Language model illusion suppression dynamic decoding method based on multi-dimensional contrast signal

By constructing a term path to organize word order and grammatical dependencies, and combining a dynamic comparison mechanism between entity reference time and knowledge node update time, the problems of semantic drift and logical fragmentation in existing technologies are solved, and semantic coherence and structural integrity of text generation are achieved.

CN121168670APending Publication Date: 2025-12-19火石创造科技有限公司
4 Cites 0 Cited by

Patent Information

Application Number
CN202511691832.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-18
Publication Date
2025-12-19

AI Technical Summary

Technical Problem

Existing technologies struggle to identify the dependency positions and expression order between terms in text generation tasks involving multi-turn action connections or cross-sentence structure tracking, leading to semantic drift and logical fragmentation, which affects content integrity and structural controllability.

Method used

By constructing term paths to sort out word order and grammatical dependencies, and combining entity reference time and knowledge node update time to establish a dynamic comparison mechanism, the system identifies repeated main clauses and structurally similar segments between sentences, selects language skeletons with stable features, and introduces time and action signals to guide the generation process, thereby enhancing semantic organization coherence, inter-sentence structural synergy, and content temporal consistency.

Benefits of technology

It effectively identifies and corrects semantic drift and logical fragmentation, improves the semantic coherence and structural integrity of text generation, and ensures the temporal consistency and controllability of content.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to the technical field of illusion suppression, in particular to a language model illusion suppression dynamic decoding method based on a multi-dimensional contrast signal, which comprises the following steps of: acquiring a language model input condition, extracting a word group connection relationship and promoting word item output, identifying a grammar sequence and an attachment mode, extracting an entity attribute and comparing reference time. And judging a semantic sequence, analyzing an action connection relationship, screening paths with the same structural form, expanding to next statement action construction, and introducing guidance to obtain a dynamic control signal set. According to the method, a word order and grammar dependency are sorted by constructing a lexical item path, a dynamic contrast mechanism is established in combination with entity reference time and knowledge node updating time, an action sequence is promoted by relying on verb phrases and conjunctions, inter-sentence repeated main sentences and structural convergence fragments are recognized, and a language skeleton with stable features is screened; time and action signals are introduced to guide the generation process, and semantic organization coherence, inter-sentence structure collaboration and content time sequence consistency are enhanced.
Need to check novelty before this filing date? Find Prior Art