Method and System of Proposition of Ancient Poems

Abstract

The embodiment of the present invention provides a method for propositioning ancient poems. The method includes: extracting at least one keyword of each poem in a poem training set based on a knowledge graph, determining it as a semantic feature of each poem, using a neural network to model the semantic feature of each poem, and obtaining the input as the semantic feature of the poem, and the output is Poetry model of other semantic features with similar semantics; use knowledge graph to model the poetry training set to obtain a structured model of the poetry information in the poem training set; based on the semantic features of each poem, associate the poetry model and the structured model, and predict the The question type is input into the associated poetry model and the structured model, and the question corresponding to the preset question type is output. The embodiment of the present invention also provides an ancient poetry proposition system. The embodiment of the present invention uses the semantic features of poems for modeling, and can distinguish related words of poems. By structuring the information of poems, rich information of poems can be obtained, and topics with higher quality can be generated.

Description

technical field

[0001] The invention relates to the field of natural language processing, and in particular, to a method and system for propositioning ancient poems. Background technique

[0002] With the application of natural language processing technology in various fields, there have been many methods for making propositional systems in the fields of education and question answering. As one of the symbols of Chinese culture, ancient poetry has always been a difficulty in natural language processing because it involves difficult and complex tasks such as pronunciation, rhythm, abstract intention, and allusions. Now some experts have used neural network technology to semantically understand ancient poems and extract abstract features, and then derive products such as reading comprehension, propositions, and generated verses. Based on the shortcomings of the neural network itself, the above tasks can still only achieve simple and basic functions.

[0003] In the process o...

Images