Information processing method and apparatus

Abstract

The present disclosure relates to an information processing method and device, the information processing method comprising: obtaining a participation time sequence of an extraction object of a user account according to historical behavior information of the extraction object of the user account; determining an abnormal extraction time in the participation time sequence, obtaining an object corresponding to the abnormal extraction time, and taking the object corresponding to the abnormal extraction time as a target extraction object of the user account; and obtaining a predicted extraction number of the user account after the target extraction object is extracted according to object extraction information of the user account in a predetermined historical period, wherein the object extraction information comprises object information extracted by the user account in the predetermined historical period. The present disclosure can accurately and efficiently find the individual differences of the target extraction object of the user account, quantify the tolerance of the user account, and facilitate subsequent data information analysis and processing of the obtained target extraction object and predicted extraction number.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of data analysis, and in particular, to an information processing method and device. BACKGROUND

[0002] For the sampling method, a sampling probability can be allocated to all samples, and then the expected target sample can be sampled. However, the sampling process is a time sequence composed of random events, and cannot guarantee the sampling time of the target sample and whether the target sample can be sampled. For example, for the case of sampling the expected target object from the user accounts on the content platform, a sampling probability can be allocated to all objects that can be sampled, and then the object is sampled. In the related art, the user accounts can be classified, and a target object expected to be sampled is determined for each category. For any user account, the object sampled in the preset sampling round can be the target object. However, this method cannot determine the target object expected to be sampled for each user account, and cannot specifically formulate the target object sampling strategy for the user account.

[0003] For example, for the user accounts on the content platform, a video is randomly sampled from a certain number of videos for playing, and each video has a probability of being sampled. For different user accounts, each user account can have a video expected to be sampled and played. In the related art, the user accounts can be classified based on historical data, such as a user account whose proportion of playing a “funny video” in total playing videos exceeds a classification threshold is directly classified as expecting to sample a certain “funny video”, and a user account whose proportion is lower than the classification threshold is directly classified as expecting to sample a certain “dance video”. For the sampling process, the classified video can be sampled in a preset number of sampling times. However, this method cannot map to the expectation of each user account, and cannot accurately predict the number of times the user account will continue to sample videos after sampling the expected video. SUMMARY

[0004] The present disclosure provides an information processing method and device to at least solve the problems in the related art described above, and can not solve any of the above problems. The technical solution of the present disclosure is as follows:

[0005] According to a first aspect of the embodiments of the present disclosure, an information processing method is provided, including: obtaining a participation time sequence of an extraction object of a user account according to historical behavior information of the extraction object of the user account; determining an abnormal extraction time in the participation time sequence, obtaining an object corresponding to the abnormal extraction time, and taking the object corresponding to the abnormal extraction time as a target extraction object of the user account; obtaining a predicted extraction number of the user account after extracting the target extraction object according to object extraction information of the user account in a predetermined historical period, wherein the object extraction information includes object information extracted by the user account in the predetermined historical period.

[0006] Optionally, the obtaining of the participation time sequence of the extraction object of the user account according to the historical behavior information of the extraction object of the user account includes: obtaining feature information of each object extracted by the user account according to the historical behavior information; constructing derived feature information of each object extracted by the user account according to the feature information; and obtaining a participation degree of each object extracted by the user account according to the derived feature information, and arranging the participation degrees into a participation time sequence in the order of extraction times of the objects.

[0007] Optionally, the obtaining of the participation degree of each object extracted by the user account according to the derived feature information includes: obtaining a feature value of each derived feature information when each object is extracted by the user account according to the derived feature information; and calculating the participation degree of each object extracted by the user account according to the feature value and a first preset weight of each derived feature information.

[0008] Optionally, the determining of the abnormal extraction time in the participation time sequence, the obtaining of the object corresponding to the abnormal extraction time, and the taking of the object corresponding to the abnormal extraction time as the target extraction object of the user account include: performing abnormal detection on the participation time sequence by at least one abnormal detection algorithm to obtain an abnormal time detected by each abnormal detection algorithm; determining the abnormal extraction time from the abnormal time detected by each abnormal detection algorithm, obtaining the object corresponding to the abnormal extraction time, and taking the object corresponding to the abnormal extraction time as the target extraction object of the user account.

[0009] Optionally, the determining of the abnormal extraction time from the abnormal time detected by each abnormal detection algorithm includes: determining a to-be-accepted abnormal time from the abnormal time detected by each abnormal detection algorithm based on a second preset weight of each abnormal detection algorithm; and determining whether to take the to-be-accepted abnormal time as the abnormal extraction time according to a confidence degree of an object extracted by the to-be-accepted abnormal time.

[0010] Optionally, the determining, according to the confidence of the object extracted at the to-be-inspected abnormal time, whether to take the to-be-inspected abnormal time as the extraction time of the abnormality, comprises: in a case where the confidence of the object extracted at the to-be-inspected abnormal time is greater than or equal to a preset threshold, taking the to-be-inspected abnormal time as the extraction time of the abnormality; in a case where the confidence of the object extracted at the to-be-inspected abnormal time is less than the preset threshold, adjusting the second preset weight and re-determining the to-be-inspected abnormal time until the confidence of the object extracted at the to-be-inspected abnormal time is greater than or equal to the preset threshold, wherein the confidence represents a confidence of taking the object extracted at the to-be-inspected abnormal time as the target extraction object of the user account.

[0011] Optionally, after the extraction time of the abnormality is determined in the participation time sequence, the object corresponding to the extraction time of the abnormality is obtained, and the object corresponding to the extraction time of the abnormality is taken as the target extraction object of the user account, the method further comprises: in a case where the object extraction information of the user account is not detected, constructing a similarity model of the user account; determining a similar user account of the user account according to the similarity model, and taking a predicted extraction frequency corresponding to the similar user account as a predicted extraction frequency of the user account after the target extraction object is extracted.

[0012] According to a second aspect of the embodiments of the present disclosure, an information processing apparatus is provided, comprising: a sequence obtaining unit configured to obtain a participation time sequence of an extraction object of a user account according to historical behavior information of the extraction object of the user account; an object determining unit configured to determine an extraction time of an abnormality in the participation time sequence, obtain an object corresponding to the extraction time of the abnormality, and take the object corresponding to the extraction time of the abnormality as a target extraction object of the user account; and a frequency predicting unit configured to obtain a predicted extraction frequency of the user account after the target extraction object is extracted according to object extraction information of the user account in a predetermined historical period, wherein the object extraction information comprises object information extracted by the user account in the predetermined historical period.

