nlp - Does the order of word appearance matters in keyword extraction using Textrank? -

i using textrank extract keywords documents. figure order of words appearance has influence on final weights of words' importance, not straight-forward understand.

let's using undirected graph, windows size 2, keyword extraction task sentence [w1, w2, w3, w4]. when constructing graph, add edges such (w1, w2), (w2, w1), (w2, w3), (w3, w2)...in case, there 2 edges word w1, 4 edges w2. given this, without further computation, w2 has better chance keyword w1, , due order of appearance. me not make sense. there method factor avoided?