Devices, systems, and methods for intelligent determination of conversational intent
Disclosed herein are devices, systems, and computer-implemented methods for intelligent conversational intent detection. Example methods include acquiring a conversational transcript input that is requested for intent detection, inputting the conversational transcript input into a model configured to decipher a conversational intent segment, and returning to a user the conversational intent segment. The conversational transcript input can include one or more conversational transcript segments. The conversational intent segment can correspond to which of the one or more conversational transcript segments is likely to indicate a conversational intent of the conversational transcript input based on the intent detection.
1 . A computer-implemented method for intelligent conversational intent detection, the computer-implemented method comprising:
acquiring a conversational transcript input that is requested for intent detection, the conversational transcript input comprising one or more conversational transcript segments;
inputting the conversational transcript input into a model configured to:
decipher a conversational intent segment corresponding to which of the one or more conversational transcript segments is likely to indicate a conversational intent of the conversational transcript input based on the intent detection;
index where the conversational intent segment is in the conversational transcript input; and
return to the user the index corresponding to the conversational intent segment; and
returning to a user at least one of the conversational intent segment,
wherein at least one of:
the model is an intelligent natural language processing model comprising a text-to-text transfer transformer, and wherein the model is further configured to index a location of the conversational intent segment in the conversational transcript input, assign a probability value to the conversational intent segment, and adjust the probability value based on at least one of a user-specific parameter and a session-specific parameter derived from prior user interaction history or session metadata indicating conversational drift; and
the conversational transcript input comprises a one or more conversational transcripts such that the conversational intent segment indicates an aggregate conversational intent of the plurality of conversational transcripts, and wherein the model is further configured to: (i) generate a set of counterfactual conversational segments from the conversational transcript input that are semantically similar to the conversational intent segment and preserve contextual relationships from the original transcript and (ii) assign sentiment-based scores to each of the counterfactual conversational segments such that the conversational intent segment is returned based on a comparative ranking of the sentiment-based scores.
2 . The computer-implemented method of claim 1 , further comprising parsing the conversational transcript input into the one or more conversational transcript segments.
3 . The computer-implemented method of claim 2 , wherein the model is further configured to parse the conversational transcript input into the one or more conversational transcript segments, and wherein parsing the conversational transcript input into the one or more conversational transcript segments is performed by the model.
4 . The computer-implemented method of claim 2 , wherein the one or more conversational transcript segments comprises one or more transcript utterances, grouped transcript utterances, or transcript utterance portions.
5 . The computer-implemented method of claim 1 , wherein the conversational intent segment comprises the one or more conversational transcript segments that is most likely to indicate the conversational intent.
6 . The computer-implemented method of claim 1 , wherein the model is further configured to index where the conversational intent segment is in the conversational transcript input, and wherein the computer-implemented method further comprises returning to the user the index corresponding to the conversational intent segment.
7 . The computer-implemented method of claim 1 , wherein the model is further configured to assign a probability value to the conversational intent segment, the probability value corresponding to a likelihood that the conversational intent segment indicates the conversational intent of the conversational transcript input, and wherein the conversational intent segment is the most likely intent.
8 . The computer-implemented method of claim 1 , wherein the one or more conversational transcripts comprises a plurality of conversational transcripts.
9 . The computer implemented method of claim 8 , wherein the conversational transcript input comprises a plurality of conversational transcripts, and wherein the computer-implemented method further comprises parsing the plurality of conversational transcripts into individual conversational transcripts such that the conversational intent segment indicates the conversational intent of each of the individual conversational transcripts.
10 . The computer implemented method of claim 9 , wherein the model is further configured to aggregate each of the conversational intent segments into intent categories that are either provided to the model or determined by the model itself.
11 . A data processing system for intelligent intent detection, the data processing system comprising:
a memory for storing one or more modules;
a processor configured to access the memory and to process:
an acquisition module that is configured to acquire a conversational transcript input that is requested for the intent detection, the conversational transcript input comprising one or more conversational transcript segments;
an input module that is configured to input the conversational transcript input into a model configured to decipher a conversational intent segment, the conversational intent segment corresponding to which of the one or more conversational transcript segments is likely to indicate a conversational intent of the conversational transcript input based on the intent detection; and
a results module that is configured to return to a user the conversational intent segment,
wherein the model is an intelligent natural language processing model comprising a text-to-text transfer transformer, and
wherein the model is further configured to index a location of the conversational intent segment in the conversational transcript input, assign a probability value to the conversational intent segment, and adjust the probability value based on at least one of a user-specific parameter and a session-specific parameter derived from prior user interaction history or session metadata indicating conversational drift.
12 . The data processing system of claim 11 , wherein the data processing system is a server running a private cloud platform or a multi-tenant platform.
13 . The data processing system of claim 11 , wherein the conversational intent segment comprises the one or more conversational transcript segments that is most likely to indicate the conversational intent.
14 . The data processing system of claim 13 , wherein the results module is further configured to return to the candidate conversational intent segments, each of which is less or equally likely to indicate the conversational intent, and to return to the user an ordered list comprising the conversational intent segment and the candidate conversational intent segments in order of likelihood that the conversational intent is indicated therein.
15 . The data processing system of claim 11 , further comprising a display that is configured to present the conversational intent segment from the results module.
16 . A non-transitory computer-readable medium that stores instructions that, when executed by one or more processors, causes the one or more processors to:
acquiring a conversational transcript input that is requested for intent detection, the conversational transcript input comprising one or more conversational transcript segments;
inputting the conversational transcript input into a model configured to decipher a conversational intent segment, the conversational intent segment corresponding to which of the one or more conversational transcript segments is likely to indicate a conversational intent of the conversational transcript input based on the intent detection; and
returning to a user the conversational intent segment,
wherein the model is further configured to:
generate a set of counterfactual conversational segments from the conversational transcript input that are semantically similar to the conversational intent segment and preserve contextual relationships from the original transcript, and
assign sentiment-based scores to each of the counterfactual conversational segments such that the conversational intent segment is returned based on a comparative ranking of the sentiment-based scores.
17 . The non-transitory computer readable medium of claim 16 , wherein the instructions, when executed by the one or more processors, further cause the one or more processors to parse the conversational transcript input into the one or more conversational transcript segments, wherein the conversational transcript input comprises one or more conversational transcripts, and wherein the model is an intelligent natural language processing model comprising a text-to-text transfer transformer.
18 . The non-transitory computer readable medium of claim 16 , wherein the model is further configured to assign a probability value to the conversational intent segment, the probability value corresponding to a likelihood that the conversational intent segment indicates the conversational intent of the conversational transcript input, and wherein the conversational intent segment comprises the one or more conversational transcript segments that is most likely to indicate the conversational intent.