International Business Machines Corporation

Paulo Rodrigo Cavalin

Ana Paula Appel

Claudio Santos Pinhanez

Maira Gatti De Bayser

Carmen Nilda Mena Paz

Renato Luiz De Freitas Cunha

Paul Borrel

Bruno Silva

Matthias Kormaksson

Luis Gregorio Moyano

Heloisa Caroline De Souza Pereira Candello

Alan Braz

Carlos R Mendes

Jaione Tirapu Azpiroz

Julio Nogima

Lucas Correia Villa Real

Marcio Ferreira Moreno

Marisa Affonso Vasconcelos

Bianca Zadrozny

Maira Athanazio De Cerqueira Gatti

Rogerio Abreu De Paula

Cicero Nogueira Dos Santos

Renan Francisco Santos Souza

Daniel Salles Chevitarese

Davi M Valladao

Mauro Carlos Pichiliani

Bruno Flach

Melina De Vasconcelos Alberio Guerra

Alexandre Rademaker

Breno William Santos Rezende De Carvalho

Helon Vicente Hultmann Ayala

Michael Raghib

Victor Henrique Alves Ribeiro

Gabriel Louzada Malfatti

Daniela De Mattos Szwarcman

Cicero N Dos Santos

Pedro Henrique Domingues

Carlos Raoni De Alencar Mendes

Flavio Vinicius Diniz De Figueiredo

Joao P Forny De Melo

Vagner F Santana

Alexandre Alves

A topic for building a new intent on which to train a chatbot can be received. A database of chatbot training data can be searched for a candidate intent having meta-knowledge similar to the received topic. Utterances associated with the candidate intent can be extracted. The received topic and the extracted utterances can be input to a trained machine learning model. The trained machine learning model generates example utterances for the new intent. The new intent with the generated example utterances can be used as training data for training the chatbot.

Training example generation to create new intents for chatbots

A privacy-preserving method of accessing a model as a service (e.g., a language model) receives an input prompt authored in a first domain having data intended to be maintained private. In response, and using a conversion model, the input prompt authored in the first domain is converted to a second domain to create a converted input prompt. The converted input prompt preserves a semantic context of the input prompt. The converted input prompt is then delivered to the model as a service, which returns a response in the second domain. Using the conversion model, the response is then converted back into the first domain to create a converted response. The converted response has the semantic context of the input prompt. The converted response is provided as a reply to the input prompt. The conversion model may be trained using data collected from user interactions with the model as a service.

Domain adaptation-based disguising of prompts for data privacy in foundation models

An ore content prediction system is provided. The system receives structured geological data that is derived based on spatial geological information that is associated with an input region. The received structured geological data includes a plurality of multidimensional tensors that are derived from spatial geological information of a plurality of sub-regions of the input region. The spatial geological information includes one or more types of data. The system trains a prediction model to produce a prediction output based on an average grade of an ore of a target mineral type at a target region by using the received structured geological data. The system identifies a relationship of the structured geological data to the prediction output and determines a revised input region based on the identified relationship.

Estimate ore content based on spatial geological data through 3D convolutional neural networks

An approach to identifying architectures of machine learning models meeting a user defined constraint. The approach can receive input associated with evaluating machine learning models from a user. The approach can determine acceptable architectural templates to evaluate the machine learning models based on the input and determine a list of architectures and metrics based on a calculation of maximum neural network sizes of the acceptable architectural templates not exceeding the constraint. The approach can send the list of architectures and metrics to the user for selection.

Automated machine learning model selection

Chatbot logs can be received. The chatbot logs can be clustered or classified into clusters of potential intents. Trending topics can be detected by analyzing at least social media content. Semantic similarities between the potential intents and the trending topics can be computed. Based on the computed semantic similarities, at least one of the potential intents to at least one of the trending topics can be correlated. Training of the chatbot using the chatbot logs associated with the correlated potential intent can be triggered.

Conversational systems content related to external events

A method of automatically creating new intents and assigning examples to those new intents in a dialog system includes generating at least one predicted intent from an intent concept graph by combining two or more nodes of the intent concept graph. A graph embedding is generated of the at least one predicted intent. A sentence embedding is generated from examples from a call log. The graph embedding of the at least one predicted intent is then matched with the examples from the sentence embedding and the at least one predicted intent, with the examples, is connected into a dialog system.

Concept prediction to create new intents and assign examples automatically in dialog systems

A computer system includes a processor configured to execute a conversational agent generator function. The conversational agent generator function is configured to receive a plurality of model input queries. Each model input query includes an intent and a corresponding answer to the intent. A database stores a corpus including sets of textual responses indicative of a common meaning. Each textual response in a given set of textual responses indicates a common meaning constructed in a different style indicative of a corresponding tone of communication. The processor is further configured to execute a rephrasing function configured determine the answer of one of the plurality of model input queries, and convert the answer of one of the plurality of model input queries into a plurality of different rephrased answers from the corpus stored in the database. The processor further outputs the plurality of different rephrased answers to the conversational agent generator function.

