information of the second candidate information, and adjusting parameters of the model to be trained according to the loss function value.
In some application scenarios, the push effect information may include a push effect characterization value. As an example, the pushing effect characterization value may be a ratio of the number of presentations to the number of clicks, where the more clicks, the smaller the pushing effect characterization value, indicating that the better the pushing effect.
As an example, the square of the push effect characterization value may be taken as the loss function value. Or the product of the square of the push effect characterization value and a preset attenuation coefficient can be used as the loss function value.
As an example, the present push effect token value may be compared with the last push effect token value, and if the push effect token value increases, a loss function may be generated according to the highest score.
If the representation value of the pushing effect is increased, the target adjustment mode is adopted to reduce the pushing effect, so that scoring corresponding to the information of the target adjustment mode can be reduced. Thus, a score corresponding to the target adjustment method can be determined as the loss function value. Based on the model parameters to be trained adjusted by the loss function value, the scoring corresponding to the target adjustment mode information can be gradually reduced.
In some embodiments, the step 304 includes adjusting parameters of the model to be trained according to pushing effect information corresponding to a second candidate information set, where the second candidate information set includes at least two second candidate information to be pushed.
It should be noted that, the training model is trained by adopting the related data of the plurality of second candidate information, so that the training model learns the related features of the second candidate information set space, when the single second candidate information is processed by using the training model, the processing mode can refer to the whole situation of the second candidate information set space, and the single second candidate information is processed, namely, a certain second candidate information is processed by referring to the whole situation of the second candidate information, so that the processed characterization value of the second candidate information is matched with the whole situation, namely, the adjustment accuracy is improved.
With further reference to fig. 4, as an implementation of the method shown in the foregoing figures, the present disclosure provides an embodiment of an information pushing apparatus, where an embodiment of the apparatus corresponds to the embodiment of the method shown in fig. 1, and the apparatus may be specifically applied to various electronic devices.
As shown in fig. 4, the information pushing apparatus of the present embodiment includes an import unit 401, a selection unit 402, a generation unit 403, and a pushing unit 404. The device comprises an input unit, a selection unit, a generation unit and a pushing unit, wherein the input unit is used for inputting first adjustment mode information and first pushing related information of first candidate information into a target model to obtain first scores corresponding to the first adjustment mode information, the selection unit is used for selecting target first adjustment mode information from at least two pieces of first adjustment mode information according to the first scores, the generation unit is used for utilizing the target first adjustment mode information to adjust a first characterization value to be adjusted of the first candidate information to generate a first characterization value after adjustment, and the pushing unit is used for pushing the first candidate information based on the first characterization value after adjustment.
In this embodiment, the specific processes of the importing unit 401, the selecting unit 402, the generating unit 403, and the pushing unit 404 of the information pushing device and the technical effects thereof may refer to the relevant descriptions of the steps 101, 102, 103, and 104 in the corresponding embodiment of fig. 1, and are not repeated herein.
In some embodiments, selecting the target first adjustment mode information from the at least two first adjustment mode information according to each first score includes determining the first adjustment mode information corresponding to the highest first score as the target first adjustment mode information.
In some embodiments, the target model is obtained through a model training step, wherein the model training step comprises the steps of importing second pushing related information of second adjustment mode information and second candidate information into a model to be trained to obtain second scores corresponding to the second adjustment mode information, the number of the second adjustment mode information is at least two, selecting target second adjustment mode information from the at least two second adjustment mode information according to the second scores, adjusting a second characterization value to be adjusted of the second candidate information by utilizing the target second adjustment mode information to generate a second adjusted characterization value, and adjusting parameters of the model to be trained based on pushing effect information, wherein the pushing effect information is used for characterizing a pushing effect of pushing the second candidate information based on the second adjusted characterization value.
