An Experimental Method and Application of Metonymy Processing Based on Quantification of Text Context Information Level

Abstract

The invention relates to an experiment method and application for metonymy processing of hierarchical quantization based on textual context information. The method comprises following steps: determining metonymy target words and obtaining textual context information to be processed; cutting natural context represented by textual context information to be processed into multiple semantic units; accessing correlations of semantic units and semantics of metonymy target words and labeling an assessment result in order to set up a corpus; using synonyms, superordinates and others to replace the semantic units in order to carry out hierarchical processing of contexts for metonymy target words in order to form an operable context information condition; operating the context information condition and carrying out an metonymy processing experiment. The method can be applied to psychological experiments (including metaphors and ironies) of other unoriginal languages and especially used for conversation development systems of man-machine (robot-human) and provides a feasible method for finding out language structures, communicative contexts and semantic meanings on the basis of qualitative and quantitative researches.

Description

technical field [0001] The present invention relates to the fields of linguistic psychological cognition technology and artificial intelligence science, in particular to an experimental method and application of metonymy processing based on quantification of text context information levels. Background technique [0002] The focus of current language psychology experiments has shifted from single-word understanding (semantics-centered) to discourse understanding / pragmatic meaning under contextual conditions, because the meaning of discourse in the context has communicative meaning. At present, robots are still stuck in the era of "machines" rather than "humans", mainly because the problem of human-machine intelligent dialogue has not yet solved the key problem: after the human-machine conducts multiple conversations, the robot will fall into a paralyzed situation that is seriously lagging behind or even answering questions that are not asked. The root cause of the inability t...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the problem that the current psycholinguistics experiments and man-machine conversation intelligence systems cannot accurately measure the influence of a single contextual information on the pragmatic meaning, the present invention realizes the hierarchical manipulation of the contextual information, and invents a text-based contextual information Hierarchical quantitative metonymic processing experimental method and application to solve the experimental technical problems of psycholinguistics and the existing problems of the human-computer conversation development system

Images