[0013] Optionally, the sequence obtaining unit is configured to obtain feature information of each object extracted by the user account according to the historical behavior information, construct derived feature information of each object extracted by the user account according to the feature information, and obtain a participation degree of each object extracted by the user account according to the derived feature information, and arrange the participation degrees into a participation time sequence in an order of extraction times of the each object.

[0014] Optionally, the sequence obtaining unit is configured to: obtain, according to the derived feature information, a feature value of each derived feature information when the user account extracts each object; and calculate, according to the feature value and a first preset weight of each derived feature information, a participation degree of the user account when extracting each object.

[0015] Optionally, the object determining unit is configured to: perform anomaly detection on the participation degree time sequence by using at least one anomaly detection algorithm to obtain an anomaly time detected by each anomaly detection algorithm; determine, from the anomaly time detected by each anomaly detection algorithm, an extraction time of the anomaly, obtain an object corresponding to the extraction time of the anomaly, and take the object corresponding to the extraction time of the anomaly as a target extraction object of the user account.

[0016] Optionally, the object determining unit is configured to: determine, based on a second preset weight of each anomaly detection algorithm, a to-be-inspected anomaly time from the anomaly time detected by each anomaly detection algorithm; and determine, according to a confidence degree of an object extracted by the to-be-inspected anomaly time, whether to take the to-be-inspected anomaly time as the extraction time of the anomaly.

[0017] Optionally, the object determining unit is configured to: in a case where the confidence degree of the object extracted by the to-be-inspected anomaly time is greater than or equal to a preset threshold, take the to-be-inspected anomaly time as the extraction time of the anomaly; and in a case where the confidence degree of the object extracted by the to-be-inspected anomaly time is less than the preset threshold, adjust the second preset weight and redetermine a to-be-inspected anomaly time until the confidence degree of the object extracted by the to-be-inspected anomaly time is greater than or equal to a preset threshold, wherein the confidence degree represents a confidence degree of taking the object extracted by the to-be-inspected anomaly time as the target extraction object of the user account.

[0018] Optionally, the method further includes a similarity constructing unit configured to: in a case where no object extraction information of the user account is detected, construct a similarity model of the user account; determine, according to the similarity model, a similar user account of the user account, and take a predicted extraction frequency corresponding to the similar user account as a predicted extraction frequency of the user account after extracting the target extraction object.

[0019] According to a third aspect of the embodiments of the present disclosure, an electronic device is provided, including: at least one processor; at least one memory storing computer executable instructions, wherein the computer executable instructions, when executed by the at least one processor, cause the at least one processor to perform the information processing method according to the present disclosure.

[0020] According to a fourth aspect of the embodiments of the present disclosure, a computer readable storage medium is provided, which stores instructions that, when executed by at least one processor, cause the at least one processor to perform the information processing method according to the present disclosure.

[0021] According to a fifth aspect of the embodiments of the present disclosure, a computer program product is provided, which comprises computer instructions that, when executed by at least one processor, implement the information processing method according to the present disclosure.

[0022] The embodiments of the present disclosure provide at least the following beneficial effects:

[0023] According to the information processing method and device, the target extraction object of the user account can be obtained through data analysis, and the tolerance (predicted extraction times) of the user account after extracting the target extraction object can be obtained in combination with the historical object extraction information, so that the target object expected to be extracted by each user account can be determined, and the object extraction strategy of the user account can be flexibly adjusted. The present disclosure can accurately and efficiently find the individual differences of the target extraction object of the user account, and quantize the tolerance of the user account. The whole process is real-time and self-adaptive, and is convenient for subsequent data information analysis and processing of the obtained target extraction object and predicted extraction times.

[0024] In addition, according to the information processing method and device, in the case of using multiple data analysis algorithms, the algorithms are not directly spliced and used, but are flexibly combined to give full play to the advantages of each algorithm to obtain the target extraction object and the predicted extraction times.

[0025] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0026] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present disclosure and, together with the specification, serve to explain the principles of the present disclosure, and do not constitute an undue limitation on the present disclosure.

[0027] Figure 1 is a flowchart of an information processing method according to an exemplary embodiment.

[0028] Figure 2 is a flowchart of another information processing method according to an exemplary embodiment.

[0029] Figure 3 is a block diagram of an information processing device according to an exemplary embodiment.

[0030] Figure 4is a block diagram of an electronic device 400 according to an exemplary embodiment. DETAILED DESCRIPTION

[0031] In order for the ordinary person in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings.

[0032] It should be noted that the terms "first", "second", and the like in the specification and claims of the present disclosure and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than that illustrated or described herein. The embodiments described in the following embodiments do not represent all embodiments consistent with the present disclosure. Rather, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0033] It should be noted herein that "at least one of a plurality of items" appearing in the present disclosure means that the three types of alternatives are included, i.e., "any one of the plurality of items", "a combination of any two or more of the plurality of items", and "all of the plurality of items". For example, "including at least one of A and B" includes the following three alternatives: (1) including A; (2) including B; and (3) including A and B. For another example, "performing at least one of step one and step two" means the following three alternatives: (1) performing step one; (2) performing step two; and (3) performing step one and step two.

[0034] For the sampling method, a sampling probability can be assigned to all samples, and then the expected target sample can be sampled. However, such a sampling process is a time sequence composed of random events, and cannot guarantee the sampling time of the target sample and whether the target sample can be sampled. For example, for the case of sampling the expected target object of the user account on the content platform, a sampling probability can be assigned to all objects that can be sampled, and then the object is sampled. The object sampling process is also a time sequence composed of random events. Each user account can have an expected target object to be sampled, and the expected target object to be sampled can change continuously with the sampling process. For some user accounts, if the target object cannot be sampled within a certain number of times, it can lead to a decrease in the activity of these user accounts on the content platform. For some user accounts, if the target object is sampled multiple times within a certain number of times, it can cause these user accounts to adjust the expected target object to be sampled. Therefore, for the convenience of subsequent information processing and strategy formulation, the determination of the expected target object to be sampled and the prediction of the number of possible subsequent samplings of the user account after sampling the target object are very important.

