2 results about "Language technology" patented technology

This invention provides a dialogue processing method, apparatus, and computer-readable storage medium, belonging to the field of natural language processing technology. The dialogue processing method includes acquiring multiple sets of dialogue streams; determining at least one topic boundary based on the semantics of the multiple sets of dialogue streams; segmenting all dialogue streams into at least two topic segments based on each topic boundary; assigning dynamic weights to memory items in each topic segment, wherein each memory item stores at least one piece of memory information; the dynamic weights are determined at least based on the timeliness of the memory item, its relevance to the current query, and its coherence with the current topic; and, in response to the current dialogue, acquiring multiple candidate memory items from all topic segments, and, based on the dynamic weight of each candidate memory item, filtering and outputting a set of memory items that meets a target length from the multiple candidate memory items.

This invention discloses a semantically anchored, approximate contrastive learning method and system, relating to the fields of vision and language technology. It involves acquiring image and text data; extracting image embedding vectors and text embedding vectors using an image encoder and a text encoder, respectively; updating dynamic image and text queues through cross-modal momentum contrast and constructing a fused multimodal representation, which serves as a shared semantic anchor; progressively calibrating the image and text embedding vectors towards the shared semantic anchor within the contrastive learning framework to generate calibrated image and text embedding vectors; and performing image-text retrieval based on the calibrated image and text embedding vectors. This invention relaxes the strict requirement for complete semantic alignment of positive sample pairs, effectively mitigating cross-modal semantic bias.