Research Interest In Intelligent Router Recommendation Algorithm

The age of Internet produce vast amounts of data every day, the router as the network traffic node, has played an important role in detection and control of the exchanging information, and is important for the users’ experience at the same time. In the current network environment, the users’ online behavior are often interferenced by the mass advertisement. But the traditional router is often used to store and transmit information, do not have the effect on the network traffic hub entirely.Users often want to get the personalized recommendation rather than the mass advertisement. Now the current intelligent router do not have the function that giving the users their favourite recommendation, and the existing advertisement recommendation techniques are almost group recommendation, do not have the target. So this thesis has designed a kind of interest-recommendation algorithm based on the user’s behavior through the data mining method on the intelligent router, and digged the user’s data to realize the function of the interest-recommendation on users with personalized recommendation. So when the users access to the Internet through the intelligent router, they can get the most suitable interest recommendation and advertising links. In this thesis, the main work is as follows:1. Obtaining the original data coming from the user, using the data interface upload the original data to the cloud server, then analyze the data on the cloud server;2. Improveing the How Net semantics similarity algorithm, adding the depth of the sememe and the density of the sememe into the algorithm to improve the accuracy of the semantic similarity between two words;3. Designed the keywords-set similarity algorithm, separation the keywords-set into some clusters, using the tracking window extract the users’ real-time interests, and divide them into long-term interests and short-term interests;4. According to the users’ interests, calculate the similarity between users, and when two users have the high similarity, recommend their different interest to each other;5. Using MATLAB to simulate the interest-recommendation algorithm designed by this thesis, collect real data from 10 users, and divided them into training set and validation set. Determine the threshold parameter in the algorithm, predicting the user’s interest and the similar users according to the algorithm and proved the result by the validation set, consistent with the real situation.At last, the interest-recommendation algorithm applied and implemented on the router. Different users obtain their interest from the recommendation algorithm, and obtain the interest by their similar users, and at last verifiy the feasibility of the algorithm designed by this thesis.