[0035] In the related art, user accounts can be classified, and a target object expected to be extracted is determined for each category. For any user account, the object extracted in a preset extraction round can be set as the target object. However, this method cannot determine the target object expected to be extracted for each user account, and cannot accurately predict the number of times the user account continues to extract after extracting the target object, so that the target object extraction strategy of the user account cannot be formulated in a targeted manner.

[0036] For example, for a user account on a content platform, a video is randomly extracted from a certain number of videos for playing. Each video has a probability of being extracted. This process can be regarded as extraction of an object. For different user accounts, each user account can have a video expected to be extracted for playing. In the related art, user accounts can be classified based on historical data. For example, a user account in which the proportion of playing a "funny video" in total playing videos exceeds a classification threshold is directly classified as expecting to extract a certain "funny video", and a user account in which the proportion is lower than the classification threshold is directly classified as expecting to extract a certain "dance video". However, for the extraction process, the preset extraction times can be set to extract the classified video. However, this method cannot map to the expectation of each user account, and cannot accurately predict the number of times the user account continues to extract videos after extracting the expected video, so that the preset extraction times cannot be set in a targeted manner to enable the user account to extract the video expected to be played in time.

[0037] To solve the problems in the related art described above, the present disclosure provides an information processing method and device. The target extraction object of a user account can be obtained through data analysis, and the tolerance (predicted extraction times) of the user account after extracting the target extraction object can be obtained in combination with historical object extraction information, so that the target object expected to be extracted for each user account can be determined in a targeted manner, and the object extraction strategy of the user account can be flexibly adjusted. The present disclosure can accurately and efficiently find the individualized differences of the target extraction object of the user account, and quantize the tolerance of the user account. The whole process is real-time, adaptive, and convenient for subsequent data information analysis and processing of the obtained target extraction object and predicted extraction times.

[0038] In the following, the information processing method and device according to the present disclosure will be described in detail with reference to Figures 1 to 4

[0039] Figure 1 is a flowchart of an information processing method according to an exemplary embodiment. Referring to Figure 1 In step 101, the participation time series of the extraction object of the user account can be obtained according to the historical behavior information of the extraction object of the user account.

[0040] ​It should be noted that the user account information involved in the present disclosure includes but is not limited to user account device information, user account personal information, and user account historical behavior information, which are all information authorized by the corresponding user account or fully authorized by all parties.

[0041] The exemplary embodiments of the present disclosure can extract features based on historical behavior information related to the extraction object of the user account, assign weights to the finally extracted features, calculate the participation degree for each extraction object action, which represents the participation activity of the user account at the time of extraction, and then obtain the participation degree time sequence, which is specifically as follows.

[0042] First, the feature information of the user account extracting each object can be obtained according to the historical behavior information.

[0043] For example, the user account on the content platform randomly extracts videos from a certain number of videos for playing, and the historical behavior information can be related behavior information of the user account historically extracting videos, such as extraction time, extraction times, video watching time of extracted videos, etc. The exemplary embodiments of the present disclosure can obtain the feature information of the user account extracting each video according to the historical behavior information, such as feature information which can include but is not limited to the time interval between the current extraction and the last extraction, the number of times of extraction before the current extraction, the extraction category (single extraction / multiple extraction / continuous extraction) of the current extraction, the number of specific data consumed by the user account for the current extraction, and the number of specific data consumed by the user account per day.

[0044] Then, the derived feature information of the user account extracting each object can be constructed according to the feature information.

[0045] The exemplary embodiments of the present disclosure need to calculate the participation degree for each extraction action (each extraction action corresponds to an object extraction event), so it is necessary to construct features that can reflect the participation degree for the extraction action, that is, the derived feature information. For example, the exemplary embodiments of the present disclosure can construct the derived feature information of the user account extracting each object according to the feature information related to the extraction video obtained in the previous step, such as derived feature information which can include but is not limited to the extraction frequency calculated in real time at the time of the current extraction and the value (excellent / good / medium / poor) of the video extracted at the current extraction based on a preset condition. It should be noted that the derived feature information of the extraction of each object can include at least one.

[0046] Finally, the participation degree of the user account extracting each object can be obtained according to the derived feature information, and the participation degree can be arranged into a participation degree time sequence in the order of the extraction time of each object.

[0047] According to an example embodiment of the present disclosure, the feature values of the derived feature information of each object extracted by the user account can be obtained according to the derived feature information. Then, the engagement of the user account in extracting each object can be calculated according to the feature values and first preset weights of the derived feature information. The first preset weights can be obtained by the coefficient of variation method according to the derived feature information.

[0048] For example, according to the derived feature information of the extracted video obtained in the previous step, the feature values of the derived feature information of each object extracted by the user account can be obtained according to an example embodiment of the present disclosure. For example, if the value of the extracted video determined based on the preset condition is good, the feature value can be 70 points. According to an example embodiment of the present disclosure, the engagement of the user account in extracting each object can be calculated according to the feature values and first preset weights of the derived feature information. For example, the derived feature information and the corresponding first preset weights can be multiplied, and then the products can be added to obtain the engagement. According to an example embodiment of the present disclosure, the engagement can be arranged into an engagement time sequence in the order of the extraction time of each object. The engagement time sequence can be a sequence with the extraction time as the horizontal coordinate and the engagement as the vertical coordinate.

[0049] According to the information processing method of the present disclosure, the engagement time sequence can be obtained through the historical behavior information, which can depict the fluctuation of the participation activity of the user account in the object extraction process, and facilitate subsequent analysis and determination of the target extraction object. The engagement can be calculated through the feature values and the first preset weights, which can accurately map the participation activity of each extraction action.

[0050] After obtaining the engagement time sequence, according to an example embodiment of the present disclosure, it can be determined which time or which times the engagement is abnormal (the abnormal time can reflect the particularity of the object extracted by the user account to the user account), and then the object extracted at the abnormal time can be taken as the target extraction object. Specifically, in step 102, the abnormal extraction time can be determined in the engagement time sequence, the object corresponding to the abnormal extraction time can be obtained, and the object corresponding to the abnormal extraction time can be taken as the target extraction object of the user account.

[0051] According to an example embodiment of the present disclosure, the engagement time sequence can be smoothed before step 102.

