Abstract: The present invention is a system and method for projective channel hopping within a customer engagement center (CEC) system. The CEC system receives a customer through a system entry point and receives, from the customer, a communication in a current customer service representative (CSR) channel in the CEC system. The CEC system assesses its ability to connect the customer with a CSR on the current CSR channel using a smart routing engine (SRE), which also assesses its current ability to connect the customer with a CSR on other equivalent CSR channels. Using the SRE, the CEC system compares its ability to connect the customer with a CSR on current and equivalent CSR channels to determine if the customer should remain on their current CSR channel or transfer to a new one. In the latter case, the CEC system offers the customer a chance to change channels or remain on their current channel.

Abstract: The present disclosure provides a sentence recommendation method and apparatus based on associated points of interest. The method includes: obtaining an input sentence from a user; extracting a keyword in the input sentence, and searching for a current point of interest matching the keyword in a preset point of interest association network determining a plurality of associated points of interest matching the current point of interest according to the preset point of interest association network, and filtering out an extension point of interest from the plurality of associated points of interest according to a preset filter strategy; and retrieving a first reply sentence and a second reply sentence from a preset corpus according to the current point of interest and the extension point of interest, and providing them to the user.

Abstract: A disclosed method includes receiving at least an indication of audio data generated based on sound captured by a microphone of a monitoring device. A set of audio event labels are generated by processing the received audio data with a model that includes audio recognition patterns. Each audio event label is generated by matching an audio pattern in the received audio data with an audio pattern of the audio recognition patterns. The method further includes identifying a hazard event type occurring within a time window of the time period by processing the set of audio event labels with a hazard detection model. In response to detecting the hazard event type, a hazard alert is generated and at least an indication of the hazard alert is communicated to an electronic device other than the monitoring device, which is authorized to receive communications from the monitoring device.

Abstract: The present invention relates to communications methods and apparatus for determining whether a received call originated from an artificial intelligence system and the call handling procedures to be used on the received call based on whether or not the call was determined to have been originated from an artificial intelligence system. An exemplary method embodiment includes the steps of: receiving a first call; determining whether said first call originated from an artificial intelligence system; and when said first call is determined to have originated from an artificial intelligence system performing one or more artificial intelligence call handling operations.

Abstract: A method includes receiving, by a question-answer system, a question in a first language and the question in a second language and predicting, by the question-answer system, a first answer to the question in the first language and a second answer to the question in the second language. The method also includes generating, by the question-answer system, a first vector representing the question in the first language and a second vector representing the question in the second language and adjusting the question-answer system based on the first and second answers and the first and second vectors such that when the question-answer system subsequently generates a third vector representing the question in the first language and a fourth vector representing the question in the second language, a distance between the third and fourth vectors is less than a distance between the first and second vectors.

Abstract: A system for assessing text content of a product. The system includes a computing device having a processor and a storage device storing computer executable code. The computer executable code, when executed at the processor, is configured to: provide text contents and confounding features of products; train a first regression model using the text content and the confounding features of the products; train the second regression model using the confounding features; operate the first regression model using the text contents and the confounding features to obtain a total loss; operate the second regression model using the confounding features of to obtain a partial loss; subtract the total loss from the partial loss to obtain a residual loss; use the residual loss to evaluate models and parameters for the regression models; and use the first regression model to obtain log odds of the words indicating importance of the words.

Abstract: The present invention relates to methods and apparatus for dynamically detecting and/or mitigating threats in communications systems. Exemplary methods and apparatus of the present invention allow for a combination of automated and operator controlled responses to threats. While an operator is provided an opportunity to provide input on how to respond to a threat, after one or more threats of a given type are identified, the system will automatically take corrective action without waiting for operator input and/or in the absence of operator input following notification of a threat.

Abstract: A system and method of creating the natural language understanding component of a speech/text dialog system. The method involves a first step of defining user intent in the form of an intent flow graph. Next, (context, intent) pairs are created from each of the plurality of intent flow graphs and stored in a training database. A paraphrase task is then generated from each (context, intent) pair and also stored in the training database. A zero-shot intent recognition model is trained using the plurality of (context, intent) pairs in the training database to recognize user intents from the plurality of paraphrase tasks in the training database. Once trained, the zero-shot intent recognition model is applied to user queries to generate semantic outputs.

Abstract: A machine learning (ML) module that analyzes multiple language-influencing factors to correct textual content in a more meaningful and efficient manner. In the context of product support, the ML module considers the factors such as a customer's persona that shapes the words a customer chooses while speaking/writing to a customer support agent, current social trends that create new words in the social media/social platforms related to the customer's support issue, and the device used by the customer to input the textual content because different words may be input by the customer when using a smart phone versus a desktop/laptop personal computer with a traditional keyboard. The ML module also may analyze the agent's persona to modify agent's response to the customer because the agent's persona can influence the content of the agent's text. The ML module automatically corrects textual content in real-time before it is sent to the relevant recipient.

Abstract: Embodiments of the present invention provide systems and methods for determining an engagement level for a customer-agent interaction (e.g., a phone call, text chat, etc.), and the engagement level can be used to evaluate the performance of the agent. For example, the engagement level can be used to determine or adjust a skill level of the agent, a performance of the communication channel that facilitates the interaction, and/or an optimal workload of the agent. According to some embodiments, the engagement level is used to determine that the agent may be experiencing difficulty satisfying a customer inquiry such that the agent should be assigned resources or assistance.

