Model-agnostic visualizations using linear programming approximation
AssigneeVerint Americas Inc.
InventorsAbhinav Aggarwal, Ian Roy Beaver, Cynthia Freeman, Jonathan Patrick Merriman
Life of this patent
Estimating the term…
Abstract
A method of determining influence of language elements in script to an overall classification of the script includes performing a sentiment analysis of the language elements. In some instances, for example, in a conversation, the language elements and turns within the conversation (e.g., in a chat bot) are analyzed for their influence in escalation or non-escalation of the conversation to a higher level of resolution, e.g., to a human representative or manager.
Claims25
A non-transitory computer readable storage medium having stored therein software executable to perform steps comprising:
- performing a perturbation of textual language elements by removing a subset of the textual language elements;
- analyzing a classification resulting from the perturbation of the textual language elements to determine if the removed subset of the language elements causes a change in classifier outcome;
- reporting that the subset is important to the classification if there is a change in the classifier outcome;
- given the conversation =(T1, . . . , TN), determining a weight wi>0 for each of the turns Ti such that given any perturbation P of C, a linear combination of the weights of the turns in P approximates a probability of escalation (CLASS(P)), and wherein the reporting comprises indicating that there is a probability of escalation;
- and
- displaying on a graphical user interface a visual scale indicative of the importance of the textual language elements to the classifier outcome, the visual scale comprising a color indicator, wherein color and intensity of the color indicator reflects relative importance of the classifier outcome for each of the textual language elements.
A graphical user interface, comprising:
- a visual representation of a language classifier outcome, the classifier outcome resulting from performing a perturbation of textual language elements in a script by removing a subset of the textual language elements and analyzing a classification resulting from the perturbation of the textual language elements to determine if the removed subset of the textual language elements causes a change in classifier outcome;
- and
- the visual representation comprising a visual scale indicative of the importance of the textual language elements to the classifier outcome, the visual scale comprising a color indicator, wherein color and intensity of the color indicator reflects relative importance of the classifier outcome for each of the textual language elements;
- wherein, given a conversation =(T1, . . . , TN), determining a weight wi>0 for each of the turns Ti such that given any perturbation P of C, a linear combination of the weights of the turns in P approximates a probability of escalation (CLASS(P)), and wherein the reporting comprises indicating that there is a probability of escalation.
A method of determining influence of language elements in script to an overall classification of the script, the language elements including a plurality of turns, the method including includes:
- performing a perturbation of the language elements by removing a subset of the language elements;
- analyzing the resulting classification to determine if the removed subset of the language elements causes a change in classifier outcome;
- reporting that the subset is important to the classification if there is a change in the classifier outcome;
- given a conversation =(T1, . . . , TN), determining a weight wi>0 for each of the turns Ti such that given any perturbation P of C, a linear combination of the weights of the turns in P approximates a probability of escalation (CLASS(P)), and wherein the reporting comprises indicating that there is a probability of escalation;
- and
- displaying on a graphical user interface a visual scale indicative of the importance of the language elements to the classifier outcome, the visual scale comprising a color indicator, wherein color and intensity of the color indicator reflects relative importance of the classifier outcome for each of the language elements.
Dependencies are inferred from the claim text.