[0052] According to an example embodiment of the present disclosure, the idea of ensemble learning can be used to analyze the engagement time sequence by an anomaly detection algorithm to find the time inflection point, so as to obtain the abnormal extraction time, and the object corresponding to the abnormal extraction time can be taken as the target extraction object. Specifically, the following steps can be taken.

[0053] Firstly, the engagement time series can be subjected to anomaly detection by at least one anomaly detection algorithm to obtain an anomaly time detected by each anomaly detection algorithm. For example, the anomaly detection algorithm can include, but is not limited to, a sigma anomaly detection method, a coefficient of variation method.

[0054] Then, an extraction time of the anomaly can be determined from the anomaly time detected by each anomaly detection algorithm, an object corresponding to the extraction time of the anomaly is obtained, and the object corresponding to the extraction time of the anomaly is taken as the target extraction object of the user account.

[0055] The exemplary embodiments of the present disclosure can select one of the anomaly times detected by all anomaly detection algorithms, determine the confidence of the selected anomaly time, and further determine the target extraction object. Specifically, the first anomaly time detected by each anomaly detection algorithm can be determined based on a second preset weight of each anomaly detection algorithm. Then, whether the anomaly time to be accepted is taken as the extraction time of the anomaly can be determined according to the confidence of the object extracted by the anomaly time to be accepted.

[0056] For example, the anomaly time detected by the anomaly detection algorithm with the largest second preset weight can be taken as the anomaly time to be accepted.

[0057] According to the exemplary embodiments of the present disclosure, the anomaly time to be accepted can be a time point or a time interval. Then, in the case where the anomaly time to be accepted is a time point, only one extracted object is included in the anomaly time to be accepted, the confidence of the object extracted at the time point can be calculated, in the case where the anomaly time to be accepted is a time interval, a plurality of extraction actions can be included in the time interval, a plurality of extracted objects can be included in the anomaly time to be accepted, a time point in the time interval can be determined, for example, the extraction time of the object with the largest number of specific data corresponding to the object extracted in the time interval can be taken as the final anomaly time to be accepted, and the final anomaly time to be accepted can also be determined according to the habit of the user account consuming the number of specific data.

[0058] According to the exemplary embodiments of the present disclosure, the confidence of the object extracted at the abnormal time to be accepted can be calculated, the confidence representing the confidence of taking the object extracted at the abnormal time to be accepted as the target extracted object of the user account, and the confidence can be calculated by performing consistency checking on the historical behavior information of the extracted object of the user account. Then, it can be determined whether to take the abnormal time to be accepted as the abnormal extracted time according to the confidence. If the confidence of the object extracted at the abnormal time to be accepted is greater than or equal to a preset threshold, the abnormal time to be accepted can be taken as the abnormal extracted time. If the confidence of the object extracted at the abnormal time to be accepted is less than the preset threshold, the second preset weight can be adjusted and the abnormal time to be accepted can be determined again until the confidence of the object extracted at the abnormal time to be accepted is greater than or equal to the preset threshold.

[0059] According to the information processing method of the present disclosure, the determination of the abnormal extracted time can be performed according to at least one abnormality detection algorithm, so that the determination of the abnormal extracted time can be more accurate. The step of determining the abnormal extracted time according to the confidence can improve the accuracy of the determination of the abnormal extracted time. The second preset weight is adjusted to be determined again when the confidence is less than the preset threshold, so that the scheme is more flexible. By smoothing the participation time sequence, the disturbance of dirty data can be reduced, and subsequent abnormal analysis can be facilitated.

[0060] In step 103, the predicted extraction times of the user account after extracting the target extracted object can be obtained according to the object extraction information of the user account in a predetermined historical period, wherein the object extraction information includes the object information extracted by the user account in the predetermined historical period, and the object information can include, but is not limited to, the object and the extraction time.

[0061] According to the exemplary embodiments of the present disclosure, the predicted extraction times of the user account after extracting the target extracted object can be obtained by regression prediction, specifically, the object extraction information of the user account in a predetermined historical period can be taken as the independent variable, and the extraction times of the user account after extracting the target extracted object can be taken as the dependent variable to perform regression prediction. The predetermined historical period can be a historical period of the day, or a historical period before the day.

[0062] However, the user account can fail to detect any historical object extraction information. Then, the exemplary embodiments of the present disclosure can find a similar user account for this case to obtain the predicted extraction times, specifically, after step 102, a similarity model of the user account can be first constructed in the case that the user account fails to detect the object extraction information of the user account. Then, a similar user account of the user account can be determined according to the similarity model, and the predicted extraction times corresponding to the similar user account can be taken as the predicted extraction times of the user account after the target extraction object is extracted. It should be noted that the similarity model can be constructed according to the portrait and historical behavior information of the user account. The exemplary embodiments of the present disclosure can optimize the similarity model to ensure the fitting accuracy of the similar user account.

[0063] According to the exemplary embodiments of the present disclosure, the object extracted by the first predicted extraction times can be set as the target extraction object in the case that the target extraction object has not been extracted all the time.

[0064] According to the information processing method of the present disclosure, the similarity model can be constructed in the case that the user account fails to detect the object extraction information, and then the predicted extraction times are determined, so that the scheme is more flexible.

[0065] Figure 2 is a flowchart of another information processing method according to an exemplary embodiment. The following describes the information processing method in the exemplary embodiments of the present disclosure with reference to Figure 2 , the information processing method in the exemplary embodiments of the present disclosure is further described.

[0066] Firstly, the feature information of the user account extracting each object can be obtained according to the historical behavior information.

[0067] Then, the derived feature information of the user account extracting each object can be constructed according to the feature information.

[0068] Next, the participation degree of the user account extracting each object can be obtained according to the derived feature information, and the participation degrees are arranged into a participation degree time sequence in the order of the extraction time of each object.

[0069] Further, the participation degree time sequence can be smoothed.

[0070] Then, the participation degree time sequence can be detected by at least one anomaly detection algorithm to obtain the anomaly time detected by each anomaly detection algorithm, and based on the second preset weight of each anomaly detection algorithm, the to-be-inspected anomaly time (time point or time interval) is determined from the anomaly time detected by each anomaly detection algorithm.

