The problem considered in this paper relates to identification of trends in a given area based on analysis of Twitter messages. The approaches currently used for Twitter trends detection are based on n-grams. We propose another approach of trend detection based on identifying trend as grammatical relation and perform the identification of trending relations on the basis of their frequency change dynamics. This paper describes our method, which evaluates grammatical relations in a flow of messages on a particular subject taking into consideration both their frequency and semantic similarity among the pairs of relations. We conducted experiments to compare the outcomes provided by our method with the trends detected by conventional Twitter algorithms. The results confirmed the effectiveness of our method. The trends identified from the application of our method are easier for human interpretation.
展开▼