In some embodiments, the second push related information includes at least one of second push basis information and second push history information, wherein
The second pushing basis information comprises at least one of attribute information of second candidate information and attribute information of a second candidate information owner;
The second pushing history information comprises at least one of second pushing process history information and second historical operation information of pushing candidate information. In some embodiments, selecting the target second adjustment mode information from the at least two second adjustment mode information according to each second score includes selecting the target second score from the second scores by adopting a random selection mode for setting a selection probability, wherein the selection probability of the highest second score is the first preset probability, and determining the second adjustment mode information corresponding to the selected target second score as the target second adjustment mode information.
In some embodiments, the model training step includes increasing the first preset probability based on an increase in the number of model training updates.
In some embodiments, the adjusting the parameters of the model to be trained based on the push effect information includes determining a loss function value based on a change in push effect information of the second candidate information, and adjusting the parameters of the model to be trained based on the loss function value.
In some embodiments, the adjusting the parameters of the model to be trained based on the push effect information includes adjusting the parameters of the model to be trained according to push effect information corresponding to a second candidate information set, where the second candidate information set includes at least two second candidate information to be pushed.
With further reference to fig. 5, as an implementation of the method shown in the foregoing figures, the present disclosure provides an embodiment of an information pushing apparatus, where an embodiment of the apparatus corresponds to the embodiment of the method shown in fig. 1, and the apparatus may be specifically applied to various electronic devices.
As shown in fig. 5, the information pushing device of the present embodiment includes an importing module 501, a selecting module 502, a generating module 503, and an adjusting module 504. The training device comprises an input unit, a selection module, a generation module and an adjustment module, wherein the input unit is used for inputting first adjustment mode information and first pushing related input modules of first candidate information, the first pushing related input modules are used for inputting second adjustment mode information and second pushing related information of the second candidate information into a model to be trained to obtain second scores corresponding to the second adjustment mode information, the number of the second adjustment mode information is at least two, the selection module is used for selecting target second adjustment mode information from the second adjustment mode information according to the second scores, the generation module is used for adjusting a second to-be-adjusted characteristic value of the second candidate information by utilizing the target second adjustment mode information to generate a second adjusted characteristic value, and the adjustment module is used for adjusting parameters of the model to be trained based on pushing effect information, wherein the pushing effect information is used for representing pushing effect of pushing the second candidate information based on the second to-be-adjusted characteristic value.
In this embodiment, the specific processing of the importing module 501, the selecting module 502, the generating module 503, and the adjusting module 504 of the information pushing device and the technical effects thereof may refer to the related descriptions of step 301, step 302, step 303, and step 304 in the corresponding embodiment of fig. 3, and are not described herein.
In some embodiments, the second push related information comprises at least one of second push basis information and second push history information, wherein the second push basis information comprises at least one of attribute information of second candidate information and attribute information of a second candidate information owner, and the second push history information comprises at least one of second push process history information and history operation information of second pair of pushed candidate information.
In some embodiments, selecting the target second adjustment mode information from the at least two second adjustment mode information according to each second score includes selecting the target second score from the second scores by adopting a random selection mode for setting a selection probability, wherein the selection probability of the highest second score is the first preset probability, and determining the second adjustment mode information corresponding to the selected target second score as the target second adjustment mode information.
In some embodiments, the model training step includes increasing the first preset probability based on an increase in the number of model training updates.
Referring to fig. 6, fig. 6 illustrates an exemplary system architecture in which the information push method of one embodiment of the present disclosure may be applied.
As shown in fig. 6, the system architecture may include terminal devices 601, 602, 603, a network 604, and a server 605. The network 604 is used as a medium to provide communication links between the terminal devices 601, 602, 603 and the server 605. The network 606 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
The terminal devices 601, 602, 603 may interact with the server 605 via the network 604 to receive or send messages or the like. Various client applications, such as a web browser application, a search class application, a news information class application, may be installed on the terminal devices 601, 602, 603. The client application in the terminal device 601, 602, 603 may receive the instruction of the user and perform the corresponding function according to the instruction of the user, for example, adding the corresponding information in the information according to the instruction of the user.