[0071] Then, the object list extracted in the abnormality acceptance time to be accepted can be obtained. In a case where there is only one object in the object list, the confidence of the object extracted in the abnormality acceptance time to be accepted can be calculated. In a case where there are multiple objects in the object list, one object can be determined from the multiple objects, and the extraction time of the object can be taken as the abnormality acceptance time to be accepted. The confidence of the object extracted in the abnormality acceptance time to be accepted can be calculated.

[0072] Then, whether the abnormality acceptance time to be accepted is taken as the extraction time of the abnormality can be determined according to the confidence of the object extracted in the abnormality acceptance time to be accepted, that is, the size relationship between the confidence of the object extracted in the abnormality acceptance time to be accepted and the preset threshold value. In a case where the confidence of the object extracted in the abnormality acceptance time to be accepted is greater than or equal to the preset threshold value, the abnormality acceptance time to be accepted is taken as the extraction time of the abnormality. In a case where the confidence of the object extracted in the abnormality acceptance time to be accepted is less than the preset threshold value, the second preset weight is adjusted, and the abnormality acceptance time to be accepted is determined again until the confidence of the object extracted in the abnormality acceptance time to be accepted is greater than or equal to the preset threshold value.

[0073] Then, the object corresponding to the extraction time of the abnormality can be taken as the target extraction object of the user account.

[0074] Next, the predicted extraction times of the user account after the target extraction object is extracted can be obtained according to the object extraction information of the user account in a predetermined historical period, or in a case where the object extraction information of the user account is not detected, a similarity model of the user account is constructed. According to the similarity model, the similar user accounts of the user account are determined, and the predicted extraction times corresponding to the similar user accounts are taken as the predicted extraction times of the user account after the target extraction object is extracted.

[0075] Finally, the target extraction object of the user account and the predicted extraction times of the user account after the target extraction object is extracted can be obtained through the above steps.

[0076] According to the example embodiments of the present disclosure, the information processing method flow shown in Figure 2 may be executed periodically, which can be every day or every extraction action.

[0077] Next, the information processing method shown in Figure 2 will be described with a specific example.

[0078] Firstly, the feature information of the user account extracting each video can be obtained according to the related behavior information of the video extraction such as extraction time, extraction times, and video watching time of the extracted video, and the feature information can include, but is not limited to, the time interval between the current extraction and the last extraction, the number of times of extraction before the current extraction, the extraction category of the current extraction (single extraction / multiple extraction / continuous extraction), the amount of specific data consumed by the user account in the current extraction, and the daily average amount of specific data consumed by the user account.

[0079] Then, the derived feature information of the user account extracting each video can be constructed according to the feature information, and the derived feature information can include, but is not limited to, the extraction frequency calculated in real time at the time of the current extraction and the value of the video extracted at the current extraction (good / medium / poor) determined based on a preset condition.

[0080] Next, the engagement of the user account extracting each video can be obtained according to the derived feature information, and the engagement is arranged into an engagement time sequence in the order of the extraction time of each video.

[0081] Further, the engagement time sequence can be smoothed.

[0082] Then, the engagement time sequence can be detected by at least one anomaly detection algorithm to obtain the anomaly time detected by each anomaly detection algorithm, and based on the second preset weight of each anomaly detection algorithm, the anomaly time (time point or time interval) to be accepted can be determined from the anomaly time detected by each anomaly detection algorithm.

[0083] Next, the video list extracted in the anomaly time to be accepted can be obtained, and in the case that there is only one video in the video list, the confidence of the video extracted in the anomaly time to be accepted can be calculated; in the case that there are multiple videos in the video list, a video can be determined from the multiple videos and the extraction time of the video is taken as the anomaly time to be accepted, and the confidence of the video extracted in the anomaly time to be accepted can be calculated.

[0084] Further, whether the anomaly time to be accepted is taken as the abnormal extraction time can be determined according to the confidence of the video extracted in the anomaly time to be accepted, i.e., the size relationship between the confidence of the video extracted in the anomaly time to be accepted and the preset threshold value. In the case that the confidence of the video extracted in the anomaly time to be accepted is greater than or equal to the preset threshold value, the anomaly time to be accepted is taken as the abnormal extraction time; in the case that the confidence of the video extracted in the anomaly time to be accepted is less than the preset threshold value, the second preset weight is adjusted and the anomaly time to be accepted is re-determined until the confidence of the video extracted in the anomaly time to be accepted is greater than or equal to the preset threshold value.

[0085] Then, the video corresponding to the abnormal extraction time can be taken as the target extraction video of the user account.

[0086] Next, the predicted extraction times of the user account after extracting the target extraction video can be obtained according to the video extraction information of the user account in a predetermined historical period, or in the case where the video extraction information of the user account is not detected, a similarity model of the user account is constructed, the similar user accounts of the user account are determined according to the similarity model, and the predicted extraction times corresponding to the similar user accounts are taken as the predicted extraction times of the user account after extracting the target extraction video.

[0087] Finally, the target extraction video of the user account and the predicted extraction times of the user account after extracting the target extraction video can be obtained through the above steps.

[0088] Figure 3 is a block diagram of an information processing device according to an exemplary embodiment. Referring to Figure 3 , the information processing device 300 includes a sequence obtaining unit 301, an object determining unit 302, and a times predicting unit 303.

[0089] The sequence obtaining unit 301 can obtain the participation time sequence of the extraction object of the user account according to the historical behavior information of the extraction object of the user account.

[0090] The sequence obtaining unit 301 can obtain the feature information of each object extracted by the user account according to the historical behavior information.

[0091] For example, for a user account on a content platform to randomly extract a video from a certain number of videos for playback, the historical behavior information can be related behavior information such as extraction time, extraction times, video watching time, etc. of the user account historically extracting videos. The sequence obtaining unit 301 can obtain the feature information of each video extracted by the user account according to the historical behavior information. For example, the feature information can include, but is not limited to, the time interval between the current extraction and the last extraction, the number of times that have been extracted before the current extraction, the extraction category (single extraction / multiple extraction / continuous extraction) of the current extraction, the number of specific data consumed by the user account for the current extraction, and the number of specific data consumed by the user account per day.

[0092] The sequence obtaining unit 301 can construct the derived feature information of each object extracted by the user account according to the feature information.

