A CORE SENTENCE EXTRACTION METHOD BASED ON A DEEP LEARNING ALGORITHM

摘要

A method for extracting a core sentence based on a deep learning algorithm is disclosed. Separating the sentences of the document; Sampling a sentence; Generating supervised learning data from the sampled data; Creating a lookup table for the sentence; Extracting a core sentence probability for each sentence using a lookup table and inputting a supervised neural network model; And extracting an upper sentence as a probability value for each sentence of the neural network model. Therefore, it is possible to extract the core sentence from the document, has the advantage of securing punctuation exceptions, can perform a pre-processing step to extract the core sentence by sampling a large number of sentences, and prepares a deep set of learning data sets Algorithms can be performed, a lookup table can be generated from sentences and input to a deep learning algorithm, and key sentences can be extracted from documents by inferring the possibility of key sentences.