Interest model update method facing to odd discovery recommendation

Abstract

The invention discloses an odd-finding-recommendation-oriented interest model updating method, which is used in a portal personalized recommendation service system; the invention is characterized in that: the odd-finding-recommendation-oriented interest model updating method comprises a long-term interest model modification rule LMR, a long-term interest model update rule LUR, a short-term interest model structure / update rule SUR, and an interest drifting detection rule IDR. The user interest model updating method of the invention can partly update interest description of the user, obtain the short-term interest situation of the user easily, satisfy the need of the odd finding recommendation, and realize the efficient and accurate updating of a user interest model; and moreover, the method can identify and process noise data, and avoid influences of the noise data on the stability of the user interest model when quickly responding to interest drifting at the same time. The updating method of the invention can identify and process the noise data, thus improving the accuracy and the stability of the user interest model.

Description

technical field [0001] The invention relates to a method for updating an interest model for a portal personalized recommendation service, more particularly, a method for updating an interest model based on interest drift detection and oriented to singular discovery and recommendation. Background technique [0002] The recommendation service technology was applied in the field of e-commerce in the 1990s, and has been further developed in the field of personalized service. Portal personalized service is the expansion of personalized service in large-scale portal applications, and it is also a new application and development field of personalized service. The user interest model in the personalized recommendation service is a computable and specific formal description of user interests, preferences, patterns, etc. The research on the user interest model covers the representation, learning, update and storage of the interest model and other related technologies . In order to s...

AI Technical Summary

Problems solved by technology

[0005] The proposed update methods of the interest model mainly include the following types: the sliding window method uses a variable-sized sliding window to select the data for model update, which improves the computational efficiency compared with using all historical data to reconstruct the interest model, but This method completely discards the historical data located outside the window and uses new data to build a new model. Therefore, the scale of feedback data is relatively high. If the feedback information does not cover all user interest categories, it cannot keep the information not involved in the feedback after the update. Description of the category of interest; the progressive forgetting method adds an age weight to each historical data, and the greater the "age", the smaller the weight of the data, which reflects to a certain extent the gradual weakening of the influence of historical data on the model, but this method only defines The desalination process of the original interest description is not involved, and it does not involve how to add new feedback to the interest model; the improved sliding window method searches for similar information in historical data according to the current recommended context, and uses similar historical information and feedback information together. Constructing a new interest model reduces the requirement for the size of the feedback data, but it still cannot maintain the description of interests not involved in the feedback information; the model update method based on genetic algorithm reconstructs the interest model according to the user feedback information, and makes the model The recommendation result is the closest to the feedback information. With the help of the characteristics of the genetic algorithm, this method can handle the noise data better, but this method also cannot retain the historical data in the model, and the update process must calculate the recommendation result many times, and its efficiency Depending on the specific recommendation algorithm, the computational efficiency cannot be guaranteed, which also reduces the generality of the algorithm; the method based on the artificial neural network uses the perceptron model to update the user's interest and complete the recommendation. There is also the problem of not being able to maintain the description of interest that is not involved in the feedback, and the requirement for the scale of the feedback data is very high. If the original interest model is updated on the basis of the original neural network interest model, the original interest model is required to use neural network representation, which weakens the model update. Algorithm versatility

[0006] To sum up, none of the existing interest model update methods can fully meet the requirements of singular discovery recommendation for the interest model update process, and there are also certain limitations in noise processing.

Images