Automatic abstract factuality enhancement method based on syntactic structure deletion

• Main Authors: HU DINGXIN, ZHOU YANQUAN, ZHANG XINGYUE, LI YIYANG, LI YUZE, LI LEI
• Format: Patent
• Language: Chinese; English
• Published: 31.10.2023
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses an automatic abstract factuality enhancement method based on syntactic structure deletion, and relates to a big data resource service and data mining technology. The method comprises the following steps: analyzing a to-be-corrected abstract text grammatical structure to obtain a corresponding grammatical dependency tree; performing post-order traversal on the grammar dependency tree, trying to prune the sub-tree according to the grammar dependency relationship to obtain a candidate abstract, and restoring the pruned sub-tree before accessing the next sub-tree in order; a plurality of fact evaluation index models are deployed in a server in advance; and calling a plurality of fact evaluation index models on the server in real time, performing factuality evaluation on each candidate abstract, preferably, when the next abstract is superior to the previous abstract in all fact evaluation indexes, considering that the next abstract is better, and outputting the optimal abstract. According t