Webpage pre-reading method, apparatus and smart terminal
A method is provided for webpage pre-reading. The method includes obtaining visiting information during a period of time for multiple webpages of each user from a plurality of users including a current user; determining respective similarity values between the current user and the plurality of users based on the visiting information; determining one or more most-similar users based on the similarity values; and when the current user is browsing a current webpage, predicting a target webpage to be accessed by the most-similar users as a target webpage for the current user and pre-reading the target webpage.
1. A method for webpage pre-reading, comprising:
obtaining visiting information during a period of time for multiple webpages of each user from a plurality of users including a current user;
determining a respective similarity values between the current user and each of the plurality of users based on the visiting information;
determining one or more most-similar users based on the respective similarity value;
determining a recommendation probability of a clicking path based on each of the one or more most-similar users, by calculating a product of the respective similarity value of each of the one or more most-similar users and a number of clicks from a first webpage to a second webpage and, wherein the first webpage and the second webpage forms the clicking path; and
when the current user is browsing a current webpage, predicting a target webpage to be accessed by the one or more most-similar users, based on the recommendation probability, as a target webpage for the current user and pre-reading the target webpage.
2. The method of claim 1 , further comprising:
when obtaining the visiting information during the period of time for the multiple webpages of each user from the plurality of users including the current user, performing a data cleaning process to screen the visiting information of each webpage.
3. The method of claim 1 , wherein determining the respective similarity value further includes:
calculating the respective similarity value between the current user and each of the plurality of users using a cosine similarity calculation formula.
4. The method of claim 3 , further comprising:
before calculating the similarity value, filtering users having a webpage visiting number that is less than a webpage visiting threshold number from the plurality of users,
wherein the webpage visiting number is determined as a lower limit of a confidence interval calculated by using Wilson interval formula.
5. The method of claim 1 , further comprising:
generating an inverted list using a same webpage visited by multiple users as a common characteristic; and
determining a similarity value between the current user and each of the plurality of users in the inverted list using a cosine similarity calculation formula.
6. The method of claim 5 , further comprising:
when a total number of users included in any inverted list exceeds a user threshold value, removing the exceeding inverted list.
7. The method of claim 1 , wherein predicting the target webpage further includes:
calculating recommendation probabilities of a plurality of clicking paths based on the one or more most-similar users, wherein each of the plurality of clicking paths is from the current webpage;
generating a pre-reading list including a subset of the plurality of clicking paths according to the recommendation probabilities of the plurality of clicking paths; and
searching the pre-reading list based on the current webpage to determine the target webpage to be accessed.
8. The method of claim 7 , wherein calculating the recommendation probabilities further includes:
obtaining a number of clicks by a most-similar user of the one or more most-similar users from the first webpage to the second webpage, the first webpage and the second webpage forming the clicking path; and
calculating a product of the number of clicks and a similarity value between the most-similar user and the current user as a recommendation probability of the clicking path.
9. The method of claim 8 , further comprising:
when multiple most-similar users click from the first webpage to the second webpage, calculating a sum of the recommendation probability of the click path of the multiple most-similar users as the recommendation probability of the clicking path.
10. The method of claim 8 , further comprising:
filtering any clicking path having a number of clicks less than a clicking number threshold.
11. The method of claim 7 , further comprising:
ranking the subset of the plurality of clicking paths in the pre-reading list in a descending order based on the recommendation probabilities of the plurality of clicking paths,
wherein pre-reading the target webpage further includes:
pre-reading a target webpage corresponding to a clicking path having a largest recommendation probability; or
pre-reading multiple target webpages corresponding to multiple clicking paths ranked in order based on their recommendation probabilities.
12. The method of claim 7 , further comprising:
filtering any clicking path having a recommendation probability less than a recommendation probability threshold.
13. An apparatus for webpage pre-reading, comprising:
an information obtaining module configured for obtaining visiting information during a period of time for multiple webpages of each user from a plurality of users including a current user;
a determination module configured for determining a respective similarity value between the current user and each of the plurality of users based on the visiting information, and determining one or more most-similar users based on the respective similarity value;
a calculating and generating module configured for determining a recommendation probability of a clicking path based on each of the one or more most-similar users, by calculating a product of the respective similarity value of each of the one or more most-similar users and a number of clicks from a first webpage to a second webpage and, wherein the first webpage and the second webpage forms the clicking path;
an analysis module configured for, when the current user is browsing a current webpage, predicting a target webpage to be accessed by the one or more most-similar users as a target webpage for the current user; and
a first pre-reading module configured for pre-reading the target webpage.
14. The apparatus of claim 13 , wherein the information obtaining module includes:
a preprocessing module configured for performing a data cleaning process to screen the visiting information of each webpage.
15. The apparatus of claim 13 , wherein the determination module includes:
a filter module configured for filtering a user corresponding to a webpage visiting number that is less than a webpage visiting threshold number; and
a similarity calculation module configured for using a cosine similarity calculation formula to calculate the similarity value between the current user and each of the plurality of users.
16. The apparatus of claim 13 , wherein the determination module further includes:
an inverted list generating module configured for generating an inverted list including a subset of the plurality of users that have visited a same webpage visited by the current user; and
a most-similar user determination module configured for determining a similarity value between the current user and each user in the inverted list.
17. The apparatus of claim 13 , wherein the calculating and generating module is further configured for:
calculating recommendation probabilities of a plurality of clicking paths based on the one or more most-similar users, wherein each clicking path is from the current webpage; and
generating a pre-reading list including a subset of the plurality of clicking paths according to the recommendation probabilities of the plurality of clicking paths.
18. The apparatus of claim 17 , wherein the calculating and generating module is further configured for:
obtaining a number of clicks by a most-similar user of the one or more most-similar users from the first webpage to the second webpage, the first webpage and the second webpage forming the clicking path;
calculating a product of the number of clicks and a similarity value between the most-similar user and the current user as a recommendation probability of the clicking path; and
when multiple most-similar users click from the first webpage to the second webpage, calculating a sum of the recommendation probability of the click path of the multiple most-similar users as the recommendation probability of the clicking path.
19. The apparatus of claim 17 , further comprising a second pre-reading module configured for:
searching the pre-reading list based on the current webpage to determine the target webpage to be accessed; and
pre-reading data of the target webpage to be accessed.
20. The apparatus of claim 19 , the second pre-reading module being further configured for:
ranking the subset of the plurality of clicking paths in the pre-reading list in a descending order based on the recommendation probabilities of the plurality of clicking paths,
wherein pre-reading the target webpage further includes:
pre-reading a target webpage corresponding to a clicking path having a largest recommendation probability; or
pre-reading multiple target webpages corresponding to multiple clicking paths ranked in order based on their recommendation probabilities.