Abstract: Methods for optimizing the routing of customer communications include receiving a customer communication; identifying a customer associated with the customer communication; accessing a profile of the identified customer to determine customer data; receiving customer metric scores for a plurality of customer metrics; identifying available agents; accessing a profile of each available agent to determine agent data; predicting interaction outcome metric values for a plurality of customer metrics based on the customer data and the agent data; calculating, in real-time, an aggregate agent-customer pairing score for each available agent; selecting a responding agent from the available agents with the highest aggregate agent-customer pairing score; and providing a routing recommendation to a communication distributor to route the customer communication to the responding agent with the highest aggregate agent-customer pairing score.

Abstract: The present invention extends to methods, systems, and computer program products for detecting online contextual evolution of linguistic terms. Within messaging sources, some users may actively attempt to (relatively quickly) shift the meaning of a word or term. Some users may attempt to perjorate a word or term to have a more toxic meaning. Other users may attempt to reappropriate a word or term to have a less toxic or even a positive meaning. Aspects of the invention identify shifts in implied meanings of words and/or phrases over time. As such, emerging forms of harassment can be identified more quickly. Aspects of the invention can utilize users' behavioral histories as well as messaging structures to improve confidence when identifying term evolution. Machine learning algorithms can be configured to identify term evolution reducing workload on human moderators.

Abstract: A computer system is described that is configured to enable customers of an organization to self-monitor account activity and modify account access settings via a contact center of the organization. The contact center control system is configured to provide an interactive call history associated with the customer's accounts with the organization. Using the call history, the customer may perform self-monitoring of calls into the contact center that access the customer's accounts, and notify the organization if any of the calls appear suspicious. The contact center control system is configured provide a user interface through which the customer may modify account access settings including changing which types of authentication methods are enabled to access the customer's accounts, and blocking account access by callers from certain phone numbers. The contact center control system may allow the customer to turn off all access to the customer's accounts via the contact center.

Abstract: Recently, there has been a demand for further improvement of accuracy of analysis. A system is provided that includes an acquisition unit that acquires call data which is a record of a call between a client and a contact center, a sound recognition unit that performs sound recognition on the call data by using a keyword registered in dictionary data, an analysis unit that performs analysis using a keyword included in a result of the sound recognition, and a registration unit that registers in the dictionary data, a keyword included in a product information database among a plurality of the keywords included in the result of the sound recognition, the product information database storing product information about each of a plurality of products.

Abstract: Systems and methods are provided for facilitating the building and use of natural language understanding models. The systems and methods identify a plurality of tokens and use them to generate one or more pre-trained natural language models using a transformer. The transformer disentangles the content embedding and positional embedding in the computation of its attention matrix. Systems and methods are also provided to facilitate self-training of the pre-trained natural language model by utilizing multi-step decoding to better reconstruct masked tokens and improve pre-training convergence.

Abstract: There is a need for more effective and efficient natural language processing (NLP) solutions. This need can be addressed by, for example, solutions for performing NLP-based document prioritization by utilizing joint sentiment-topic (JST) modeling.

Abstract: In an approach for detecting customer usage of a device and validating a customer claim about the device, a processor receives a statement from a user describing usage of a device. A processor identifies correlating events of the usage of the device via applying natural language processing techniques to the statement. A processor analyzes sensor data from the device via applying a learning model, the learning model being pre-trained to associate the sensor data with physical events. A processor provides a credibility score to the statement based on the analysis of the sensor data and the correlating events exacted from the statement.

Abstract: Methods, systems, and devices supporting data processing are described. In some systems, a data processing platform may support communication message analysis using machine learning. For example, a system may receive a set of communication messages (e.g., social media messages) and perform a machine learning process on the message contents and message interaction data to train a machine learned model. The system may further receive a subject line for a communication message for analysis, input the subject line into the machine learned model, and receive, as an output of the machine learned model, an engagement score based on the subject line. The engagement score may indicate an estimated probability that a user receiving the communication message opens the communication message (e.g., based on the subject line). A user—or the system—may modify the subject line based on the analysis to improve the engagement score.

Abstract: A system includes at least one memory configured to store multiple free-form text-based descriptions comprising descriptions of upcoming travels by multiple users. The system also includes at least one processor coupled to the at least one memory. The at least one processor is configured to perform natural language processing to generate vectors representing the free-form text-based descriptions in a feature space. The at least one processor is also configured to identify similarities between multiple pairs of the vectors in the feature space. The at least one processor is further configured to identify at least two of the free-form text-based descriptions as being related based on the identified similarities. In addition, the at least one processor is configured to generate an output indicating that the at least two free-form text-based descriptions are associated with related travels by at least two users.

Abstract: A word embedding method and apparatus and a word search method are provided, wherein the word embedding method includes training a word embedding model based on characteristic information of a chemical substance, and acquiring an embedding vector of a word representing the chemical substance from the word embedding model, wherein the word embedding model is configured to predict a context word of an input word.