The terminal devices 601, 602, 603 may be hardware or software. When the terminal devices 601, 602, 603 are hardware, they may be various electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablet computers, electronic book readers, MP3 players (Moving Picture Experts Group Audio Layer III, dynamic video expert compression standard audio plane 3), MP4 (Moving Picture Experts Group Audio Layer IV, dynamic video expert compression standard audio plane 4) players, laptop and desktop computers, and the like. When the terminal devices 601, 602, 603 are software, they can be installed in the above-listed electronic devices. Which may be implemented as multiple software or software modules (e.g., software or software modules for providing distributed services) or as a single software or software module. The present invention is not particularly limited herein.
The server 605 may be a server that provides various services, for example, receives information acquisition requests sent by the terminal devices 601, 602, 603, and acquires presentation information corresponding to the information acquisition requests in various ways according to the information acquisition requests. And related data showing the information is transmitted to the terminal devices 601, 602, 603.
It should be noted that, the information pushing method provided by the embodiments of the present disclosure may be performed by the terminal device, and accordingly, the information pushing apparatus may be set in the terminal devices 601, 602, 603. In addition, the information pushing method provided by the embodiment of the present disclosure may also be performed by the server 605, and accordingly, the information pushing device may be disposed in the server 605.
It should be understood that the number of terminal devices, networks and servers in fig. 6 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Referring now to fig. 7, a schematic diagram of an electronic device (e.g., a terminal device or server in fig. 6) suitable for use in implementing embodiments of the present disclosure is shown. The terminal devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and the like, and stationary terminals such as digital TVs, desktop computers, and the like. The electronic device shown in fig. 7 is merely an example and should not be construed to limit the functionality and scope of use of the disclosed embodiments.
As shown in fig. 7, the electronic device may include a processing means (e.g., a central processing unit, a graphics processor, etc.) 701, which may perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 702 or a program loaded from a storage means 708 into a Random Access Memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the electronic device 700 are also stored. The processing device 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. An input/output (I/O) interface 705 is also connected to bus 704.
In general, devices including input devices 707 such as a touch screen, touch pad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc., output devices 707 such as a Liquid Crystal Display (LCD), speaker, vibrator, etc., storage devices 708 such as magnetic tape, hard disk, etc., and communication devices 709 may be connected to the I/O interface 705. The communication means 709 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. While fig. 7 shows an electronic device having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead.
In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a non-transitory computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via communication device 709, or installed from storage 708, or installed from ROM 702. The above-described functions defined in the methods of the embodiments of the present disclosure are performed when the computer program is executed by the processing device 701.
It should be noted that the computer readable medium described in the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of a computer-readable storage medium may include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present disclosure, however, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to electrical wiring, fiber optic cable, RF (radio frequency), and the like, or any suitable combination of the foregoing.
In some embodiments, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperTextTransfer Protocol ), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.
The computer readable medium may be included in the electronic device or may exist alone without being incorporated into the electronic device.
The computer readable medium carries one or more programs, when the one or more programs are executed by the electronic device, the electronic device is caused to import first adjustment mode information and first pushing related information of first candidate information into a target model to obtain first scores corresponding to the first adjustment mode information, the number of the first adjustment mode information is at least two, target first adjustment mode information is selected from the at least two first adjustment mode information according to the first scores, the first to-be-adjusted characteristic value of the first candidate information is adjusted by the target first adjustment mode information to generate a first adjusted characteristic value, and the first candidate information is pushed based on the first adjusted characteristic value.
Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, including, but not limited to, an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units involved in the embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. The name of the unit does not in some cases define the unit itself, and for example, the import unit may be described as an "import target model unit".
The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic that may be used include Field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems-on-a-chip (SOCs), complex Programmable Logic Devices (CPLDs), and the like.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by persons skilled in the art that the scope of the disclosure referred to in this disclosure is not limited to the specific combinations of features described above, but also covers other embodiments which may be formed by any combination of features described above or equivalents thereof without departing from the spirit of the disclosure. Such as those described above, are mutually substituted with the technical features having similar functions disclosed in the present disclosure (but not limited thereto).
Moreover, although operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the present disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.
Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are example forms of implementing the claims.