Abstract: An enhanced information retrieval system takes a customer utterance and constructs a contextually-enriched content-based query allowing the system to retrieve the most relevant documents to assist an agent in a real-time conversation with the customer. Phrases in the utterance are classified as informational or non-informational using a machine learning system trained with phrases from prior conversations of multiple users. Content phrases are extracted from the informational phrases using keyword extraction (ranking noun phrases), intent/action extraction (semantic role labeling), and topic label extraction (clustering of historical logs). Emotional content is identified using a sequence tagging model and removed. Contextual information from prior conversations with this user is combined with the updated content phrases to create the contextually-enhanced content-based query, which can then be submitted to the information retrieval system.

Abstract: A dialer in a contact center is managed by a Communication Attempt Management System (“CAMS”) for tracking the total number of communications attempts, including voice calls, originated to an individual within a period of time. A list comprising account data of a plurality individuals may be processed by the CAMS to identify those accounts exceeding a weekly limit on the number of allowable communication attempts and are not authorized to receive further communications in that period of time. The list is modified to reflect those accounts which have not exceeded their corresponding limit and are authorized to be contacted. The modified list is provided to the dialer, which then originates a call to each account. The CAMS may then update each account to reflect each communication attempt made by the dialer and whether the communication is considered exempt or non-exempt with respect to the communication limits.

Abstract: A pipeline is provided for management of a pool of chat message templates for an automated dialog system. The pool of chat messages may be managed using machine learning-based clustering and feedback-based modifications. A set of chat messages may be analyzed using a machine learning model to generate different clusters of messages that are semantically related. Representative messages may be selected from each cluster and used in chat sessions according to the semantic context of the chat sessions. Based on feedback obtained during the chat sessions, metrics generated based on the feedback, and/or other data, modifications may be made to the clusters and/or the representative messages to improve the performance of the automated dialog system.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for automatic text report concept generation. Generating concepts from text reports includes receiving a collection of text reports; performing a clustering process for a plurality of different cluster sizes; evaluating each of the plurality of different cluster sizes to select an optimal cluster size; generating, from the collection of text reports, clusters using the selected optimal cluster size; aggregating text associated with text reports in each cluster; maintaining a training dataset comprising the aggregated text; and generating a predictive model from the training dataset to generate a concept for an input text report.

Abstract: This application discloses a summary generation using natural language processing via method, apparatus, and a computer device. The summary generation is implemented by obtaining a document, the document including at least one sentence. The summary generation is further implemented by extracting m candidate sentences from the document by using an extractive model. The summary generation is further implemented by outputting target words according to the m candidate sentences by using an abstractive model, and generating a summary according to the target words.

Abstract: Provided are a device and a method for providing a voice recognition service based on artificial intelligence. Provided is a device for recognizing a voice based on artificial intelligence for waking up an electronic device in a user specific manner including: an input unit configured to receive voice signals having different frames specific to different users; and a processor configured to: convert the voice signal into a frequency domain signal having a plurality of voice frames in a time domain to extract at least one spectrum associated with each voice frame of the frequency domain signal; configure a template associated with a wake-up voice specific to a first user; and receive a first voice signal and compare the first voice signal with the template to determine whether the first voice signal matches the first user, and to determine whether to wake-up the electronic device based on the determination result.

Abstract: An evaluation system includes a speech data acquisition unit configured to acquire speech data from speech of a member engaged in a task, a speech recognition unit configured to recognize content of the speech from the speech data and convert the content into text data, a task data acquisition unit configured to acquire data indicating an action related to the task of the member and data indicating a task status, an extraction unit configured to extract conversation data related to the task from the text data based on the content of the speech recognized by the speech recognition unit, and an analysis unit configured to analyze correlations between the conversation data, the data indicating the action, and the data indicating the task status.

Abstract: An apparatus has a network interface circuit to receive a source sentence from a network connected client device. A processor is connected to the network interface circuit. A memory is connected to the processor. The memory stores parameters of a multilingual neural review system and instructions executed by the processor to operate the multilingual neural review system trained on a corpus of source sentences, draft target sentences and corrected target sentences. The multilingual neural review system produces a corrected target sentence from a draft target sentence representing a proposed translation of the source sentence. The draft target sentence and the corrected target sentence are supplied to the network connected client device. Approval for the corrected target sentence is received from the network connected client device. Parameters of the multilingual neural review system are updated based upon the approval for the corrected target sentence.

Abstract: A computerized method for evaluating agent-call-logging, in a contact center, is provided herein. The method includes operating an evaluating agent-call-logging module. The operating of an evaluating agent-call-logging module includes: (i) receiving a recorded interaction between an agent and a customer, stored in the database of recorded interactions and a corresponding agent-call-logging, having ‘n’ sentences, stored in the database of call loggings; (ii) operating a speech-to-text algorithm on the received recorded interaction to yield a summarized text thereof. The yielded summarized text is comprised of sentences expressed by the customer; and (iii) operating a semantic comparison unit, to compare each sentence in the yielded summarized text with a corresponding sentence in the agent-call-logging, to yield a Sentence Similarity Score (SSS) for each sentence of the agent-call-logging.

Abstract: Electronic communication methods and systems for determining sentiment associated with a communication, conveniently displaying indicia of the sentiment information, and scoring the sentiment information are disclosed. The methods and systems can include associating an emoji with a sentiment and annotating information, for example, a highlight reel or a waveform, with the emoji.

