Research On IP Address Location Technology Based On Semantic Understanding And Deep Neural Network

With the development of the Internet,IP address positioning technology is playing an increasingly important role in today's society.Advertisers can obtain the user's location information by locating the IP address used by the user to perform accurate advertising.Operators can quickly locate network faults by locating the IP addresses used by network equipment to determine where a network fault is located.Since the development of IP address positioning technology,there have been many excellent IP positioning products and positioning algorithms,but these products have their own limitations.Among them,how to use more data to improve the accuracy of IP address positioning and how to use delay to improve the positioning accuracy have become hot topics in current research.Therefore,this thesis conducts research from these two aspects.Specifically completes the following work:1.Aiming at the problem that the data used for IP address location is relatively limited,this thesis proposes to use IP social data published on the Internet by IP addresses to locate IP addresses and designs a method for IP address location based on semantic understanding.Firstly,fusioning matrix and edit distance to calculate the similarity between the word segmentation and address words to identify the address word segmentation in the sentence.At the same time,using the similarity to design an address-sentence filtering algorithm to screen out address sentences containing address words.In order to obtain the final location of the target IP,this thesis designs the sentence processing rules and final position determination method.The experimental results show that our method can effectively locate IP addresses from the level of social data,and the positioning accuracy and positioning accuracy are higher than the database search results.2.Aiming at the problem of how to use time delay to improve the positioning accuracy,this thesis proposes a method of IP address positioning based on deep neural networks.Firstly,building a deep neural network model and train it to obtain a positioning model that can calculate the latitude and longitude of the target IP location through time delay.In order to reduce the positioning error,this thesis proposes an improved algorithm based on adaptive weighted K-nearest neighbor(WKNN)to add constraint intervals and reference coordinates to the positioning coordinates of the deep neural network.Finally,the gradient approximation method is used to modify the initial latitude and longitude coordinates calculated by the deep neural network.The experimental results show that the positioning accuracy of the positioning method in this thesis is higher than the existing IP positioning method,and the stability is better.