Data-driven dialogue enabled self-help systems

A method for configuring an automated self-help system based on prior interactions between a plurality of customers and a plurality of agents of a contact center, the method comprising:

• recognizing, by a processor, speech loaded from a call recording storage medium in the prior interactions between customers and agents to generate recognized text, the recognized text comprising a plurality of phrases, the phrases being classified into a plurality of clusters;
• extracting, automatically by the processor, a plurality of sequences of clusters, each of the sequences of clusters corresponding to the phrases of one of the prior interactions;
• filtering, automatically by the processor, the sequences of clusters based on a criterion, the criterion comprising at least one of:
• success rate;
• frequency of use;
• or
• interaction length;
• mining, automatically by the processor, a preliminary dialogue tree from the sequences of clusters;
• invoking configuration of the automated self-help system based on the preliminary dialogue tree;
• and
• outputting a dialogue tree, automatically generated based on the preliminary dialogue tree, for configuring the automated self-help system,
• wherein the mining the preliminary dialogue tree from the sequences of clusters and a current prefix comprises:
• generating sequences of cluster identifiers;
• identifying frequent clusters of the clusters from the sequences of cluster identifiers;
• and
• matching prefixes of the sequences to generate the preliminary dialogue tree, and
• wherein the matching prefixes of the sequences comprises, based on the current prefix, a projected database for the current prefix, and an input tree, for each cluster in the frequent clusters:
• concatenating the current prefix with the cluster to generate an appended prefix;
• building a projected database for the appended prefix;
• creating a subtree by recursively repeating the process using the appended prefix, the projected database for the appended prefix, and the input tree;
• and
• extending the input tree by connecting the subtree as a child to a current parent node corresponding to the current prefix.

A system comprising

• a processor;
• and
• a memory, wherein the memory stores instructions that, when executed by the processor, cause the processor to:
• recognize speech loaded from a call recording storage medium in a plurality of prior interactions between a plurality of customers and a plurality of agents of a contact center to generate recognized text, the recognized text comprising a plurality of phrases, the phrases being classified into a plurality of clusters, the plurality of clusters comprising agent clusters corresponding to agent speech and customer clusters corresponding to customer speech;
• extract, automatically, a plurality of sequences of clusters, each of the sequences of clusters corresponding to the phrases of one of the prior interactions;
• filter, automatically, the sequences of clusters based on a criterion, the criterion comprising at least one of:
• success rate;
• frequency of use;
• or
• interaction length;
• mine, automatically, a preliminary dialogue tree from the sequences of clusters by matching prefixes of the sequences to generate the preliminary dialogue tree, the preliminary dialogue tree comprising alternating layers of agent clusters and customer clusters;
• invoke configuration of an automated self-help system based on the preliminary dialogue tree;
• and
• output a dialogue tree, automatically generated based on the preliminary dialogue tree, for configuring the automated self-help system, the dialogue tree being customized for the contact center based on the prior interactions between the customers and agents of the contact center, the automated self-help system being configured to:
• present one or more options to a customer in accordance with a current node in the dialogue tree;
• receive an input from a customer;
• route the customer to a new node in the dialogue tree in accordance with the input;
• and
• generate, automatically, a response in accordance with the input and the new node.

A system comprising

• a processor;
• and
• a memory, wherein the memory stores instructions that, when executed by the processor, cause the processor to:
• recognize speech loaded from a call recording storage medium in a plurality of prior interactions between a plurality of customers and a plurality of agents of a contact center to generate recognized text, the recognized text comprising a plurality of phrases, the phrases being classified into a plurality of clusters;
• extract, automatically, a plurality of sequences of clusters, each of the sequences of clusters corresponding to the phrases of one of the prior interactions;
• filter, automatically, the sequences of clusters based on a criterion, the criterion comprising at least one of:
• success rate;
• frequency of use;
• or
• interaction length;
• mine, automatically, a preliminary dialogue tree from the sequences of clusters;
• invoke configuration of an automated self-help system based on the preliminary dialogue tree;
• and
• output a dialogue tree, automatically generated based on the preliminary dialogue tree, for configuring the automated self-help system, the dialogue tree being customized for the contact center based on the prior interactions between the customers and agents of the contact center,
• wherein the instructions that cause the processor to mine the preliminary dialogue tree from the sequences of clusters and a current prefix include instructions that, when executed by the processor, cause the processor to:
• generate sequences of cluster identifiers;
• identify frequent clusters of the clusters from the sequences of cluster identifiers;
• and
• match prefixes of the sequences to generate the preliminary dialogue tree
• wherein the instructions that cause the processor to match prefixes of the sequences includes instructions that, based on the current prefix, a projected database for the current prefix, and an input tree, for each cluster in the frequent clusters, cause the processor to:
• concatenate the current prefix with the cluster to generate an appended prefix;
• build a projected database for the appended prefix;
• create a subtree by recursively repeating the process using the appended prefix, the projected database for the appended prefix, and the input tree;
• and
• extend the input tree by connecting the subtree as a child to a current parent node corresponding to the current prefix.