International Business Machines Corporation

Stephen Arthur Boxwell

Stanley John Vernier

Keith Gregory Frost

Kyle Matthew Brake

Timothy A Bishop

Benjamin L Brumfield

Nirav P Desai

Charles Evan Beller

Kristen Maria Summers

Christopher F Ackermann

Edward Graham Katz

Christopher Munjal Nolan

David E Wilson

William G Visotski

Jared M Smythe

Chris F Ackermann

Ashok Kumar

John J Thomas

Octavian Florin Filoti

Nils R Holzenberger

Rafael A Leano

Cheyenne E Parsley

An embodiment includes tokenizing an input passage into an n-gram sequence of tokens. The embodiment also includes evaluating the input passage using a trained classification model that generates an output indicative of a classification of the input passage. The embodiment also includes generating a first token vector for a first token of the sequence of tokens and projecting the first token vector to a higher dimensional space, resulting in a first projected token vector. The embodiment also includes generating a first similarity score for the first projected token vector based on comparisons of the first projected token vector to each of a plurality of class representations. The embodiment also includes generating a ranked list of the tokens, wherein the generating of the ranked list includes ranking the first token among others of the tokens based on the first similarity score.

Explainable passage classification

Mechanisms are provided to minimize personally identifiable information (PII) in an electronic document. An iterative personally identifiable information minimization (IPIIM) engine receives an electronic document comprising natural language content having a mention of a protected entity and obfuscates the mention of the protected entity to thereby generate a minimized natural language content. A question answering system processes the minimized natural language content to generate a listing of candidate answers and corresponding confidence scores and the IPIIM engine determines whether or not the minimized natural language content is sufficiently obfuscated based on the listing of candidate answers and corresponding confidence scores. In response to determining that the minimized natural language content is sufficiently obfuscated, the minimized natural language content is provided for processing by a requestor computing device.

Cognitive iterative minimization of personally identifiable information in electronic documents

Embodiments relate to a computer platform and corresponding process and program code to assess multimedia files with respect to similarity and duplicate media content. Data streams are converted into sequences of events, and audio representation with in the streams is identified and subject to processing with respect to the event sequences. A similarity assessment is conducted between two or more of the data streams, and a corresponding distance measurement to quantify similarity is produced. Duplicate data is selectively identified in response to the similarity assessment and the produced distance measurement.

Identifying duplication multimedia entities

Generating automated conversation responses by receiving a conversation input message, determining an intent associated with the conversation input message, detecting content associated with the intent in a data stream in response to determining the intent, and generating a conversation output according to the content and the intent.

Generating responses for live-streamed questions

Embodiments provide a computer implemented method for ranking supporting passages in a question answering (QA) system. The method includes the following steps: receiving an input question and a plurality of passages, wherein the plurality of passages are extracted from the document corpus by the QA system; identifying a plurality of candidate answer occurrences from the plurality of passages; assigning a plurality of first local features to each candidate answer occurrence; merging the plurality of candidate answer occurrences to generate a group of candidate answers; assigning a plurality of second local features and a plurality of nonlocal features to each candidate answer; ranking the group of candidate answers using a linear answer scoring model; and ranking the plurality of candidate answer occurrences using the linear answer scoring model.

Supporting passage ranking in question answering (QA) system

A method provides for assistance in recognition of an entity. A set of data and associated information corresponding to a plurality of entities known to an assisted user is received, such that an instance of the set of data and associated information includes identification of a respective entity of the plurality of entities known to an assisted user. Real-time data corresponding to a first entity is received from one or more devices capturing the real-time data. The real-time data is compared to the set of data and associated information corresponding to the plurality of entities known to the user to determine whether the first entity has a known relevance to the user, and in response to determining the first entity does have a known relevance to the user, the processor provides the identity and relevance of the first entity to the user.

Recognition assistant

A computer-implemented method for optimizing search query results is disclosed. The computer-implemented method includes receiving a first natural language query, selecting a plurality of search query patterns based on the first natural language query, assigning a weight to each search query pattern in the plurality of search query patterns, generating one or more weighted keyword queries from each search query pattern based on the weight assigned to each search query pattern and one or more keywords from the first natural language query, evaluating a relevancy of one or more search results returned by each of the one or more weighted keyword queries for the first natural language query, and iteratively altering the weight assigned to a search query pattern until an optimal number of relevant search results are returned from the search query pattern for the first natural language query.

