Seat call content analysis method

Abstract

The invention provides a seat call content analysis method. The method comprises the following steps: acquiring voice data of a conversation performed when a seat calls out or accesses the conversation; performing text transcription processing on the voice data to obtain a voice text; calling a dialogue analysis model to perform call overall analysis and local information extraction on the voice text, the call overall analysis being used for identifying at least one dialogue description attribute corresponding to the voice text, and the local information extraction being used for extracting atleast one key segment in the voice text; and creating the dialogue description attribute and an indexing mapping between the key fragment and the voice text. In addition, the invention also providesa corresponding computer medium. Compared with the prior art, the evaluation reference can be provided for the service handling effect of the agent call more comprehensively, so that an operator of acall center can know the agent communication performance, customer requirements and service hotspots more objectively and comprehensively.

Description

technical field [0001] The invention relates to the field of intelligent traffic processing in a call center, in particular to an analysis method for seat conversation content. Background technique [0002] At present, the call center has become an important way for enterprises to provide online comprehensive business information. The agents of the call center carry out outbound or inbound traffic services according to the needs of the enterprise to accept customer feedback, consultation and suggestions, or for The company's products conduct market research, telephone sales, after-sales tracking and other services. [0003] Generally, it is considered to record and evaluate the handling effect of the business conducted by the agent during the call, so that the call center can perform follow-up steps such as business management, quality assessment, and hot spot analysis according to the recorded handling effect, and further improve the service quality. The above-mentioned ha...

AI Technical Summary

Problems solved by technology

[0004] Some solutions for recording and evaluating the above-mentioned handling effects in the prior art are: after the agent’s call ends, manually select a defined call completion code in combination with the customer’s feedback data to mark the handling effect, so that the handling effect is consistent with the selected call completion code. Corresponding to the completion code, it is used to mark the type of handling effect, and realize the simple classification of communication history by calling the completion code. The shortcomings of this solution are also obvious. The method of relying on the agent to manually select the type of handling effect has obvious advantages. Subjectivity and limitations, and only evaluated the processing effect from the perspective of results, and failed to objectively evaluate the processing effect from the perspective of process

[0005] Other solutions for recording and evaluating the handling effect in the prior art are: after the agent call is over, perform simple clustering statistics on the partial specific sentence content of the agent call according to the selected keywords to obtain statistical results, and pass the statistics To evaluate the handling effect based on the results, this solution still has obvious defects, because the selection of keywords is difficult to completely cover various possibilities that may arise in daily conversations, and relying on this keyword can only count the local details of the agent's calls. sentence, it is impossible to realize the overall cognition of the agent's call, which leads to the unsatisfactory accuracy of the evaluation of the handling effect

Images