International Business Machines Corporation

Li Na Wang

Nan Chen

Ju Ling Liu

June-Ray Lin

Shun Xian Wu

Yu Zhu

Yu Meng Li

Si Bin Fan

Ning Zhang

Lei Tian

Na Zhao

Jing Wen

Ye Chuan Wang

Yan Xin Li

Rong Hong Wan

Han Zhang

Yi Cen Pan

Juan Y Zhai

Yin Xiang

Determining a class-specific predictor indicator importance can include generating, using machine learning, a prediction model having a tree structure including nodes, class label subgroups, and class-specific predictor indicators. A class-specific predictor indicator importance can be generated for a selected class-specific predictor indicator by merging a term predictor indicator frequency and a purity predictor indicator frequency at each instance of the selected class-specific predictor indicator within one or more nodes of the tree structure. A representation of the tree structure displayed in a graphical user interface can depict each path between a root node of the tree structure and a leaf node mapped to the selected class label subgroup having a node containing an instance of the selected class-specific predictor indicator in a manner that contrasts with other paths between the root node and other leaf nodes of the tree structure.

Class-specific predictor importance

A computer-implemented method, system and computer program product for identifying equivalent technical terms. A deep learning model is trained to identify equivalent technical terms. The deep learning model is then applied to a new document. The sentences of the document are analyzed to identify technical terms. Text is then analyzed surrounding the technical term identified in the document to determine the meaning of such text. A glossary list is then reviewed to determine if the identified meaning of the analyzed text matches a meaning/concept in the glossary list linked to a technical term. In response to determining that the meaning of the analyzed text matches a meaning/concept in the glossary list linked to an equivalent technical term, the technical term in the document is annotated with the equivalent technical term. In this manner, non-standard equivalent technical terms in different documents with the same meaning/concept are able to be identified.

Identifying equivalent technical terms in different documents

Content for products can be identified. For each identified content, at least one class to which the first content pertains can be predicted using an artificial intelligence multiclass model. For each identified first content that corresponds to the at least one class, a support level of the product with regard to at least one class can be predicted using artificial intelligence binary class models. For each identified product, data indicating the product and the support level of the product with regard to the at least one class can be added to a data table. A product features map based on the data indicating the products and a support level of each product with regard to the at least one class can be generated and output.

Product features map

Embodiments of the invention are directed to a computer-implemented method of operating a multi-screen virtual reality environment. The computer-implemented method includes performing a wall arrangement and transmission (WA&T) protocol that includes receiving at a second module a function transmitted by a first module over a network to the second module. The second module and the function received over the network are used to generate priority data that identifies a priority of each of a plurality of individual video streams generated by a plurality of video sources. Based at least in part on the priority, the second module is used to generate reduced-size video streams that include selected ones of the plurality of individual video streams. The second module transmits the priority data and a multi-screen video stream that includes the reduced-size video streams and non-reduced-size video streams of the plurality of individual video streams.

Smart multi-screen wall arrangement in a virtual reality environment

Determination of a dynamic meet location for multiple individuals using current conditions. Taking into account a variety of external conditions (including traffic, travel speed, etc.), a meet location is selected and dynamically updated for multiple individuals based, at least in part, on a set of initial conditions.

Dynamic determination of meet location

The present disclosure discloses a method and system for data deduplication. The method comprises: acquiring meta data and multiple data chunks corresponding to at least one original data object, which are generated by using a data deduplication method; combining the acquired multiple data chunks into a new data object; performing deduplication on the new data object to generate new meta data and new data chunks corresponding to the new data object; and storing the meta data corresponding to the at least one original data object, the new meta data corresponding to the new data object, and the new data chunks. The method and system can further improve deduplication ratio, lower data storage amount, and save costs.

Method and system for data deduplication

Growing community of inventors

Beijing, China

Li Na Wang