Tuning query generation patterns

Mechanisms are provided for implementing a proximity based candidate answer pre-processor engine that outputs a sub-set of candidate answers to a question and answer (QA) system. The mechanisms receive a lexical answer type (LAT) and an entity specified in an input natural language question as well as an ontology data structure representing a corpus of natural language content. The mechanisms identify a set of candidate answers having associated nodes in the ontology data structure that are within a predetermined proximity of a node corresponding to the entity, and a sub-set of candidate answers in the set of candidate answers having an entity type corresponding to the LAT. The mechanisms output, to the QA system, the sub-set of candidate answers as candidate answers to the input natural language question for evaluation and selection of a final answer to the input natural language question.

LAT based answer generation using anchor entities and proximity

Embodiments relate to an intelligent computer platform to use ground truth data to rate source reliability, and application of the source reliability. A computer system, computer program product, and computer implemented method apply the ground truth data to evaluate a source, assess source reliability, and dynamically apply the assessed reliability. The assessed reliability is dynamically applied to response data, with the application yielding selective presentation of the response data based on the applied reliability assessment.

Dynamic source reliability formulation

An input question enhancement mechanism is provided for merging sub-question results in Question and Answer (QA) systems in order to answer an input question submitted from a user. A plurality of sub-questions associated with the input question is identified. Responsive to submitting the set of questions to a QA pipeline of the QA system, a set of answers and a set of passages related to the set of answers corresponding to each of the plurality of sub-questions are received. A mapping is applied to the set of passages to form a consolidated set of passages. The set of answers is then merged into the mapping so as to generate a consolidated set of answers. A factoid question response report is generated with answers from the consolidated set of answers and passages from the consolidated set of passages. The factoid question response report is then provided to the user.

Sub-question result merging in question and answer (QA) systems

A method trains a neural network to identify an event based on discrepancies in answers to factoid questions at different times. One or more processors identify answers to a series of factoid questions. The processor(s) compare the answers from the series of factoid questions in order to determine discrepancies in the answers at different times, and then train a neural network to identify an event based on the discrepancies in the answers at the different times.

Training a neural network based on temporal changes in answers to factoid questions

A method learns approximate translations of unfamiliar measurement units during deep question answering (DeepQA) system training and usage. The DeepQA system receives a training set containing Question-Answer (QA) pairs having known unit-of-measurement terms, where each QA pair contains an answer having a known numeric value for a corresponding question from the QA pair. The DeepQA system receives a question from each QA pair from the training set to the DeepQA system in order to find answers and passage phrases to the question from each QA pair, and then identifies all found answers and passage phrases having values that are within a predetermined range of answer values of the training set, where one or more of the identified all found answers and passage phrases contain unfamiliar unit-of-measurement terms, in order to learn approximate translations of the unfamiliar unit-of-measurement terms.

Learning approximate translations of unfamiliar measurement units during deep question answering system training and usage

Embodiments for managing the utilization of software releases are provided. Information associated with a software release and at least one early adopter of the software release is analyzed to calculate a severity score for the software release. A time to utilize the software release is determined based on the calculated severity score.

Methods and systems for preventing utilization of problematic software

Generating a controllable panoramic image while eliminating unsuitable dynamic elements by receiving a plurality of images of a location from a user device, wherein the plurality of images includes images of a location at various times, identifying an object of one or more images of the plurality of images, wherein the object corresponds to an unsuitable condition for a database, determining a score of the one or more images of the plurality of images based at least in part on the identified object, determining a base image from the one or more images of the plurality of images, and generating a set of replacement images of the location based at least in part on respective determined scores of the one or more images of the plurality of images.

Privacy-protecting multi-pass street-view photo-stitch

A question and answer (QA) system, computer program product, and computer-implemented method configured to determine an answer to a question that includes a measurement value. In one example, the QA system receives a question and analyzes the question to identify a measurement value specified in the question. The QA system determines relevant passages to the question. The QA system assigns a measurement value confidence score to a relevant passage based on a comparison of the measurement value specified in the question and a second measurement value specified in the relevant passage. The QA system determines an order of the relevant passages using the measurement value confidence score of each of the relevant passages. The QA system determines an answer to the question based on the order of the relevant passages.

Growing community of inventors

Columbus, OH, United States of America

Stephen Arthur Boxwell