International Business Machines Corporation

American Chemical Society

Mark Ryan Grabau

Mark A Freeman

Sayed Yousef Hashimi

Yi Deng

Christopher Ryan Gessner

Zakeer Rahman Shaik

Rodney Laroy Fulford

Michael E Girata

Dmitrii Arkadyevich Polshakov

Philippe Yves Ayala

John David Fleig

Jeffrey Michael Wilson

Donald Eugene Swartwout

In some embodiments, a computer-implemented method for retrieving a similar document from a corpus of documents is provided. The method may include receiving a search document comprising a set of words; applying a first encoder to generate a first vector; applying a second encoder to generate a second vector; determining a first similarity between the first vector of the search document and the first vector of each document of the corpus of documents; determining a second similarity between the second vector of the search document and the second vector of each document of the corpus of documents; generating a first ranked list of documents based on the first similarity; generating a second ranked list of documents based on the second similarity; applying a voting algorithm to determine a score associated with each document; and outputting a third ranked list of documents based on the determined score.

Systems and methods for performing a computer-implemented prior art search

For a time-series, a baseline error value is reduced to compute a target forecast error value by maximizing a net benefit value of a forecasted value of the time-series. For each forecasting model in a set of models, a corresponding model error value related to the time-series is computed. From the set, a subset of models is selected where each model in the subset has a cost that will produce a positive value for the net benefit. A selected model from the subset is associated with the time-series such that a model error value of the selected model is at most equal to the target forecast error. The time-series is forecasted using the selected model such that the forecasted value has an error of less than the baseline error at a future time, and the forecasted value produces a positive net benefit at the future time.

Adaptive forecasting of time-series

Large amounts of data resulting from periodic sampling and capture of data corresponding to variable operational parameters of one or more systems, such as a plurality of vehicles, and stored in a large number of rows of data corresponding to the number of captured samples are summarized in an analytically meaningful way by simply counting the instances of particular values of the operational parameters of interest in the form of a histogram and stored as a single row of data. The reduction in number of rows of data allows acceleration of processing of later queries by a factor of one thousand (assuming an average trip length of ten minutes) or more.

Data collection for usage based insurance

Growing community of inventors

Westerville, OH, United States of America

Mark Ryan Grabau