A Similarity Calculation Model of Weak Link Web Pages Based on Keyword Location Influence

Abstract

Weakly linked web pages are a group of related pages with weak connections. With the advent of the self-media era, the related information links between various platforms have become weaker, the calculation of the similarity of weakly linked web pages is becoming more and more important. The current method for calculating the similarity of weakly linked webpages is to combine the similarity of webpage content and the relevance of links, and then simply add the results. There is a large deviation in the calculation results of this method. To this end, this paper proposes a similarity calculation model for weak link webpages based on the influence of keyword position (WIWKLI). First, we start with the division of web pages, through URL matching deduplication, keyword similarity calculation to form a keyword sequence of web pages, then through the improvement of the RNN model to calculate the positional influence of keywords, so as to realize the precise definition of the webpage and form a link pseudo-network between related webpages accordingly, Finally, the similarity between web pages is calculated through the pseudo-network. By applying this calculation model to the focused crawler experiment, Experimental results show that the method proposed in this paper can calculate the similarity of weakly linked web pages, and the accuracy and page relevance have been improved to a certain extent.