International Business Machines Corporation

Nuance Communications, Inc.

Hironori Takeuchi

Futoshi Iwama

Kohichi Kamijoh

Taiga Nakamura

Yuichi Nakamura

Hitomi Takahashi

Yoichi Hatsutori

Ken Mizuno

Shiho Negishi

Hiroshi Horii

Kiyokuni Kawachiya

Tetsuya Nasukawa

Shohei Ohsawa

Keisuke Inoue

Shourya Roy

Satoshi Masuda

Tsuyoshi Ide

L Venkata Subramaniam

Toshiyuki Yamane

Kohtaroh Miyamoto

Yohei Ikawa

Mei Kobayashi

Masaki Aono

Hikaru Samukawa

Hiroaki Nakamura

Hideo Watanabe

Masato Noguchi

Akira Koseki

Nobuyasu Itoh

Toshinari Itoko

Kei Sugano

Toshiyuki Sanuki

Risa Nishiyama

Tamotsu Sogo

Diwakar Punjani

Verified snapshots are generated by obtaining, from one of a plurality of first nodes, a difference between a common data at a first time point and the common data at a second time point that is different from the first time point, generating a first snapshot of the common data at the first time point based on the difference, obtaining a hash of the common data at the first time point from one of the plurality of first nodes, and verifying the first snapshot at the first time point with the hash of the common data at the first time point.

Management of snapshot in blockchain

A computer-implemented method includes receiving a strategy associated with a new machine-learning (ML) project. There exist a plurality of ML projects, each of which includes artifacts, and for each such candidate project, the following are performed: iterations of the candidate ML project are divided into a first phase, including a first set of iterations, and a second phase, including a second set of iterations; a workload to generate the candidate ML project in the first phase is determined; a performance of the candidate ML project in the first phase is determined; an additional workload to generate the candidate ML project in the second phase is determined; and an increased performance of the candidate ML project in the second phase is determined. Final ML projects are selected from the candidate ML projects, based on the strategy. Artifacts of the final ML projects are incorporated into the new ML project.

Sourcing a new machine-learning project by reusing artifacts from reference machine learning projects

Methods and systems for text classification include identifying a text sample, classified according to a first class in a set of hierarchical graphs of classes, that is more similar to a second class in the set of hierarchical graphs of classes than to the first class according to a similarity metric. A word is deleted from the identified text sample that is associated with the second class. A word is added to the identified text sample that is associated with the first class. A text classification model is trained using the text sample as training data. Text classification is performed on an input text using the trained text classification model implemented on a processor device.

Verifying and correcting training data for text classification

A method for emergency communication includes encoding a message for visual display including a message to field and a message from field. The visual display is revealed to a reading device in communication with a broadcast center, which stores the visual display. The messages are decoded and sorted from visual displays at the broadcast center. On an index channel, a time and channel number for when a message body of the message will be broadcast on a message channel is broadcasted. The message body is broadcasted on the message channel.

Peer-to-peer emergency communication using public broadcasting

A computer-implemented method for constructing and searching structured data of question sentences in a question-answer database using machine learning and natural language processing includes receiving one or more question-answer articles as input from one or more question-answer databases over at least one network, extracting one or more profile keywords from a profile associated with a user, assigning a personalization score corresponding to the user to each of the one or more question-answer articles based on the one or more profile keywords, including applying a morphological and dependency analysis to a body section of the question-answer article, receiving a query for obtaining question-answer data relevant to the user, and outputting question-answer data based at least in part on the personalization score of each question-answer article in satisfaction of the query.

Constructing and searching structured data in a question-answer database using machine learning and natural language processing

A method for answering consecutive questions, using a system which can access a database is provided. The method includes retrieving a questioner identifier associated with a questioner in response to receipt of a question from the questioner. The method also includes obtaining, using the questioner identifier, a history on class labels which were assigned to questions made by the questioner in the past and choosing a class label among the class labels in the history. The method further includes obtaining a list of keywords of the question and judging whether or not each keyword in the keyword list matches a representative keyword in any node among one or more nodes which exist in a hierarchy level in the database, or matches a synonym of the representative keyword.

Answering of consecutive questions

A method, computer system, and computer program product for dividing a term with appropriate granularity includes extracting an element word specifying granularity from content by parsing, and, if the term includes at least one element word in a part thereof, dividing the term at a position where the at least one element word exists.

Method, system, and computer program product for dividing a term with appropriate granularity

Methods and systems for identifying conditional actions in a business process are disclosed. In accordance with one such method, text fragments are extracted from input documents. In addition, a plurality of pairs of the text fragments that respectively include text fragments that are similar according to a pre-defined similarity standard are determined. For each pair of at least a subset of the pairs, at least one difference between the text fragments of the corresponding pair is determined. Further, at least two particular pairs of the subset of the pairs are merged in response to determining that the particular pairs have at least one of the determined differences in common. Additionally, the merged particular pairs are output to indicate the conditional actions in the business process.

Identifying and amalgamating conditional actions in business processes

To provide a technique for generating, at a high speed, a smaller-sized set that satisfies an intended property such as, for example, being pair-wise, and includes many test cases that match a set of existing test cases given as an input, candidates to be used from a set of existing input test cases are determined in the following manner: for some parameters, values to be held by test case candidates are determined; test cases having the determined values, among those included in the set of input test cases, are selected as the candidates. A test case having the highest score among one or more test case candidates generated with the method of the related art and one or more test case candidates selected from the set of input test cases is added to a set of output test cases.

Method, program, and system for generating test cases

Growing community of inventors

Kanagawa-ken, Japan

Hironori Takeuchi