[0093] The sequence acquisition unit 301 needs to calculate the participation degree for each extraction action (each extraction action corresponds to an object extraction event), so it needs to construct features reflecting the participation degree for the extraction action, that is, derived feature information. For example, the sequence acquisition unit 301 can construct the derived feature information of the user account extracting each object according to the feature information, such as the derived feature information can include, but is not limited to, the extraction frequency calculated in real time when the extraction, the value (good / normal / poor) of the video extracted at this time based on the preset condition. It should be noted that the derived feature information of extracting each object can include at least one.

[0094] The sequence acquisition unit 301 can obtain the participation degree of the user account extracting each object according to the derived feature information, and arrange the participation degree into a participation degree time sequence in the order of the extraction time of each object.

[0095] According to the example embodiment of the present disclosure, the sequence acquisition unit 301 can first obtain the feature value of each derived feature information when the user account extracts each object according to the derived feature information. The sequence acquisition unit 301 can then calculate the participation degree of the user account extracting each object according to the feature value and the first preset weight of each derived feature information. The first preset weight can be obtained by the coefficient of variation method according to the derived feature information.

[0096] For example, the sequence acquisition unit 301 can obtain the feature value of each derived feature information when the user account extracts each object according to the derived feature information, such as the value of the video extracted at this time based on the preset condition is normal, and the feature value can be 70 points. The sequence acquisition unit 301 can also calculate the participation degree of the user account extracting each object according to the feature value and the first preset weight of each derived feature information, such as multiplying each derived feature information and its corresponding first preset weight, and then adding the products to obtain the participation degree. The sequence acquisition unit 301 can also arrange the participation degree into a participation degree time sequence in the order of the extraction time of each object, and the participation degree time sequence can be a sequence with the extraction time as the horizontal coordinate and the participation degree as the vertical coordinate.

[0097] According to the information processing device of the present disclosure, the sequence acquisition unit 301 can obtain the participation degree time sequence through the historical behavior information, which can depict the fluctuation of the participation activity of the user account in the object extraction process, and facilitate subsequent analysis and determination of the target extraction object. The sequence acquisition unit 301 can calculate the participation degree through the feature value and the first preset weight, which can make a more accurate mapping of the participation activity of each extraction action.

[0098] The object determination unit 302 can determine the abnormal extraction time in the participation time sequence, obtain the object corresponding to the abnormal extraction time, and take the object corresponding to the abnormal extraction time as the target extraction object of the user account.

[0099] According to the example embodiments of the present disclosure, a smoothing processing unit is further included, which can perform smoothing processing on the participation time sequence. The smoothing processing unit can be executed before the object determination unit 302.

[0100] The object determination unit 302 can find the time inflection point by analyzing the participation time sequence through the abnormality detection algorithm based on the idea of ensemble learning, thereby obtaining the abnormal extraction time, and taking the object corresponding to the abnormal extraction time as the target extraction object. Details are described as follows.

[0101] The object determination unit 302 can perform abnormality detection on the participation time sequence through at least one abnormality detection algorithm to obtain the abnormal time detected by each abnormality detection algorithm. For example, the abnormality detection algorithm can include, but is not limited to, the sigma abnormality detection method and the coefficient of variation method.

[0102] The object determination unit 302 can determine the abnormal extraction time from the abnormal time detected by each abnormality detection algorithm, obtain the object corresponding to the abnormal extraction time, and take the object corresponding to the abnormal extraction time as the target extraction object of the user account.

[0103] The object determination unit 302 can first determine the abnormality time to be accepted from the abnormal time detected by each abnormality detection algorithm based on the second preset weight of each abnormality detection algorithm. The object determination unit 302 can then determine whether to take the abnormality time to be accepted as the abnormal extraction time according to the confidence of the object extracted by the abnormality time to be accepted.

[0104] For example, the object determination unit 302 can take the abnormal time detected by the abnormality detection algorithm with the largest second preset weight as the abnormality time to be accepted.

[0105] According to an example embodiment of the present disclosure, the abnormality acceptance time can be a time point or a time interval. When the abnormality acceptance time is a time point, only one extracted object is included in the abnormality acceptance time, and the object determination unit 302 can calculate the confidence of the object extracted at the time point. When the abnormality acceptance time is a time interval, multiple extraction actions can be included in the time interval, and multiple extracted objects can be included in the abnormality acceptance time. The object determination unit 302 can determine a time point in the time interval, for example, the extraction time of the object corresponding to the largest number of specific data of the objects extracted in the time interval can be taken as the final abnormality acceptance time, and the final abnormality acceptance time can also be determined according to the habit of the user account consuming the number of specific data.

[0106] According to an example embodiment of the present disclosure, the object determination unit 302 can calculate the confidence of the object extracted at the abnormality acceptance time, which represents the confidence of taking the object extracted at the abnormality acceptance time as the target extraction object of the user account. The confidence can be calculated by consistency checking the historical behavior information of the extraction object of the user account. Further, the object determination unit 302 can determine whether to take the abnormality acceptance time as the abnormal extraction time according to the confidence. When the confidence of the object extracted at the abnormality acceptance time is greater than or equal to a preset threshold, the object determination unit 302 can take the abnormality acceptance time as the abnormal extraction time. When the confidence of the object extracted at the abnormality acceptance time is less than the preset threshold, the object determination unit 302 can adjust the second preset weight and redetermine the abnormality acceptance time until the confidence of the object extracted at the abnormality acceptance time is greater than or equal to the preset threshold.

[0107] According to the information processing device of the present disclosure, the object determination unit 302 can determine the abnormal extraction time according to at least one abnormality detection algorithm, so as to more accurately determine the abnormal extraction time. The object determination unit 302 can improve the accuracy of determining the abnormal extraction time by determining the abnormal extraction time through the confidence. The object determination unit 302 can make the scheme more flexible by adjusting the second preset weight to redetermine when the confidence is less than the preset threshold. The smoothing processing unit can reduce the disturbance of dirty data and facilitate subsequent abnormal analysis by smoothing the participation time sequence.

[0108] The number of times prediction unit 303 can obtain the predicted extraction number of times of the user account after extracting the target extraction object according to the object extraction information of the user account in a predetermined historical period, wherein the object extraction information includes the object information extracted by the user account in the predetermined historical period.

[0109] According to an example embodiment of the present disclosure, the number of times prediction unit 303 can obtain the predicted number of times of extraction of the user account after extracting the target extraction object through regression prediction, specifically, the number of times prediction unit 303 can perform regression prediction by taking the object extraction information of the user account in a predetermined historical period as the independent variable and taking the number of times of extraction of the user account after extracting the target extraction object as the dependent variable. The predetermined historical period can be a historical period of the day or a historical period before the day.

