Video user gender classification method and device for method
A classification method and gender technology, applied in special data processing applications, instruments, electrical digital data processing, etc., can solve problems such as user attribute unknown state
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0108] A video site contains a set of videos { V 1 ,...,V K}, Each video is marked with a number of tags when uploaded by the user. The website obtained a small number of viewing records of users with gender tagging through registered users and questionnaire surveys. It is required to extract the tag features of the video tags and filter out the effective tag set.
[0109] First, extract all the tags of each video to get the tag set of all videos;
[0110] Then, according to the viewing records of gender-marked users, the number of views of male and female users on each label is counted;
[0111] After that, calculate the ratio of male and female users of each tag, calculate the tag attributes and tag features, and at the same time, calculate the total number of views of the tag;
[0112] Finally, the absolute value of the difference in the proportion of men and women is used as the discriminative expression of the label, and the total number of views on each label is expre...
example 2
[0115] Example 2: Training example of user classification model
[0116] When predicting user gender, you need to use a trained classification model. The classification model can be constructed by learning and training a classifier on a set of user viewing behavior features extracted from gender-marked user viewing records. The specific implementation process is as follows:
[0117] First, collect a number of gender-labeled users as a training data set;
[0118] Then, obtain the viewing records of users in the training data set for a period of time (for example: one week);
[0119] After that, construct the viewing behavior characteristics of the training data, that is: extract the video tags watched by each user, and calculate the viewing behavior characteristics of the user during this period through the tag attribute table;
[0120] Finally, using the feature set of viewing behavior characteristics of the training data, by maximizing the objective function:
[0121] ...
example 3
[0123] Example 3: An example of gender prediction for an unknown user on a video website
[0124] A certain user watches videos V1,...,Vn within a week and asks for gender prediction.
[0125] First, extract the tag set {tag 1 ,...,tag m}, and count the number of times the user viewed each tab .
[0126] Then, according to the label attribute table, look up the label characteristics of each label. If there is no certain label in the label attribute table, it is considered that the label has no effect on determining the user’s gender, and it is filtered out from the label set; if it exists, the regular The optimized feature value is used as the viewing behavior feature of the user on the label dimension .
[0127]
[0128] in, for this user in tag i views on , is the total number of valid views of the user during this period, is the label feature obtained from the table lookup. e.g. user U i After watching the tags {beauty and fitness (1 time), slimming y...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More