Abstract: A natural language sentence is generated for a radiology report. One or more words are obtained where the one or more words were produced based on image processing of a radiology image. A computer implemented text analysis process is used to analyse the one or more words to generate a natural language sentence representing the radiology image. The natural language sentence is output. The computer implemented text analysis process includes: determining, for each of the one or more words, and using word modified embeddings, a vector representing the word; and determining, based on the determined one or more vectors, and using a text generator model, the natural language sentence.

Abstract: Systems and methods for providing customized automatic speech recognition (ASR) in a customer support system are disclosed. In an example method, one or more data sources for training an ASR language model associated with the customer support system are identified, and one or more weighting models are selected, each weighting model applying a corresponding weight to each data source of the one or more data sources. The ASR language model is then trained based at least in part on the one or more data sources and the one or more weighting models, and a transcript may be generated for one or more customer support calls of the customer support system using the trained ASR language model.

Abstract: Techniques for processing a user input are described. Text data representing a user input is processed with respect to at least one finite state transducer (FST) to generate at least one FST hypothesis. Context information may be required to traverse one or more paths of the at least one FST. The text data is also processed using at least one statistical model (e.g., perform intent classification, named entity recognition, and/or domain classification processing) to generate at least one statistical model hypothesis. The at least one FST hypothesis and the at least one statistical model hypothesis are input to a reranker that determines a most likely interpretation of the user input.

Abstract: A call handler in a contact center is configured to temporarily derogate the skill level of one or more agents in a group having a common skill. During the time period of derogation, incoming calls will be routed to agents with the highest skill level, such that the normally lower skill level agents will be selected to receive calls. If there are more than one agent meeting this criteria, then the agent that has been waiting the longest to receive a call is selected. This allows the lower skill level agents to gain experience in handling calls and improving their call handling capabilities. After a set threshold, the derogation of the agents is terminated, and their skill level is restored to the pre-derogation level. In this manner the lower skill level agents are given limited opportunities for improvement by handling calls.

Abstract: Techniques for conducting a support session and determine suitable instructions for resolving a certain technical mal-function in a device/equipment of a user. Imagery data associated the technical mal-function is received from a user's device and used for determining at least one improperly setup property associated with the mal-function in the mal-functioning device/equipment based on a comparison of the received imagery data with reference data. Instructions comprising augmented imagery for resolving the mal-function can be then generated, or fetched form a database, based on the determined at least one improperly setup property. A new database record can be generated comprising the augmented imagery data for use in future support sessions associated with the mal-function.

Abstract: A voice awakening method and device are provided. According to an embodiment, the method includes: receiving voice information of a user; obtaining an awakening confidence level corresponding to the voice information based on the voice information; determining, on the basis of the awakening confidence level, whether the voice information is suspected wake-up voice information; and performing, in response to determining the voice information being the suspected wake-up voice information, a secondary determination on the voice information to obtain a secondary determination result, and determining whether to perform a wake-up operation on the basis of the secondary determination result. The embodiment implements a secondary verification on the voice information, thereby reducing the probability that the smart device is mistakenly awakened.

Abstract: A conversation interaction method and apparatus, and a computer-readable storage medium are provided. The method includes: converting a speech to be recognized into a first text; inputting the first text into a semantic analysis model, to obtain intention information and slot information of the first text; and inputting the intention information and the slot information of the first text into a conversation state machine, to obtain interaction information corresponding to the first text. By using a semantic analysis model, intention information and slot information of a first text are obtained directly from the first text. The process in the existing technology, where a semantic analysis model needs to be used immediately after a language model, is avoided, thereby shortening processing time and making it possible to respond faster to a user. Further, by using the above scheme, calculation complexity and the cost of a whole system are reduced.

Abstract: A system is provided for streaming media content in a vehicle. The system includes a personal media streaming appliance system configured to connect to a media delivery system and receive media content from the media delivery system at least via a cellular network. The personal media streaming appliance system operates to transmit a media signal representative to the received media content to a vehicle media playback system so that the vehicle media playback system operates to play the media content in the vehicle.

Abstract: Methods and systems are provided for the AI-based computer-aided persuasion system (CAPS). The CAPS obtains inputs from both the target and the agent for an object, dynamically generates persuasion references based on analysis of the input. The CAPS obtains content output by analyzing the agent audio stream and the target audio stream using a recurrent network (RNN) model, obtains sentiment classifiers based on a convolutional neural network (CNN LSTM) model, updates a conversation matrix, and generates a persuasion reference based on the updated conversation matrix. The persuasion reference is based on an acceptance likelihood result generated from the conversation matrix using the RNN model. The CAPS further generates a target profile using DNN (deep neural net) with input of target Big Data, wherein the target profile includes one or more objects, and wherein the agent is selected based on the generated profile and one or more selected objects.

Abstract: The present disclosure provides a method, apparatus for implementing a far-field speech function, system and a storage medium, wherein the method comprises: a speech detecting unit located on a smart device performing speech signal detection in real time; upon detecting an awakening word, the speech detecting unit awakening an algorithm unit located on the smart device and being in a standby state; the speech detecting unit transmitting the obtained speech signal to the algorithm unit so that the algorithm unit performs arithmetic processing for the speech signal in a predetermined manner, and sends a processed speech signal to a control system of the smart device, to complete a corresponding control operation. The solution of the present disclosure can be applied to save energy consumption and improve the acoustic effect, break away from the constraints of the remote controller and facilitate the user's operation.