[0110] However, the user account can fail to detect any historical object extraction information. Then, an example embodiment of the present disclosure can find a similar user account for this case to obtain the predicted number of times of extraction, specifically, the example embodiment of the present disclosure can further include a similarity construction unit, which can first construct a similarity model of the user account in the case where the object extraction information of the user account is not detected. The similarity construction unit can then determine a similar user account of the user account according to the similarity model and take the predicted number of times of extraction corresponding to the similar user account as the predicted number of times of extraction of the user account after extracting the target extraction object. It should be noted that the similarity construction unit can construct the similarity model according to the portrait and historical behavior information of the user account.

[0111] According to the information processing device of the present disclosure, the similarity construction unit can construct the similarity model in the case where the object extraction information of the user account is not detected, and further determine the predicted number of times of extraction, so that the scheme is more flexible.

[0112] Figure 4 is a block diagram of an electronic device 400 according to an example embodiment.

[0113] Referring to Figure 4 The electronic device 400 includes at least one memory 401 having a set of computer executable instructions stored therein and at least one processor 402, which, when executing the set of computer executable instructions, performs the information processing method according to an example embodiment of the present disclosure.

[0114] As an example, the electronic device 400 can be a PC computer, a tablet device, a personal digital assistant, a smart phone, or other devices capable of executing the above-mentioned set of instructions. Here, the electronic device 400 is not necessarily a single electronic device, but can also be a collection of any devices or circuits capable of executing the above-mentioned instructions (or set of instructions) individually or jointly. The electronic device 400 can also be part of an integrated control system or a system manager, or can be configured as a portable electronic device that interfaces with local or remote (e.g., via wireless transmission) devices.

[0115] In electronic device 400, processor 402 can include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. By way of example and not limitation, the processor can also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0116] Processor 402 can execute instructions or code stored in memory 401, which can also store data. The instructions and data can also be transmitted and received via a network through network interface device, which can employ any known transmission protocol.

[0117] Memory 401 can be integrated with processor 402, e.g., disposed within an integrated circuit microprocessor, etc. Further, memory 401 can include a separate device, such as an external disk drive, a storage array, or other storage device usable by any database system. Memory 401 and processor 402 can be operatively coupled, or can communicate with each other, e.g., through I / O ports, network connections, etc., such that processor 402 can read files stored in memory.

[0118] Further, electronic device 400 can also include a video display, such as a liquid crystal display, and a user interface, such as a keyboard, a mouse, a touch input device, etc. All components of electronic device 400 can be connected via a bus and / or network.

[0119] According to an exemplary embodiment of the present disclosure, there can also be provided a computer-readable storage medium in which instructions stored therein are executable by at least one processor to cause the at least one processor to execute an information processing method according to an exemplary embodiment of the present disclosure. Examples of the computer-readable storage medium herein include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disk memory, a hard disk drive (HDD), a solid state drive (SSD), a card-type memory such as a multimedia card, a secure digital (SD) card, or an extreme digital (XD) card, a magnetic tape, a floppy disk, a magneto-optical data storage device, an optical data storage device, a hard disk, a solid state disk, and any other device configured to store a computer program and any associated data, data files, and data structures in a non-transitory manner and provide the computer program and any associated data, data files, and data structures to a processor or computer so that the processor or computer can execute the computer program. The computer program in the computer-readable storage medium described above can be executed in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc., and, in addition, in one example, the computer program and any associated data, data files, and data structures are distributed over a networked computer system so that the computer program and any associated data, data files, and data structures are stored, accessed, and executed in a distributed manner by one or more processors or computers.

[0120] According to an exemplary embodiment of the present disclosure, there can also be provided a computer program product in which instructions executable by a processor of a computer device can execute an information processing method according to an exemplary embodiment of the present disclosure.

[0121] According to the information processing method and device, the target extraction object of the user account can be obtained through data analysis, and the tolerance (predicted extraction times) of the user account after extracting the target extraction object can be obtained in combination with historical object extraction information, so that the target object expected to be extracted can be determined for each user account, and the object extraction strategy of the user account can be flexibly adjusted. The disclosure can accurately and efficiently find the personalized differences of the target extraction object of the user account, and quantize the tolerance of the user account. The whole process is real-time and adaptive, and is convenient for subsequent data information analysis and processing of the obtained target extraction object and predicted extraction times.

[0122] In addition, according to the information processing method and device, in the case of using multiple data analysis algorithms, the algorithms are not directly spliced for use, but are flexibly combined to give full play to the advantages of each algorithm to obtain the target extraction object and the predicted extraction times.

[0123] Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses or adaptive changes of this disclosure that follow the general principles of the disclosure and include common knowledge or conventional technical means in the art not disclosed in the disclosure. The specification and examples are only regarded as exemplary, and the true scope and spirit of the disclosure are indicated by the following claims.

[0124] It should be understood that the disclosure is not limited to the precise structures described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the disclosure is limited only by the appended claims.

Claims

1. An information processing method, characterized in that, include: Based on the historical behavior information of the user account's extracted objects, a time series of the user account's participation in the extracted objects is obtained, wherein the time series of participation is a sequence with the extraction time on the horizontal axis and the participation degree on the vertical axis, wherein the participation degree represents the user account's participation activity at each extraction object. In the participation time series, identify the abnormal extraction time, obtain the object corresponding to the abnormal extraction time, and use the object corresponding to the abnormal extraction time as the target extraction object of the user account. Based on the object extraction information of the user account in a predetermined historical period, the predicted number of extractions after the target object is extracted is obtained. The object extraction information includes object information extracted from the user account during the predetermined historical period.

2. The information processing method as described in claim 1, characterized in that, The step of obtaining the participation time series of the user account's extraction targets based on the historical behavior information of the user account's extraction targets includes: Based on the historical behavior information, feature information of each object in the user account is extracted; Based on the aforementioned feature information, the derived feature information for each object in the user account is constructed. Based on the derived feature information, the participation degree of each object extracted by the user account is obtained, and the participation degree is arranged into a participation degree time series according to the extraction time of each object.