Adapting conversational agent communications to different stylistic models

A product quality analysis and control system is provided. The product quality analysis and control system includes a processing system that is configured to receive data of objective qualities of a product from multiple data gathering modules. Each of the multiple data gathering modules is respectively positioned at corresponding nodes of a supply chain and is configured to receive from sensors data of objective qualities of a product when the product is at each of the corresponding nodes. The processing system includes a model repository and a data evaluation module. The data evaluation module is configured to select, from the model repository, a model associated with a type of the product and to use the model to analyze the data to identify changes in the objective qualities along the supply chain.

Product quality analysis and control

A method of providing examples to a computerized conversation agent includes associating one or more triplets of two concepts and a relationship therebetween with an intent related to a query. The triplet is embedded in a knowledge graph and the concepts in the knowledge graph are mapped to the intent. A call log is scanned for examples of the intent based on the concepts in the knowledge graph and the examples are mapped to the intent.

Management of concepts and intents in conversational systems

A system, method and program product for stratigraphic layer identification using a stratigraphic knowledge base for machine learning. Reservoir data includes seismic data and well log data for a reservoir area. The well log data is processed to identify well stratigraphic layer features and the seismic data is processed to identify seismic stratigraphic layer features. A feature matching algorithm based on a stratigraphic knowledge base is selected to match the well stratigraphic layer features to the seismic stratigraphic layer features. The matched features are used to define a stratigraphic layer interpretation for the reservoir area and the interpretation is presented to a user.

Stratigraphic layer identification from seismic and well data with stratigraphic knowledge base

A system, method and program product for processing borehole images to delineate between natural fractures and induced fractures. A system is disclosed that includes: an image analysis platform that inputs a noisy image from a borehole, processes the image using a set of filtering strategies, and renders a set of suggested filtered images via a user interface, the user interface including a mechanism for allowing a user to choose a selected filtered image from the set of suggested filtered images that best delineates between natural fractures and induced fractures, and wherein the image analysis platform further includes a feedback system for packaging and outputting the noisy image and selected filtered image as feedback; and a learning platform having a knowledge registration system that collects and stores training data and the feedback and in a knowledgebase, and a machine learning system that generates filtering strategies.

Identification of natural fractures in wellbore images using machine learning

A computer-implemented method includes extracting a plurality of topics from a plurality of unlabeled social media posts, mapping the plurality of topics to a plurality of frequencies, each frequency in the plurality of frequencies indicating how often a corresponding topic in the plurality of topics occurs in the plurality of unlabeled social media posts, and predicting, based in part on the plurality of frequencies, a future social media posting behavior of a specific social media user, wherein the future social media posting behavior includes a specific topic about which the specific social media user is likely to post at a time in the future and a frequency with which the specific topic is likely to occur in posts of the specific social media user that are created at the time in the future.

Predicting user posting behavior in social media applications

A fracture pattern characterization method, system, and computer program product, include classifying fracture patterns of reservoir layers using a correlation of reservoir information associated with target images and the target images extracted from a database and determining a fracture pattern characterization of a new reservoir layer input based on the classified fracture patterns.

System, method and computer program product for fracture pattern characterization in naturally fractured reservoirs

A method for answering a natural-language user query includes receiving a query from a user by a central agent, parsing the query, and determining if an answer is known to the central agent as being within a local database. If the answer is known, the query is answered with a known natural-language answer using the local database. If not, artificial conversational entity (ACE) invocation is performed. The ACE invocation includes determining a topical field for the query, determining an ACE, of a plurality of ACEs, to send the query to, based on the topical field, and sending the query to the ACE. If the ACE is capable of answering, a natural-language answer is received from the ACE and presented to the user. If not, a natural-language answer is created using social knowledge and presented to the user. Social knowledge includes data crawled from subscriber-provided content of social media services.

System and method for managing artificial conversational entities enhanced by social knowledge

Identifying life events within social network feeds. The method may include receiving social media data. The method may include identifying life event data within the social media data. The method may include determining a life event probability score associated with a life event class for the life event data using metadata. The method may include assigning a first life event class to each item of life event data based on the life event probability score. The method may include creating conversations by grouping the life event data. The method may include extracting metadata. The method may include determining a conversation probability score for each conversation based on the metadata and the life event probability score associated with each item of life event data. The method may include assigning a second life event class based on the conversation probability score. The method may include displaying the conversations based on user preferences.

Growing community of inventors

Rio de Janeiro, Brazil

Paulo Rodrigo Cavalin