User recommendation method and recorded medium storing program for implementing the method
View Patent ↗A method of recommendation to a user is disclosed. In the method in accordance with an embodiment of the present invention, a recommendation system recommends a product to a user of an e-commerce site, which sells the product through the Internet. The method can include: collecting user behavior information of the user; analyzing the collected user behavior information; creating a recommendation tree for predicting a user preference for a certain product based on the analyzing; and recommending the product to the user based on the recommendation tree. The user behavior information is a command signal inputted by the user in the e-commerce site while the user is accessed to the e-commerce site, and the product can be classified into a plurality of items, which represent the product. The method in accordance with the present invention can analyze the user's preferences more quickly and accurately without inconveniencing the user.
1. A method of recommending a product to a user of an e-commerce site selling the product through the Internet, the product being recommended in a user recommendation system, the method comprising the steps of:
collecting user behavior information, wherein the user behavior information is comprised within a command signal inputted using an input device by the user in communication with the e-commerce site while the input device is accessing the e-commerce site in the user recommendation system;
computing a user preference on each of a plurality of items using usefulness of the user behavior information in the user recommendation system, wherein the product is classified into a plurality of items representing the product;
calculating a conditional probability value for each element included in the plurality of items in the user recommendation system;
creating a recommendation tree for predicting a user preference of the user for the product, the user preference of the user for the product being predicted based on the user preference on each of the plurality of items and the conditional probability value in the user recommendation system; and
recommending the product to the user based on the recommendation tree in the user recommendation system,
wherein the computing of the user preference uses an ID 3 algorithm and comprises:
calculating the usefulness by giving a score to the user behavior information in the user recommendation system; and
computing the user preference on each of the plurality of items based on the usefulness of the user behavior information in the user recommendation system,
wherein an equation for computing the user preference on each of the plurality of items is Entropy(S)=−p u≧3 log 2 p u≧3 −p u<3 log 2 p u<3 , and wherein Entropy(S) is the user preference on each of the plurality of items, p u≧3 is a probability when the usefulness of the user behavior information is greater than or equal to 3, and p u<3 is a probability when the usefulness of the user behavior information is smaller than 3.
2. The method of claim 1 , wherein the creating of the recommendation tree in the user recommendation system comprises the steps of:
forming a hierarchy level by arranging the plurality of items in a descending order of the user preference in the user recommendation system; and
arranging the elements in a descending order of the conditional probability value, the elements being included in the plurality of items in the user recommendation system.
3. The method of claim 1 , wherein the recommending of the product to the user based on the recommendation tree in the user recommendation system comprises the steps of:
calculating a similarity between the user preference and the product in the user recommendation system; and
recommending the product to the user in a descending order of the similarity in the user recommendation system.
4. The method of claim 3 , wherein the calculating of the similarity in the user recommendation system comprises the steps of:
computing a unit similarity by multiplying the preference of a hierarchy level and the conditional probability value of the element, the element corresponding to the product disposed in the hierarchy level in the user recommendation system; and
computing the similarity by adding the unit similarities in the user recommendation system.
5. A computer program product embodied in a non-transitory computer readable medium, the computer program for executing a method of recommending a product to a user of an e-commerce site selling the product through the Internet, the product being recommended in a user recommendation in system, when executed by a computer performs the steps of:
collecting user behavior information, wherein the user behavior information is comprised within a command signal inputted using an input device by the user in the e-commerce site while the input device is accessing the e-commerce site in the user recommendation system;
computing a user preference on each of a plurality of items using usefulness of the user behavior information in the user recommendation system, wherein the product is classified into a plurality of items representing the product;
calculating a conditional probability value for each element included in the plurality of items in the user recommendation system;
creating a recommendation tree for predicting a user preference of the user for the product, the user preference of the user for the product being predicted based on the user preference on each of the plurality of items and the conditional probability value in the user recommendation system; and
recommending the product to the user based on the recommendation tree in the user recommendation system,
wherein the computing of the user preference uses an ID 3 algorithm and comprises in the user recommendation system:
calculating the usefulness by giving a score to the user behavior information in the user recommendation system; and
computing the user preference on each of the plurality of items based on the usefulness of the user behavior information in the user recommendation system,
wherein an equation for computing the user preference on each of the plurality of items is Entropy(S)=−p u≧3 log 2 p u≧3 −p u<3 log 2 p u<3 , and wherein Entropy(S) is the user preference on each of the plurality of items, p u≧3 is a probability when the usefulness of the user behavior information is greater than or equal to 3, and p u<3 is a probability when the usefulness of the user behavior information is smaller than 3.
6. The computer program product within the computer readable medium of claim 5 , wherein the creating of the recommendation tree in the user recommendation system comprises the steps of:
forming a hierarchy level by arranging the plurality of items in a descending order of the user preference in the user recommendation system; and
arranging the elements in a descending order of the conditional probability value, the elements being included in the plurality of items in the user recommendation system.
7. The computer program product within the computer readable medium of claim 5 , wherein the recommending of the product to the user based on the recommendation tree in the user recommendation system comprises the steps of:
calculating a similarity between the user preference and the product in the user recommendation system; and
recommending the product to the user in a descending order of the similarity in the user recommendation system.
8. The computer program product within the computer readable medium of claim 7 , wherein the calculating of the similarity in the user recommendation system comprises the steps of:
computing a unit similarity by multiplying the preference of a hierarchy level and the conditional probability value of the element, the element corresponding to the product disposed in the hierarchy level in a user recommendation system; and
computing the similarity by adding the unit similarities in a user recommendation system.