Analysis of Sentence Boundary of the Host's Spoken Language Based on Semantic Orientation Pointwise Mutual Information Algorithm
Xueyujie Wang · 2020
Detecting dependencies in spontaneous speech is much more challenging than in written texts. The biggest problem is the unclear boundary of sentences. In this paper, an algorithm that relies on the semantic orientation pointwise mutual information is proposed to achieve an F-measure of 84.9, which has improved the accuracy of sentence boundary detection. Based on the sentence boundaries detected automatically, the accuracy of dependency structure analysis is also improved from 75.2% to 77.2%. Through the interactive use of dependency structures and sentence boundaries detected automatically, the nine-tenths of dependency structure analysis and sentence boundary detection has also been improved.