Text processing method, system and equipment and medium

Abstract

The invention provides a text processing method, system and equipment and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps of recognizing obtained audio data, and generating an original recognition text; combining the original recognition texts, carrying out serial number marking on the combined texts, and recording a starting serial number and an ending serial number of the original recognition texts in the combined texts to obtain a serial number interval; associating all the sequence number intervals to form a target dictionary; performing target text recognition on the intercepted text, determining interval positions of a start sequence number and an end sequence number of the intercepted text in a target dictionary, and determining the number of text paragraphs of the intercepted text crossing the original recognition text; and comparing the number of the text paragraphs with a preset threshold value, and performing paragraph segmentation on the intercepted text according to a comparison result to obtain a paragraph segmentation result. According to the invention, the obtained paragraph segmentation result cannot be different from the original recognition text, segmentation is completed only once, and the segmentation result does not need to be confirmed again.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a text processing method, system, equipment and medium. Background technique [0002] In some consulting project dialogues, algorithmic models or regular engines are generally used to judge whether there are illegal sentences in the dialogue between customer service (such as manual customer service, intelligent customer service) and customers. Since customer service and customers generally complete the corresponding dialogue consultation through voice dialogue or text dialogue, for voice dialogue, the conventional recognition method is to convert the voice formed by the dialogue between customer service and customer into voice text, and then Recognize speech text to determine if there are any offending sentences in the conversation between the agent and the customer. However, when the speech formed by the dialogue between the customer service and the cust...

AI Technical Summary

Problems solved by technology

[0003] However, the inventors found that when segmenting the illegal sentence recognition results in the prior art, there may be wrong or inefficient segmentation. For example, there may be the following types of incorrect or inefficient segmentation: Situations: (1) When multiple partially overlapping original recognition texts appear in the illegal sentence recognition results, if the illegal sentence recognition results are segmented directly according to the partially overlapping original recognition texts, it may lead to segmentation Subsequent paragraphs differ from the original recognized text

(2) When segmenting the illegal sentence recognition results into paragraphs, if there is a substring relationship, only keep the longest original recognition text to segment the illegal sentence recognition results, and the segmented paragraphs may also There are differences from the original recognition text

(3) By recording the paragraph numbers of multiple original recognition texts, and then segmenting the recognition results of illegal sentences according to all the paragraphs with the most consecutive numbers, although this method of paragraph segmentation can get the correct original recognition text, but in the presence of For repeated paragraphs or texts, the confirmation of the specific serial number will be performed again from multiple original recognition texts, so that there will be an additional layer of loop operation when recording the serial number, which will increase the amount of calculation when judging the most consecutive serial numbers later, resulting in the overall Not very efficient

Images