3. The information processing method as described in claim 2, characterized in that, The step of obtaining the participation degree of each object extracted from the user account based on the derived feature information includes: Based on the derived feature information, the feature values ​​of each derived feature information when extracting each object from the user account are obtained; Based on the feature value and the first preset weight of each derived feature information, the participation degree of the user account in extracting each object is calculated.

4. The information processing method as described in claim 1, characterized in that, The step of determining the abnormal extraction time in the participation time series, obtaining the object corresponding to the abnormal extraction time, and using the object corresponding to the abnormal extraction time as the target extraction object of the user account includes: Anomalies are detected in the participation time series by at least one anomaly detection algorithm, and the abnormal times detected by each anomaly detection algorithm are obtained. The extraction time of the anomaly is determined from the anomaly time detected by each anomaly detection algorithm, the object corresponding to the extraction time of the anomaly is obtained, and the object corresponding to the extraction time of the anomaly is used as the target extraction object of the user account.

5. The information processing method as described in claim 4, characterized in that, Determining the extraction time of the anomaly from the anomaly times detected by each anomaly detection algorithm includes: Based on the second preset weight of each anomaly detection algorithm, the anomaly time to be accepted is determined from the anomaly time detected by each anomaly detection algorithm. Whether to use the abnormal time to be inspected as the extraction time of the abnormality is determined based on the confidence level of the object extracted from the abnormal time to be inspected.

6. The information processing method as described in claim 5, characterized in that, The step of determining whether to use the pending acceptance anomaly time as the extraction time of the anomaly based on the confidence level of the object extracted from the anomaly time includes: If the confidence level of the object extracted during the pending acceptance anomaly time is greater than or equal to a preset threshold, the pending acceptance anomaly time shall be used as the extraction time of the anomaly. If the confidence level of the object extracted during the abnormal acceptance time is less than the preset threshold, the second preset weight is adjusted and the abnormal acceptance time is redefined until the confidence level of the object extracted during the abnormal acceptance time is greater than or equal to the preset threshold. The confidence level represents the confidence level of using the object extracted from the abnormal time to be accepted as the target extraction object of the user account.

7. The information processing method as described in claim 1, characterized in that, After identifying the abnormal extraction time in the participation time series, obtaining the object corresponding to the abnormal extraction time, and using the object corresponding to the abnormal extraction time as the target extraction object of the user account, the method further includes: In the absence of detected object extraction information for the user account, a similarity model for the user account is constructed. Based on the similarity model, similar user accounts of the user account are determined, and the predicted number of extractions corresponding to the similar user accounts is used as the predicted number of extractions of the user account after the target extraction object is extracted.

8. An information processing device, characterized in that, include: The sequence acquisition unit is configured to: obtain the participation time series of the user account's extracted objects based on the historical behavior information of the user account's extracted objects, wherein the participation time series is a sequence with the extraction time on the horizontal axis and the participation degree on the vertical axis, wherein the participation degree represents the user account's participation activity at each extraction object. The object determination unit is configured to: determine the abnormal extraction time in the participation time series, obtain the object corresponding to the abnormal extraction time, and use the object corresponding to the abnormal extraction time as the target extraction object of the user account. The number prediction unit is configured to: obtain the predicted number of times the user account will extract the target object after the user account has extracted it, based on the object extraction information of the user account in a predetermined historical period, wherein the object extraction information includes the object information extracted by the user account in the predetermined historical period.

9. The information processing apparatus as described in claim 8, characterized in that, The sequence acquisition unit is configured as follows: Based on the historical behavior information, feature information of each object in the user account is extracted; Based on the aforementioned feature information, the derived feature information for each object in the user account is constructed. Based on the derived feature information, the participation degree of each object extracted by the user account is obtained, and the participation degree is arranged into a participation degree time series according to the extraction time of each object.

10. The information processing apparatus as described in claim 9, characterized in that, The sequence acquisition unit is configured as follows: Based on the derived feature information, the feature values ​​of each derived feature information when extracting each object from the user account are obtained; Based on the feature value and the first preset weight of each derived feature information, the participation degree of the user account in extracting each object is calculated.

11. The information processing apparatus as described in claim 8, characterized in that, The object determination unit is configured as follows: Anomalies are detected in the participation time series by at least one anomaly detection algorithm, and the abnormal times detected by each anomaly detection algorithm are obtained. The extraction time of the anomaly is determined from the anomaly time detected by each anomaly detection algorithm, the object corresponding to the extraction time of the anomaly is obtained, and the object corresponding to the extraction time of the anomaly is used as the target extraction object of the user account.

12. The information processing apparatus as claimed in claim 11, characterized in that, The object determination unit is configured as follows: Based on the second preset weight of each anomaly detection algorithm, the anomaly time to be accepted is determined from the anomaly time detected by each anomaly detection algorithm. Whether to use the abnormal time to be inspected as the extraction time of the abnormality is determined based on the confidence level of the object extracted from the abnormal time to be inspected.

13. The information processing apparatus as described in claim 12, characterized in that, The object determination unit is configured as follows: If the confidence level of the object extracted during the pending acceptance anomaly time is greater than or equal to a preset threshold, the pending acceptance anomaly time shall be used as the extraction time of the anomaly. If the confidence level of the object extracted during the abnormal acceptance time is less than the preset threshold, the second preset weight is adjusted and the abnormal acceptance time is redefined until the confidence level of the object extracted during the abnormal acceptance time is greater than or equal to the preset threshold. The confidence level represents the confidence level of using the object extracted from the abnormal time to be accepted as the target extraction object of the user account.

14. The information processing apparatus as described in claim 8, characterized in that, It also includes similar building blocks, configured as follows: In the absence of detected object extraction information for the user account, a similarity model for the user account is constructed. Based on the similarity model, similar user accounts of the user account are determined, and the predicted number of extractions corresponding to the similar user accounts is used as the predicted number of extractions of the user account after the target extraction object is extracted.

15. An electronic device, characterized in that, include: At least one processor; At least one memory that stores computer-executable instructions. The computer-executable instructions, when executed by the at least one processor, cause the at least one processor to perform the information processing method as described in any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that, When the instructions stored in the computer-readable storage medium are executed by at least one processor, the at least one processor causes the at least one processor to perform the information processing method as described in any one of claims 1 to 7.

17. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by at least one processor, they implement the information processing method as described in any one of claims 1 to 7.

Images