Avaya Inc.

Pearson Education, Inc.

Lee Becker

David Skiba

George William Erhart

Valentine C Matula

Scott Hellman

Peter Foltz

Kyle D Habermehl

Alok Baikadi

Mark Rosenstein

Jill Budden

William M Murray

Stephen Hopkins

Andrew Gorman

Luis Manuel Oros

Reinhard P Klemm

Samuel Downs

Marcia Derr

Systems and methods for automated custom training of a scoring model are disclosed herein. The method include: receiving a plurality of responses received from a plurality of students in response to providing of a prompt; identifying an evaluation model relevant to the provided prompt, which evaluation model can be a machine learning model trained to output a score relevant to at least portions of a response; generating a training indicator that provides a graphical depiction of the degree to which the identified evaluation model is trained; determining a training status of the model; receiving at least one evaluation input when the model is identified as insufficiently trained; updating training of the evaluation model based on the at least one received evaluation input; and controlling the training indicator to reflect the degree to which the evaluation model is trained subsequent to the updating of the training of the evaluation model.

Systems and methods for automated machine learning model training quality control

Systems and methods for interface-based automated custom authored prompt evaluation

Systems and methods for interface-based machine learning model output customization

Systems and methods for automated evaluation model customization

Real time development of auto scoring essay models for custom created prompts

Systems and methods for automated machine learning model training for a custom authored prompt

A semantic translation model system is described along with various methods and mechanisms for administering the same. The semantic translation model system proposed herein creates an intermediate representation and a knowledge base in multiple languages, reducing the amount of time and expensive resources typically required for translation and automatic response to written communications. The system also removes the problem of a translation being influenced by a person's writing style and human misinterpretation and provides ongoing translation to keep the system current.

Semantic translation model training

The sentiment of a message may not be obtainable from the message itself. However, many messages have an associated context that provides information useful in determining the sentiment of a message. Messages may include links to other resources, such as graphics or videos, which in turn include titles, comments, viewer ratings or other attributes that may provide a sentiment of the message.

Unspoken sentiment

A dialog aggregator provided by a contact center communication system for text-based interaction chains is described along with various methods and mechanisms for administering the same. The dialog aggregator produces a summary, in real-time, of questions posed and existing answers in the interaction chain while identifying outstanding questions that have not been answered for display to an agent. The display includes any current answer the agent is working on as well as completed items and additionally executes rules based on the status of the remaining questions. The display in canonical form of the summary and outstanding question set enables a contact center agent or other observer of the interaction to quickly and efficiently assess the interaction history.

Aggregated multi-topic agent desktop

An automated system for message analysis whereby messages within a given category may be identified and processed as a category connote. While a domain of messages may be monitored and processed in the due course of business, connote message are different. For example, a number of messages may fall into a domain of 'poor airline food.' Such messages may be processed in the due course of business. However, a message with a different aspect, such as, 'I found glass in my food,' may be initially identified as begin within the domain of &#x201c;poor airline food,&#x201d; and processed further to distinguish the message as being a connote with regard to the &#x201c;poor airline food&#x201d; category and warranting special handling.

Anti-trending

Contact centers may incorporate automated agents to respond to inquiries. The inquiries may solicit a substantive response, for example, by providing a time when the inquiry asks for the departure time for a flight. Such responses omit the normal conversational subject matter used to embellish person-to-person conversations and appear are very machine-like. Herein, a source of user context, such as a social media website, customer database, or other data, is accessed. Certain aspects of the customer may then be identified and used to embellish the reply with additional and/or alternative content. As a result, the reply may be more conversational.

Natural language processing (NLP) and natural language generation (NLG) based on user context for enhanced contact center communication

A contact center system can receive messages from social media sites or centers. The messages may include derogatory or nefarious content. The system can review messages to identify the message as nefarious and identify the poster as a social media provocateur. The system may then automatically respond to the nefarious content. Further, the system may prevent future nefarious conduct by the identified social media provocateur by executing one or more automated procedures.

Social media provocateur detection and mitigation

Contact centers employ agents to provide services to customers. In particular, artificial agents are provided that have a rich background and continuing life with the realm of social media. The artificial agent's profile is selected in accord with the profiles of current or prospective customers. As the customers age and progress in life, the artificial agent profile is updated in accord with the customers' changing profiles and similarly ages and progresses in life. When a customer desires to interact with artificial agent, for a purpose provided by a contact center, a human agent may be provided the profile and/or history of the artificial agent so as to impersonate the artificial agent and promote the relationship with the customer.

Pseudo agent matching

A contact center system can receive messages from social media sites or centers. The messages may be in a foreign language. The system can review messages by identifying content in the social media messages with negative/positive sentiment and then identify a seed term in the messages. A seed term can be a word in another language, different from the message body. The seed term is then used to find one or more other words, in the foreign language, that are correlated with the seed term. The identification of the found words in other messages can then be used to determine sentiment in the foreign language.

Cross-lingual seeding of sentiment

A contact center system can receive messages from social media sites or centers. The system can review long messages by identifying content in the long message with negative sentiment. The content with negative sentiment is further analyzed to determine whether the identified content is actionable. If the identified content is actionable, the communication system can automatically routed the long message to an agent for response.

Growing community of inventors

Boulder, CO, United States of America

Lee Becker