Feature effectiveness assessment method and apparatus, electronic device, and storage medium

An electronic device, comprising:

• a processor, a memory, and executable instructions stored in the memory;
• and
• the executable instructions stored in the memory, when executed by the processor, causing the electronic device to perform a feature effectiveness assessment method including:
• constructing a feature combination set comprising a plurality of feature combinations, each feature combination being a combination of original features of candidate recommendation content;
• obtaining feature value sets each of which corresponds to a corresponding feature combination of the plurality of feature combinations, each feature value set comprising feature values of combined features based on the corresponding feature combination;
• for each of the combined features, determining a positive sample statistic corresponding to each feature value of the combined feature and a negative sample statistic corresponding to the feature value of the combined feature, wherein the determining the positive sample statistic and the negative sample statistic comprises:
• determining a corresponding first prediction accuracy when positive samples comprising the feature value are training samples and a corresponding second prediction accuracy when negative samples comprising the feature value are training samples,
• determining the positive sample statistic corresponding to the feature value based on the corresponding first prediction accuracy, which indicates an accuracy of performing recommendation based on the positive samples of the feature value, and
• determining the negative sample statistic corresponding to the feature value based on the corresponding second prediction accuracy, which indicates an accuracy of performing recommendation based on the negative samples of the feature value;
• obtaining weight values corresponding to the combined features based on the positive sample statistics and the negative sample statistics corresponding to the feature values of the combined features;
• constructing weight value sets corresponding to the plurality of feature combinations based on the weight values of the combined features;
• and
• respectively determining effectivenesses of the plurality of feature combinations based on the weight value sets of the plurality of feature combinations, the effectiveness being used for predicting an accuracy of performing content recommendation according to features obtained based on the corresponding feature combination.

A non-transitory computer-readable storage medium, storing executable instructions, the executable instructions, when executed by a processor of an electronic device, causing the electronic device to perform a feature effectiveness assessment method including:

• constructing a feature combination set comprising a plurality of feature combinations, each feature combination being a combination of original features of candidate recommendation content;
• obtaining feature value sets each of which corresponds to a corresponding feature combination of the plurality of feature combinations, each feature value set comprising feature values of combined features based on the corresponding feature combination;
• for each of the combined features, determining a positive sample statistic corresponding to each feature value of the combined feature and a negative sample statistic corresponding to the feature value of the combined feature, wherein the determining the positive sample statistic and the negative sample statistic comprises:
• determining a corresponding first prediction accuracy when positive samples comprising the feature value are training samples and a corresponding second prediction accuracy when negative samples comprising the feature value are training samples,
• determining the positive sample statistic corresponding to the feature value based on the corresponding first prediction accuracy, which indicates an accuracy of performing recommendation based on the positive samples of the feature value, and
• determining the negative sample statistic corresponding to the feature value based on the corresponding second prediction accuracy, which indicates an accuracy of performing recommendation based on the negative samples of the feature value;
• obtaining weight values corresponding to the combined features based on the positive sample statistics and the negative sample statistics corresponding to the feature values of the combined features;
• constructing weight value sets corresponding to the plurality of feature combinations based on the weight values of the combined features;
• and
• respectively determining effectivenesses of the plurality of feature combinations based on the weight value sets of the plurality of feature combinations, the effectiveness being used for predicting an accuracy of performing content recommendation according to features obtained based on the corresponding feature combination.