International Business Machines Corporation

Robert Leslie Yates

David E Wilson

William G Visotski

Mustansir Banatwala

Joseph Anthony Russo

Mary Ellen Zurko

Michael D Whitley

Christopher Munjal Nolan

David Andrew Brooks

Tin Kam Ho

Thomas Schaeck

Neil Rohit Mallinar

Abhishek Shah

Rajendra G Ugrani

Bridget Briana Beamon

Harold B Moss, Iii

Ayush Gupta

Patrick Y Lin

Blake McGregor

Anastas Stoyanovsky

Luis Alfonso Lastras-Montano

Nataraj Nagaratnam

Sachindra Joshi

Anthony Joseph Nadalin

Evelyn Duesterwald

Pankaj Dhoolia

Ming Tan

Darrell Christopher Reimer

Saloni Potdar

Haoyu Wang

Matthew R Arnold

Steven Michael Pritko

Carol Ann Jones

Michael Desmond

Harold Leon Ossher

Yang Yu

Dinesh Raghu

Haode Qi

Daniel C Gurney

Navneet N Rao

Scott M Lewis

Robert C Johnson, Jr

Prashant S Kulkarni

John J Rawls

Alfred Zollar

Roxana Gheorghiu

Angela Swindell

Adilakshmi Veerubhotla

A combination of propagation operations and learning algorithms is applied, using a selected set of labeled conversational logs retrieved from a subset of a plurality of conversational logs, to a remaining corpus of the plurality of conversational logs to train an automated response system according to an intent associated with each of the conversational logs. The combination of propagation operations and learning algorithms may include defining the labels by a user for the selected set of the subset of the plurality of conversational logs; training a probabilistic classifier using the defined labels of features of the selected set, wherein the probabilistic classifier produces labeling decisions for the subset of conversational logs; weighting the features of the selected set in a model optimization process; and/or training an additional classifier using the weighted features of the selected set and applying the additional classifier to the remaining corpus.

Intent authoring using weak supervision and co-training for automated response systems

The generating of actionable recommendations for tuning model metrics of an Artificial Intelligence (AI) system includes partitioning a key performance indicator (KPI) range associated with a target system into a plurality of buckets. Log data including at least one KPI of the target system and one or more AI model metrics is partitioned and distributed across the plurality of buckets. For each bucket, an aggregate value of the one or more AI model metrics across the log data is computed and weighted according to the volume of log data in that bucket. A correlation factor between the aggregate value and a representative KPI value for each bucket is determined. A model tuning recommendation to increase ranking of the AI model metrics according to the determined correlation factor is provided to an output device and/or to the AI system for updating the one or more AI model.

Generating model insights by progressive partitioning of log data across a set of performance indicators

A domain-specific document is retrieved. The domain-specific document is related to the performance of a process in the domain and includes a plurality of document spans and a plurality of annotations that annotate the spans. A goal-based framework is generated from the domain-specific document. The generation is based on the plurality of annotations and based on the plurality of document spans. The goal-based framework includes entities and entity relationships related to performance of the process. The goal-based framework is transformed into a platform agnostic dialog model based on the entities and the entity relationships. At least one platform specific dialog model is provided to at least one automated response system based on the platform agnostic dialog model.

Assistant dialog model generation

A mechanism is provided to implement a text classifier training augmentation mechanism for incorporating unlabeled data into the generation of a text classifier. For each term of a plurality of terms in each document of a plurality of documents in a set of unlabeled data, a term frequency value is determined. The term is normalized by dividing the term frequency value by a total number of terms in the document. An inverse document frequency (idf) value is determined for each term based on the term frequency value. A subset of terms is filtered from the plurality of terms based the determined idf values. The idf values for the remaining terms are transformed into feature weights. Terms from a set of labeled data are re-weighted based on the feature weights determined from the set of unlabeled data. The text classifier is then generated using the re-weighted labeled data.

Feature reweighting in text classifier generation using unlabeled data

A method overfits a word vector generating process to identify implicit relationships between two or more terms in a corpus. A server identifies instances of multiple user-generated pairs of terms in an original corpus of documents, in which the terms are labeled but a relationship between two or more of the corpus terms are not identified. The server then extracts sentences, from the original corpus of documents, that contain one or more of the multiple user-generated pairs of terms, and combines the sentences into a training corpus, which is used to purposely overfit a word embedding model. This word embedding model leads to a vector that is used to identify other terms that have a same type of relationship as that found in the multiple user-generated pairs of terms, such that search corpus of documents can be searched for similar terms that trained the word embedding model.

Implicit relation induction via purposeful overfitting of a word embedding model on a subset of a document corpus

A mechanism is provided to implement a summarization mechanism for summarizing an identified natural language document using hyperbolic embeddings. Responsive to receiving a query from a user for a summarization of the identified natural language document, the summarization mechanism produces a hyperbolic embedding model of embeddings of the query. The summarization mechanism compares the embeddings of the query to each of a set of embeddings associated with a set of sentences of the identified natural language document. Responsive to identifying a subset of embeddings associated with the set of sentences of the identified natural language document having a semantic specificity to a subset of embeddings associated with the query, the summarization mechanism adds the sentence to a summary of the identified natural language document. The summarization mechanism then outputs the summary to the user.

Natural language document summarization using hyperbolic embeddings

Embodiments for training an automated response system using weak supervision and co-training in a computing environment are provided. A plurality of conversational logs comprising interactive dialog sessions between agents and clients for a given product or service are received. A subset of the plurality of conversational logs are retrieved according to a defined criterion, and a selected set of the subset of the plurality of retrieved conversational logs are labeled by a user. The labeling is associated with a semantic scope of intent considered by the clients. A combination of propagation operations and learning algorithms using the selected set of labeled conversational logs are applied to a remaining corpus of the plurality of conversational logs to train the automated response system according to the semantic scope of intent.

Mechanisms are provided for generating an answer to an input question. An input question is received and a set of candidate answers is generated along with, for each candidate answer in the set of candidate answers, a corresponding selection of one or more selected evidence portions from a corpus of information providing evidence in support of the candidate answer being a correct answer for the input question. The candidate answers are ranked based on an application of a justifying passage model (JPM) to the selected evidence portions for each of the candidate answers in the set of candidate answers. The JPM identifies whether a candidate answer is justified by a selected evidence passage corresponding to the candidate answer. A candidate answer is output as the correct answer for the input question based on the ranking of the candidate answers.

Justifying passage machine learning for question and answer systems

According to an aspect, a question answering computer system determines a set of search terms for a given question. A plurality of passages that each includes at least one search term in the set of search terms is identified in response to a search of a document for search terms in the set of search terms. A passage from the plurality of passages is selected. A context based score of the selected passage that indicates the likelihood that the selected passage contains an answer to the question is generated based on criteria that specify a relationship among the set of search terms, content of the passage, and content of at least one other passage in the plurality of passages.

Context based passage retreival and scoring in a question answering system

Mechanisms are provided, in association with a question and answer system, for generating answers to an input question. An input question is received and processed to generate at least one query. The at least one query is applied to the corpus to generate a set of candidate answers and corresponding selections of evidence portions of the corpus of information providing evidence in support of the at least one candidate answer being a correct answer for the input question. A graphical user interface (GUI) is output comprising a first GUI sub-section comprising the set of candidate answers, and a second GUI sub-section comprising the evidence portions of the corpus. User input modifying at least one of set of candidate answers or the evidence portions of the corpus of information is received via the GUI and an operation of the data processing system is adjusted based on the user input.

User collaboration for answer generation in question and answer system

Context based passage retrieval and scoring in a question answering system

An information sharing paradigm for a cloud computing solution enables flexible organizational boundaries with respect to cloud resources. Cloud service customers manage their own organization boundary but can extend that boundary selectively by associating cloud resources they own with sets of domain names that may be associated with requests for cloud resources that the organization may be willing to share with other organizations that are using the cloud environment, and by ensuring that any such requests for resources that are shared in this manner are associated with one or more message handling policies that have been defined by (or otherwise associated with) the resource-owning organization. Cloud resources owned by an organization (even those marked as 'internal only') may be selectively shared with one or more other organizations using the cloud environment depending on the domain names associated with the requests. Message handling policies are enforced with respect to shared resources.

Extending organizational boundaries throughout a cloud architecture

Mechanisms are provided for identifying commonalities between candidate answers generated by a Question and Answer (QA) system in response to an input question. The mechanisms receive a plurality of candidate answers for an input question from the QA system and identify terms present in the candidate answers. The mechanisms determine relationships between terms in each of the candidate answers and determine a common relationship between a first term and a second term, the common relationship being common amongst at least a subset of the plurality of candidate answers, based on the determined relationships between terms in each of the candidate answers. The mechanisms present the plurality of candidate answers and the common relationship to a user.

Identifying and displaying relationships between candidate answers

Growing community of inventors

Arlington, MA, United States of America

Robert Leslie Yates